Fanout queries, in particular, deserve special attention. These are queries ChatGPT supposedly searches to help generate its response, and if you look in the right parts of your browser’s console, ChatGPT tells you what they are. So, the logical conclusion folks have reached is that if you rank for them, you’ll have a good chance of ChatGPT seeing, and possibly citing your article. On the surface, this makes sense.

But fanout queries cannot fully explain how to get cited by ChatGPT, for two reasons: 

1. A simple manual check can show you that there are tons of sources in a ChatGPT response that aren’t ranking for the fanout queries. Where does ChatGPT get those from? 
   
2. Fanout queries shift with subtle changes in the user’s prompt. And as we’ve established before, you can’t predict users’ prompts to ChatGPT. So that means you also can’t predict the fanout queries. Then how do you know what to target in SEO?

So, in trying to figure out how to get cited by ChatGPT, we wanted to look beyond fanout queries and ask a more practical question:

Does ranking for traditional SEO keywords help you get cited by ChatGPT?

To answer this, we took 70 sources that ChatGPT cited across 10 BOTF prompts and ran them through Ahrefs to see what organic keywords each URL was already ranking for. 

(We focused on BOTF prompts only because, as we’ve written about before, those are the only AI search topics where ChatGPT mentions brands. And most of the time, these are the only queries where it cites sources rather than relying solely on its training data. So, that’s where the opportunity is for brands to get traffic and conversions.)

Of these 70 sources, 60% were ranking organically for at least one relevant keyword, and most of them were ranking for three or more. These are traditional, common-sense SEO keywords that one can easily discover doing traditional keyword research in any SEO tool. 

Additionally, half of them were ranking on page one and a third were in the top three positions. 

That’s good evidence that ranking for bottom-of-funnel keywords related to your customers’ pain points helps you get cited by ChatGPT. 

You don’t need to guess what fanout queries to target. You just need to understand the underlying intent behind them and focus on that, which is what successful BOTF SEO already does. 

Below, we look more closely at where ChatGPT citations appear to come from, what this means for your GEO strategy, and where the sources that don’t appear in live web search may be coming from.

Every source ranking for a fanout query also ranked organically for a traditional SEO keyword

While our main takeaway from this study is that you don’t need to know the exact fanout queries in order to get mentioned by ChatGPT, looking at the fanout query data adds a layer that helps us understand the relationship between traditional SEO and cited sources. 

We’ll start with the sources that did rank for fanout queries. 40% of the sources we studied were ranking in the top 10 pages of the SERP for known fanout queries and every one of them also ranked for relevant, more traditional, SEO keywords. 

For example, one of the fanout queries for the prompt shown in the screenshot below was “CRM software that helps reduce dropping leads between stages best CRM lead management features”. And the article outlined in yellow below ranked in Google’s SERP for that fanout query.

Of course no SEO would intentionally go after a query that long and specific. But checking in Ahrefs, we see it also is ranking for extremely normal keywords related to CRMs. 

Every source ranking for a long, hard-to-guess fanout query was also ranking for more traditional, discoverable SEO keywords.

Another 20% of the sources we studied also ranked for relevant SEO keywords but weren’t found in the SERPs for fanout queries. And the last 40% didn’t have any trackable organic presence at all. 

Here’s what that looks like at a glance:

We draw two conclusions from this:

1. The 20% with organic presence and not in the SERP for fanout queries was likely drawn from fanout queries we can’t see. This isn’t a new idea. Others have also noted that it’s highly likely that we are only able to see some of the fanout queries ChatGPT uses.
   
2. If you aren’t ranking for traditional SEO keywords, it’s highly unlikely that you’ll get in the SERPs for fanout queries, and therefore won’t be found via live search.

Overall this data supports the key thesis of our Prioritized GEO strategy: that ranking for relevant, bottom-of-funnel SEO keywords is a key mechanism by which you can get noticed and cited by LLMs (in this case, ChatGPT). However, as we’ll discuss in a later section, citations may not necessarily influence ChatGPT’s answers. 

But very few of the cited sources are ranking for the same keywords

We also checked whether cited sources were ranking for the same organic keywords. If most of them were ranking for the same one or two terms, those keywords would likely be worth targeting.

But we didn’t find any significant overlap, meaning most of the cited sources were ranking organically for different keywords. 

There was some overlap where one or two URLs were ranking for the same keyword, but overall there wasn’t enough of a pattern to tell us anything useful. So, our conclusion is that the more relevant keywords you rank for, the better your chances of getting picked up, but no single keyword seems to be the magic ticket to ChatGPT citations. 

This continues the trend we keep seeing: GEO or AIO seems to be less hackable than SEO, you just need to do the work of publishing helpful content on the topics your customers care about and, slowly, more of them start to be found by LLMs. 

Not all cited sources come from live search

It’s a common assumption that if ChatGPT cites a source, it found it through a live web search. But as we found in our fanout query SERP study, that’s only true about 40% of the time.

So where do the other 60% of cited sources come from? We covered the full picture in that study, but the short version is that no one really knows for certain. 

The leading theories are:

• Additional queries running behind the scenes. Like we covered in the previous section, the fanout queries that we can see are likely just a subset of what ChatGPT is actually searching. 
• Response caching. ChatGPT may be reusing sources from past queries rather than running fresh searches every time, particularly for common buying-intent prompts.
• Training data. There’s evidence that suggests ChatGPT can remember specific URLs from its training data. 
• Hallucinations. Roughly 10% of citations in our study led to error pages. Some sources simply don’t exist.

The honest answer is that there’s a randomness to it. We tend to assume AI is operating logically and systematically, but in reality it’s trained on patterns and probability and that makes its behavior harder to predict and control than traditional search.

Cited sources don’t necessarily influence answers

This entire article has been focused on where citations come from, and overall, we’re interested in anything that helps us understand how LLMs work and gives us practical information to work off of. But, as we’ve written about before, being mentioned has way more impact on traffic and conversion than being cited as a source. 

So the question is does getting cited correlate to being mentioned? 

We’re currently working on a separate study to answer this more directly, but for now we can say that having worked with dozens of clients to increase AI visibility over the past few years and looking at the evidence we’ve seen from other studies, we’ve come to the conclusion that that’s not the case.

No one really knows for sure how these LLM chats work, but evidence tells us that tools like Perplexity and Google AIO retrieve sources first and then build the answer. ChatGPT on the other hand seems to rely much more heavily on training data and generates the answer before adding citations. Because the answer is already written before sources get attached, those citations aren’t shaping what users read. And because your brand is tucked away as a citation (rather than a recommendation ChatGPT is including in its answer), it’s not driving meaningful traffic back to you.

We do track citations for our clients, but we put more weight on mentions because that’s where conversions actually come from. 

To learn more about how we increase and track mentions for our clients, reach out. Or you can sign up for our mailing list and we’ll send you studies and articles about what we’re learning as we go.

That’s exactly where Toro TMS was when they came to us.

Toro TMS makes trucking management software (“TMS”) for bulk haulers — think dump trucks, fuel tankers, and construction materials. It’s a subset of the trucking industry that most TMS platforms ignore because the larger software players are built for general freight. Toro saw that gap and built their product specifically for bulk haulers.

When we met their head of sales, he was building the outbound motion, hiring reps, and running the sales process himself. He knew inbound leads would be critical for long-term growth, but he didn’t have the time or expertise to build an inbound SEO and content marketing engine. So he hired us to build that side of the business while he focused on what he did best: sales.

When we started in early 2025, roughly 5% of Toro’s revenue came from branded inbound. Eighteen months later, that number is over 25%.

Here’s how we did it.

The Starting Point: Branded Keywords Only

When we kicked off in January 2025, here’s what Toro ranked for:

That’s it. Every keyword driving real traffic was someone who already knew Toro’s name. The handful of non-branded keywords they showed up for were all on page 2 or beyond, driving essentially zero traffic.

This is what most B2B companies look like before investing in content: you rank for your own name and nothing else. People who already know you can find you. Everyone else can’t.

What We Rank for Today (18 months later)

Here’s a sample of what they rank for today:

This is just a sample. In total, Toro went from 0 first-page rankings for non-branded keywords to 43 today. Out of the 46 articles we’ve published, 43 of the keywords we’ve targeted now rank on page 1. 

That’s a 93% first-page ranking rate in 18 months. The remaining 3 pieces not on page 1 are mostly newer publications that should reach page 1 in due time.

The Strategy: Pain Point SEO Across Every Keyword Type

Our approach with Toro was the same strategy we use with every client: Pain Point SEO. We identify the keywords that people search when they’re actively looking to buy, and we write content targeting those keywords.

The keywords we’ve targeted for them fall into three types:

Category keywords are the core of the strategy. In short, they’re keywords that indicate someone is shopping for products in your category (read more about Category Keywords). These range from the broadest version of what the product is (the head term, now ranking #1) down to specific features the product offers: payroll, dispatch, load management, ticket management. Each one maps to a real capability of the product. We also layer in specificity where Toro has a niche advantage. Most competitors are fighting over generic software keywords. Toro ranks for keywords specific to their sub-industry because they have content that goes deep on those use cases. Their competitors don’t.

Comparison keywords capture people actively comparing options. These include keywords like “toro tms vs. [competitor]” or “[competitor] alternatives.” 

Template keywords are a type of jobs-to-be-done keyword and catch people one step before the software search. Someone downloading a dispatch spreadsheet template is managing their operations manually. They’re likely outgrowing spreadsheets and will eventually need software.

For Toro, that meant mapping out every way a trucking company owner or fleet manager might search for software. Not just the obvious head terms, but every variation: category keywords for every feature the product offers (payroll, dispatch, accounting, compliance), comparison keywords (best X, X for small companies), and even template keywords that catch people one step before the software search.

These keywords aren’t massive in volume. The biggest gets about 1,500 searches per month. Many get 100 to 200. But most people searching those terms are potential buyers. That’s the whole point of Pain Point SEO. You don’t need 50,000 visitors. You need 500 of the right ones.

How We Wrote the Articles for Toro TMS

We’ve published 46 bottom of funnel, product-centric articles over the course of the engagement. As we’ve written about before, writing quality is equally as important as keyword and content strategy. Even if you target the right keywords, if the writing is low quality (whether from humans or AI), it will massively hurt your chances to both (a) rank and (b) impress your prospects well enough to convert. 

So for Toro, every article we wrote relied on a few foundational principles that we’ve relied on for 8+ years:

(1) We didn’t write generic product comparisons like a third party would. We wrote from Toro’s perspective as a company that deeply understands bulk hauling. The category keyword article doesn’t pretend to be an objective third-party review. It opens by discussing what matters in trucking management, steers toward the specific needs of bulk haulers, and then explains how Toro built their product to address those needs.

This is the approach we laid out in our self-promotional listicles article: you don’t need to pretend you’re unbiased. You don’t need to rank yourself #1 on a fake list. You just need to be honest about what your product does and who it’s for, and go deep enough that the content is genuinely useful to the reader.

(2) We interviewed Toro’s team to get the details. Every article is built on interviews with the people at Toro who know the product and the customer. That’s how we write about things like compliance reporting, ticket management, or the specific challenges of hauling construction and demolition waste. You can’t get that level of detail from Google research or AI generation. It has to come from people who actually know the industry and have a strong opinion on how and why their product or service approaches the problem the way it does. People want these unique perspectives. They live inside key employees’ heads in your organization. You need a process to extract those perspectives and get them into your content.   

(3) We wrote at the depth of a sales conversation. When a bulk hauler is evaluating TMS options, they want to know how it solves their problems in extreme detail. For example, for Toro they want to know exactly how the software handles dispatch, invoicing, driver management, and compliance. They don’t want a 300-word overview or a basic table with checkmarks, they can ask ChatGPT for that. Your content needs to go deeper. So our articles did. They go deep on features, workflows, and specific use cases because that’s what converts readers into demo requests.

The Results: Rankings

We covered the keyword types above. Here’s the distribution of where those keywords rank:

30 keywords in positions 1 through 3; 13 more in positions 4 through 10. That’s 43 out of 46 keywords on page 1, an 93% hit rate.

The Results: Demos Booked

This is the number that actually matters.

In January and February 2025, Toro booked zero inbound demos. By June, they were at 4 per month. By October, they hit 17. After a natural dip in November, the numbers climbed back to 15 in March 2026 and have been consistently in double digits since.

Over the trailing 12 months, Toro is averaging roughly 10 inbound demos per month from content. For a niche B2B software company selling to bulk haulers, that’s a significant pipeline. Every one of those demos is someone who found Toro through a blog post, read about the product, and decided to book a call. The sales team didn’t have to prospect them.

The result: inbound went from 5% of revenue to over 25%.

In even more detail on results that matter to their business, from Q1 ‘25 to Q1 ‘26:

• Opportunity volume is up 350%
• Booked revenue is up 362%

The Results: AI Visibility

As we’ve written about before, this product-centric content strategy improves AI visibility. The same content that ranks on Google also gets Toro recommended by AI search tools. We track this through Traqer, our AI visibility tool.

Across 18 topics and 90 prompts, Toro TMS has:

• 91% visibility on Google AI Overviews — when someone asks Google’s AI about trucking software, Toro appears in 91% of relevant responses
• 81% visibility on AI Mode — Google’s newest AI search feature
• 59% visibility on Perplexity
• 50% visibility on Gemini
• 35% visibility on ChatGPT — lower here, but growing

Of 90 tracked topics, Toro has some visibility in 86 of them and high visibility (appearing in more than 50% of prompts) in 65. That’s not a handful of lucky mentions. That’s broad, consistent AI coverage.

This matters because Toro’s customers are exactly the kind of people who are starting to use AI search to find software. A trucking company owner asking ChatGPT “what’s the best TMS for a dump truck fleet” is going to get Toro in the answer because we’ve published content that gives the LLM the details it needs to make that recommendation.

The content that ranks on Google is the same content that gets cited by LLMs. If you write detailed, specific content about your product and your market, AI search picks it up. That’s the thesis behind our topic-based GEO approach, and Toro is a clean example of it working.

Building an SEO and AI Inbound Channel via Content

Toro’s situation is extremely common among our clients. A head of sales, a VP of marketing who just got hired, a founder who knows they need inbound but doesn’t have the team to build it.

The playbook is straightforward:

1. Identify the buying-intent keywords in your space. For Toro, that’s every variation of “trucking software” plus adjacent searches around payroll, dispatch, invoicing, and compliance.
   
2. Write content that goes deep on your product’s strengths. Not generic overviews. Detailed, interview-driven content that explains exactly what your product does and who it’s for.
   
3. Be patient. Toro’s first two months were zero demos. That’s normal. SEO takes time. But once the rankings hit, the pipeline compounds month over month.
   
4. Create content that works across AI and SEO. We target high intent keywords on the SEO side, but make the content specific and detailed enough that LLMs will find it and recommend that brand to specific use cases like we talk about in our Invisible Prompts post.

Across many of these engagements, we’ve also heard the same story from founders, content marketing teams, and SaaS marketers that their previous SEO agency ran a monthly technical audit and showed them a running list of issues they were fixing or “actively working on.” Most of those issues were real, but the ones that actually impacted the site’s ability to rank for keywords that drive organic traffic, qualified leads, trial signups, demo requests, and pipeline usually required engineering or web development work to fix, which rarely got prioritized.

In our experience, most technical SEO audits mix together two very different types of problems. That’s why when we audit SaaS websites, we separate these issues into two categories: 

• Structural technical SEO: These are decisions made during a site redesign, replatforming, or major launch that can compound for years and are often expensive to reverse. For SaaS companies, it includes decisions such as whether to host the blog on a subdomain or subfolder, whether the marketing site renders server-side or client-side, and how the overall site structure is organized.
• Maintenance technical SEO: These are recurring, smaller fixes that audit tools tend to flag, such as broken links, redirect chains, missing meta descriptions, schema, and other crawl hygiene issues. These issues are real, but they usually don’t compound the same way and can be handled through quarterly audits or basic monitoring.

In this article, we walk you through the technical SEO issues we most often see on SaaS websites, in roughly the order they impact rankings and pipeline.

Issue 1: The Blog Is Hosted on a Subdomain Instead of a Subfolder

Google has publicly said that subdomains and subfolders are treated equally, but across many SaaS SEO engagements, we’ve seen that hosting content in a subfolder consistently outperforms a blog on a subdomain in both organic traffic and rankings.

That’s why we treat subfolder hosting as the default recommendation for any SaaS company starting a blog or planning a migration, and we often recommend migrating existing subdomain blogs back to subfolders when the SEO upside justifies the engineering effort.

We usually see this problem on SaaS websites where the marketing site is built on a custom framework or static site generator, and the engineering team doesn’t want to maintain a CMS as part of the main site. So they just stick a blog running on WordPress, Webflow, or another CMS on a subdomain. But this comes with a serious SEO tradeoff that engineering teams usually don’t consider. 

The problem is that subdomains behave like separate domains, and the backlinks you’ve earned through link building, the brand authority, and the topical relevance built on your main domain don’t carry over cleanly to the subdomain blog, even though both appear to be the same brand to users. As a result, the blog ranks with less authority than it would if it were a subfolder on the root domain.

Our Circuit case study is one of the clearest examples. Circuit had previously moved its blog to a subdomain and saw rankings drop. When they moved the blog back to a subfolder on the root domain, traffic recovered and, within six months, grew from around 920 sessions per month to over 14,577. 

How to Fix?

Host your blog at /blog/ on your root domain.

If your blog is already on a subdomain, plan a migration that includes:

• A reverse proxy or CMS setup that serves the blog from /blog/ on the main domain
• 301 redirects from every existing subdomain URL to its new subfolder URL
• Updated internal links across the marketing site, footer, and product to point to the new URLs
• An updated XML sitemap submitted to Google Search Console

If the content is intended to rank in organic search and contribute to your overall SaaS SEO strategy, content marketing, and link building efforts, it should usually reside on the main domain. 

We also recommend the same principle to other content hubs that often end up on subdomains for engineering convenience, such as help centers, community forums, resource libraries, and changelogs.

Issue 2: The Marketing Site Renders Client-Side Instead of Server-Side

Most SaaS marketing sites are built by the same engineering team that builds the product, on the same JavaScript framework (typically React, Next.js, Vue, or Nuxt). That choice makes sense from a stack-consolidation and engineering-ownership perspective. 

The problem is that when the marketing site renders entirely on the client side, search engines and their crawling algorithms have to work much harder to see your content, which usually results in delayed indexing, partial content rendering, and product pages that struggle to rank for keywords they should easily own.

The technical issue is that search engine crawlers don’t render JavaScript the way browsers do. Google does have a second-pass rendering stage for JavaScript, but it runs on a delay (sometimes days or weeks after the initial crawl), and pages with heavier JavaScript or rendering errors often don’t make it through cleanly. Other search engines, including Bing and the crawlers behind AI search products like ChatGPT and Perplexity, render JavaScript less reliably or not at all. 

We see this most often on SaaS sites built on Nuxt, Angular, or custom React frameworks, where the team has defaulted to client-side rendering for the marketing pages. In these engagements, we typically find that a meaningful portion of web pages are missing from Google’s index, even though every page is internally linked and submitted in the XML sitemap. 

Once the marketing site is switched to a server-rendered or statically generated configuration, indexation usually recovers within weeks.

You can diagnose this on your own site in under a minute. Disable JavaScript in your browser’s developer tools (in Chrome, open DevTools, then use the command menu to search for “Disable JavaScript”) and reload any marketing page. If the main content, headings, or internal links disappear, search engines that don’t fully execute JavaScript are seeing the same empty page.

How to Fix?

For the marketing site, use server-side rendering (SSR) or static site generation (SSG). Both deliver fully rendered HTML to crawlers on the first request, eliminating rendering delays and reliability issues.

If you’re on Next.js, this means using getServerSideProps or getStaticProps (Pages Router) or server components and static export (App Router). On Nuxt, it means running in ssr: true mode rather than SPA mode. For custom React or Vue setups, it means adding a server-rendering layer or migrating to a framework that handles it.

The deeper structural fix is to separate the marketing site stack from the product app stack. The product app can remain a client-rendered SPA since users authenticate before they see it, and SEO doesn’t apply. The marketing site should be its own deployment, ideally on a static site generator or an SSR framework, with content management handled through a CMS that the marketing team can edit without filing engineering tickets.

If switching frameworks isn’t realistic in the near term, the interim fix is dynamic rendering or pre-rendering specifically for crawlers. Services like Prerender.io serve a pre-rendered HTML version to crawlers while serving the SPA to users, which solves the indexation problem without changing the underlying stack. This works as a bridge but isn’t a long-term substitute for proper SSR or SSG optimization.

Issue 3: Thin and Duplicate Pages Don’t Have a Clear Canonical or Noindex Policy

Most SaaS websites accumulate thin pages or duplicate content over time, but many don’t have a consistent canonical or indexation policy for handling them.

We’ve seen login-gated marketing pages appearing in search results, internal search pages getting indexed, and UTM-tagged URLs attracting external links that should have pointed to the canonical version.

Individually, these issues may look minor. Collectively, they create a search profile where the marketing site has far more indexable web pages than it should. This can dilute link authority across near-duplicate pages, waste crawl budget, and make the site appear lower-quality because of its thin and duplicate content footprint.

Over time, this can also lead to indexing and ranking issues, especially on larger SaaS sites with hundreds or thousands of blog posts, feature pages, integration pages, comparison pages, resource pages, and template-based URLs.

How to Fix?

Apply canonical and noindex rules consistently at the template level so that every URL of the same type receives the same treatment automatically.

• Internal search URLs: Add noindex tags at the template level and exclude these URLs from XML sitemaps. Also, verify your robots.txt file isn’t accidentally blocking important sections of the site while handling internal search URLs.
• Filter, sort, and parameter URLs: Add a <link rel=”canonical”> tag pointing back to the clean, unparameterized version of the page. For very heavy filter combinations, consider adding noindex to specific parameter groups.
• UTM-tagged URLs: Make sure UTM parameters canonicalize back to the clean URL. If these URLs are being linked externally, update the source links when possible so authority flows directly to the canonical page.
• Paginated pages: Use a self-referential canonical tag on each paginated URL rather than canonicalizing page 2, page 3, and beyond back to page 1. Keep paginated pages crawlable and indexable so search engines can reach the deeper content linked from them. Avoid blanket noindexing paginated URLs, since that can also prevent crawlers from following links to deeper content. rel=next and rel=prev tags are no longer required for Google, though leaving them in place generally doesn’t hurt.
• Login-gated marketing pages: Add a noindex directive or serve a 401 or 403 status code, depending on whether you want the URL discoverable in any form.

Once these rules are in place, you can audit the indexation report in Google Search Console quarterly. Look for sudden growth in indexed URLs, especially when that growth is not tied to new content production. That usually means a new template, parameter group, or URL pattern is being indexed without the right canonical or noindex rule.

If your indexed URL count is growing faster than your content volume, there’s almost certainly a canonical or noindex policy issue worth investigating.

Issue 4: Pages Don’t Follow a Scalable URL Taxonomy

Most SaaS product sites already have CMS collections for features, integrations, comparison pages, use cases, industries, FAQ pages, and help docs. The issue we often see is that these page types are created in isolation.

For example, different teams produce feature pages, integration pages, comparison pages, or industry pages, and each team organizes its pages around its own campaign or internal workflow.

The site has organized page types, but no consistent structure for how commercial pages relate to one another.

For example, a SaaS company might have:

• A feature page targeting “[feature] software”
• A use case page targeting “[job to be done] software”
• An industry page targeting “[feature] software for [industry]”
• An integration page targeting “[partner] integration”
• A comparison page targeting “[competitor] alternative”

Without a clear taxonomy, these pages often overlap, compete, or remain disconnected from one another. The site ends up with multiple commercial pages targeting adjacent intent, but no clear parent-child structure, internal linking model, or rule for which page should rank for which query type.

This is where URL architecture becomes more than a formatting issue. A strong SaaS URL taxonomy is part of your overall site structure and defines how commercial intent is organized across the site.

When this isn’t defined, several issues show up over time:

• Feature, solution, and industry pages start targeting the same keywords.
• Integration pages sit isolated from relevant feature and use case pages.
• Comparison pages are launched as one-off landing pages rather than as part of a scalable competitor hub.
• Internal linking becomes manual instead of template-driven.
• Breadcrumbs don’t reinforce the commercial hierarchy.
• XML sitemaps group pages by CMS origin rather than by search intent.
• Reporting can’t easily show which commercial page type is driving rankings, qualified traffic, trials, demos, or pipeline in Google Analytics or your internal dashboards.

How to Fix?

Build a commercial URL taxonomy and a clear site structure before scaling new SaaS landing pages. 

This means defining the role of each page type, the query class it should target, the URL pattern it should use, and how it should link to related pages.

A simple taxonomy might look like this:

Once the taxonomy is defined, apply it across the site:

• Move existing pages into the correct URL structure when the SEO benefit outweighs the migration risk.
• Add 301 redirects from old URLs to new URLs.
• Update internal links so related features, use cases, industry, integration, and comparison pages reinforce each other.
• Add breadcrumb logic that matches the taxonomy.
• Segment XML sitemaps by commercial page type.
• Track rankings, traffic, conversion rates, and pipeline by page type, not just by individual URL.

For SaaS companies, this matters most once the site has more than a handful of product-led SEO pages. At that point, every new feature, integration, use case, industry, and comparison page either strengthens the site’s commercial search architecture or adds another disconnected URL that has to be manually managed later.

A strong taxonomy lets the site compound and supports a scalable, high-quality content strategy, consistent on-page SEO, and ongoing optimization across the site. Each new page fits into a known structure, links to related pages by default, appears in the right sitemap, inherits the right breadcrumbs, and has a clear role in the broader keyword strategy.

Issue 5: 404s, Redirect Chains, and Broken Internal Links

For SaaS sites that undergo frequent product changes, redesigns, and migrations, this is one of the clearest examples of a maintenance-related technical SEO issue. It may not require daily attention, but it does need regular cleanup.

The most common causes of these issues are when features are renamed, and the old feature pages start returning 404s, when blog posts are deleted but remain in the XML sitemap, or when site migrations leave behind redirect chains where one URL passes through two or three intermediate URLs before reaching the final destination. 

None of these issues is critical or usually breaks the site on its own. But over time, they create technical drag, leading to crawlers wasting time on dead URLs, internal links passing through unnecessary redirects, and users occasionally landing on pages that no longer exist, which hurts user experience and crawl efficiency.

The most common patterns we see are:

• 404s with internal or external links pointing to them: Pages get deleted, but the links pointing to them don’t get updated. If the old URL had backlinks from your link building efforts, rankings, or meaningful internal links, redirect it to the closest relevant live page instead of letting those signals die on a 404.
• Redirect chains: URL A redirects to URL B, which redirects to URL C. The cleaner setup is for URL A to redirect directly to URL C.
• Redirect loops: URL A redirects to URL B, and URL B redirects back to URL A. This makes the page inaccessible to users and crawlers.
• Internal links pointing to redirected URLs: The link technically works, but every click and crawl has to pass through a redirect. 
• 302 redirects used for permanent moves: If a page has permanently moved, use a 301 redirect instead of a 302 so search engines understand that the new URL should replace the old one.
• Soft 404s: These are pages that return a 200 status code but display “not found,” “page no longer exists,” or similarly thin content. The response code says the page exists and functions normally, but the content says otherwise.

How to Fix?

Use Screaming Frog, Ahrefs Site Audit, Semrush, or a similar crawler to identify 404s, redirect chains, redirect loops, soft 404s, and internal links pointing to redirected URLs.

Then clean up each issue based on the URL type:

• For 404s, decide whether to restore the page, redirect it to the closest relevant live equivalent, or leave it as a 404 if the page was intentionally removed and has no meaningful links or traffic.
• For pages that are permanently removed and should not come back, consider using a 410 Gone status code instead of a 404. This gives search engines a clearer signal that the URL has been intentionally removed.
• For redirect chains, update the first redirect to point directly to the final destination URL.
• For redirect loops, fix the redirect rule causing the loop.
• For internal links pointing to redirected URLs, update the source links so they point directly to the final URL.
• For 302 redirects, replace them with 301 redirects when the move is permanent.
• For soft 404s, either improve the page so it serves a real purpose, redirect it to a relevant live page, or return a proper 404 or 410 status code.

After the cleanup, remove dead URLs from XML sitemaps.

For SaaS businesses, especially fast-moving startups that undergo multiple redesigns, migrations, feature launches, and pricing changes, this cleanup should happen right after each change goes live, so issues are caught immediately rather than accumulating into broader crawl and internal linking problems.

Issue 6: Schema Markup Is Missing or Doesn’t Match the Page Type

Schema markup is structured data that tells search engines what type of content is on a page, beyond what they can infer from the HTML alone. For a SaaS company, a well-implemented schema can unlock rich results and rich snippets in the SERPs and feed into Google’s knowledge panel for the brand.

Most SaaS websites have either no schema markup at all or generic schema generated by the theme or website template they use (usually the case on WordPress websites). 

The implementation is inconsistent across page types, and the schema often doesn’t match the page’s actual content.

The most common patterns we see are:

• No schema markup at all. The site has no structured data on any page, which doesn’t break anything but means it misses out on rich results opportunities and provides less context to AI search products.
• Wrong schema type used. A blog post tagged with Product schema, or a feature page tagged with Article schema. The schema technically validates but doesn’t accurately represent the content.
• Required fields missing. Schema types have required and recommended fields. When required fields are missing, the schema may not be eligible for rich results, even when it validates as syntactically correct.
• Product or SoftwareApplication schema added to the homepage to surface star ratings. A common misuse of schema on SaaS sites is adding the Product or SoftwareApplication schema with aggregateRating fields to the homepage, often pulling those ratings from third-party review sites such as G2, Capterra, or TrustRadius. Google’s structured data guidelines explicitly prohibit this. The reviewed item must be the main subject of the page, the ratings must be backed by reviews visible to users on the same page, and the reviews can’t be self-serving. Sites that misuse review markup this way can lose rich result eligibility across the entire site, often without warning.

How to Fix?

Audit the existing schema, then implement appropriate schema types for each page type at the template level so that every page of the same type receives the same treatment automatically.

• Audit: Run the Rich Results Test and Schema Markup Validator on a representative page of each template (homepage, blog post, feature page, integration page, comparison page, pricing page, help doc). Note what schema exists, what’s missing, and what’s flagged with errors.
• Implement at the template level: Add schema to each page template so new pages inherit the correct markup automatically. Don’t add schema page by page.

The most common schema types for SaaS sites are:

• Organization site-wide: company name, logo, social profiles, contact information. Add to every page that describes the organization (homepage, about page, etc.) through a global template.
• Article or BlogPosting on blog posts: author, publish date, modified date, headline, image.
• SoftwareApplication on product or feature pages: name, application category, operating system, offers. Only include aggregateRating if the page itself displays legitimate first-party reviews. Don’t pull ratings from third-party sites like G2 or Capterra or use self-reported numbers.
• HowTo on procedural content like setup guides or tutorials.

For SaaS sites, schema is a low-effort, low-risk investment that complements high-quality content with meaningful upside in search visibility and click-through rates from rich snippets. The goal shouldn’t be to add schema everywhere for its own sake, but to ensure every page type has schema that matches its purpose, and that the markup actually reflects the content visible to users.

Issue 7: Chasing Perfect PageSpeed Scores

Core Web Vitals are rarely the main ranking bottleneck for SaaS websites.

Most modern SaaS sites already use decent hosting, CDN delivery, caching, compressed assets, and frontend frameworks that make pages fast enough to meet basic performance thresholds for site speed and load times. 

Because of that, we rarely see cases where moving a page from a “good” PageSpeed score to a perfect score materially changes rankings. The issue is that teams often focus on the overall PageSpeed score rather than the specific performance problem affecting user experience on the page.

The most common patterns we see are:

• Oversized HTML documents: Googlebot can read large HTTP responses, but that does not mean large HTML documents are good for performance. HTML documents over 2MB often point to bloated templates, inline base64 images, large JSON blobs in the page source, excessive inline CSS or JavaScript, or uncompressed HTML.
• Hero image and web font issues hurting LCP: Largest Contentful Paint is often the hero image or above-the-fold heading on a SaaS marketing page. If the hero image is too large, served in the wrong format, not preloaded, or paired with render-blocking web fonts, the page can feel slow even if the overall score looks acceptable.
• Cumulative Layout Shift from cookie banners, popups, and lazy-loaded content: CLS often comes from elements that load late and push existing content down. Cookie consent banners, popups, embedded widgets, and lazy-loaded images without dimensions are common causes.
• Render-blocking JavaScript and CSS: Scripts and stylesheets that block the initial render delay everything else. This often comes from third-party tags loaded synchronously, large CSS files without critical CSS extraction, and JavaScript that is not deferred or loaded asynchronously.
• INP issues from heavy JavaScript: Heavy JavaScript bundles, third-party scripts, chat widgets, and analytics libraries can make a page feel unresponsive even when LCP and CLS are fine.

How to Fix?

Treat Core Web Vitals as a threshold to pass, not a score to perfect. Start by auditing different page types, not just the homepage.

Then fix the issue based on what the audit shows:

• For oversized HTML documents: Compress HTML with gzip or Brotli, move large JSON blobs out of the initial HTML, replace inline base64 images with proper image URLs, and reduce unnecessary inline CSS and JavaScript.
• For LCP issues: Compress hero images, serve them in WebP or AVIF, set explicit width and height attributes, preload the LCP image, use font-display: swap, and preload critical fonts where needed.
• For CLS issues: Add explicit dimensions or aspect-ratio CSS to images and embeds, reserve space for cookie banners and embedded widgets, and avoid inserting content above existing content after the page loads.
• For INP issues: Break up long JavaScript tasks, reduce unused JavaScript, load third-party scripts with async or defer, audit tracking libraries, and move non-critical JavaScript out of the initial bundle.
• For render-blocking resources: Defer non-critical JavaScript, extract and inline critical CSS, reduce large stylesheet dependencies, and delay non-essential third-party scripts like chat widgets, heatmaps, and marketing tags until after the page is interactive.

For SaaS sites, page speed optimization should be targeted rather than exhaustive. Fix the templates and scripts that create real performance problems on important web pages. Don’t spend weeks chasing a perfect score when the site already passes Core Web Vitals, and the same engineering time could be spent on higher-impact structural SEO efforts.

Issue 8: XML Sitemaps and Robots.txt Don’t Match What Should Be Crawled and Indexed

XML sitemaps and robots.txt files are supposed to help search engines understand which URLs should be discovered, crawled, and indexed.

We often see XML sitemaps that include redirected URLs, 404 pages, noindexed pages, parameter URLs, staging URLs, thin CMS-generated pages, or old blog URLs that should no longer be submitted.

At the same time, robots.txt files sometimes block important sections of the site from being crawled, including blog directories, integration pages, JavaScript assets, CSS files, or documentation pages that should be accessible to search engines.

The most common issues we see are:

• Noindexed URLs in the XML sitemap: These pages are being submitted to Google even though the page itself says not to index them.
• Redirected URLs in the XML sitemap: The sitemap includes old URLs that now redirect elsewhere instead of the final canonical URLs.
• 404 or 410 URLs in the XML sitemap: Deleted pages remain in the sitemap long after they have been removed.
• Parameter or filtered URLs in the XML sitemap: URLs with filters, sorting parameters, tracking parameters, or internal search parameters get submitted even though they are not meant to rank.
• Important pages missing from the XML sitemap: Feature pages, integration pages, comparison pages, or new blog posts are live and internally linked, but not included in the sitemap.
• Robots.txt blocking important content: Blog directories, docs, integration pages, or page assets are blocked from crawling, even though they need to be accessible for proper rendering and indexation.
• Robots.txt used where noindex would be more appropriate: If a page is blocked in robots.txt, Google may not be able to crawl the page and see the noindex directive. For pages that need to be removed from the index, a page-level noindex is usually the cleaner solution.

How to Fix?

Audit your XML sitemap and robots.txt file at least quarterly, and after any redesign, migration, CMS change, or major template launch.

Your XML sitemap should only include URLs that are:

• Canonical
• Indexable
• 200-status
• Important enough to be discovered and crawled
• Not blocked by robots.txt
• Not parameterized or duplicated versions of another page

Remove any URLs that are noindexed, redirected, broken, canonicalized to another URL, blocked by robots.txt, or not meant to rank.

Then review the robots.txt file to ensure it is not blocking important pages or assets. In most cases, SaaS sites should avoid blocking major content sections like /blog/, /features/, /integrations/, /compare/, /solutions/, /docs/, or /help/ unless there is a clear reason. 

While reviewing robots.txt, also confirm the sitemap URL is declared at the bottom of the file (e.g., Sitemap: https://www.yoursite.com/sitemap.xml) so crawlers can find it even when they don’t check Google Search Console.

Also, avoid using robots.txt as your primary indexation control. Robots.txt controls crawling, not indexing. If a page should not appear in search results, use a noindex directive instead, and make sure the page is crawlable long enough for search engines to see that directive.

For SaaS websites, the sitemap should list the URLs you want search engines to crawl and index, and robots.txt should only block areas that genuinely should not be crawled. 

Issue 9: Staging, Preview, or App URLs Are Indexable

For SaaS companies, this issue usually becomes more important as the engineering and marketing stack gets more complex. The more environments, preview systems, CMS collections, and app surfaces the company has, the easier it is for non-production URLs to leak into search.

There may be a production marketing site, a staging site, CMS preview URLs, Vercel or Netlify deployment URLs, a documentation subdomain, a help center, a public product app, partner landing pages, and old campaign pages that were never fully removed. Over time, some of these URLs get discovered and indexed by search engines.

The issue is not just that these pages look messy in Google but that they can create duplicate versions of important marketing pages, expose thin or unfinished content, and make indexation harder to manage.

The most common patterns we see are:

• Staging subdomains indexed in Google: URLs like staging.company.com, dev.company.com, or test.company.com appear in search results.
• Preview deployment URLs indexed: Vercel, Netlify, Webflow, or CMS preview URLs get indexed alongside the production page.
• Duplicate marketing pages on alternate hosts: The same homepage, feature page, or blog post exists on both the production domain and a staging or preview URL.
• Public app URLs showing in search results: Product app URLs, login-gated pages, onboarding screens, or public user-generated pages get indexed even though they are not part of the marketing site’s SEO strategy.
• Old campaign or landing page environments still live: Temporary pages created for launches, ads, partner campaigns, or sales enablement stay accessible long after the campaign ends.
• Sitemaps include non-production URLs: XML sitemaps accidentally include staging, preview, or alternate-domain URLs instead of only canonical production URLs.

How to Fix?

Start by auditing what Google has indexed across your main domain and common subdomains.

Search for indexed URLs across different environments, like:

• site:staging.company.com
• site:dev.company.com
• site:company.vercel.app
• site:company.netlify.app
• site:app.company.com
• site:preview.company.com

Then decide how each environment should be handled.

For staging and preview environments, the cleanest fix is to require authentication so these URLs aren’t publicly crawlable in the first place. If authentication is not possible, add a noindex directive at the template level and make sure the pages are not included in XML sitemaps.

For duplicate production content on preview or deployment URLs, add canonical tags pointing to the production URL, but don’t rely on them alone in staging environments. If the environment should not be public, block access or noindex it.

For app URLs, decide whether the content has search value. Most authenticated app pages should not be indexable. Public app pages, user profiles, dashboards, reports, or shared assets should have a clear indexation policy based on whether they are useful landing pages or thin or duplicate content.

For old campaign pages, either keep them live and indexable if they still serve a purpose, redirect them to the closest relevant page, or noindex or remove them if they are no longer useful.

After cleanup, remove non-production URLs from XML sitemaps, and use the Removals tool only when you need to temporarily hide sensitive or urgently problematic URLs until the permanent fix takes effect.

Improve Your Technical SEO With Grow and Convert

Across hundreds of B2B SaaS engagements, we’ve found that technical SEO lays the foundation for the rest of the work to compound. When the structural decisions are right, every piece of content you publish, every internal link you add, and every keyword you target builds on a stable base. When they’re wrong, the rest of the work fights against a site that isn’t built to rank.

That’s why we treat technical SEO as a finite set of decisions to make well, not an ongoing retainer-level discipline. We make recommendations and fix the structural issues once, and audit the maintenance items quarterly. Then,  we put our time and budget into the work that actually moves the pipeline, like keyword research, content production, internal linking, and external link building tied to keywords that drive trials, demos, and revenue.

For more on the broader SaaS SEO strategy that technical work supports, see our articles on B2B SaaS SEO and Pain Point SEO.

• Our Agency: If you want to hire us to execute a conversion-focused SEO strategy built around driving demo requests, trial signups, and pipeline, not just traffic, you can learn more about working with us here.
• Join Our Team: If you’re a content marketer, writer, or SEO who’d love to do SEO and content marketing in this way, we’d love to have you apply to join our team.
• Our Course and Community: Individuals looking to learn our agency’s SEO and content strategy and become better marketers, consultants, or business owners can join our private course and community, taught via case studies and presented in both written and video formats.

This made some waves. It was syndicated or summarized by outlets like Search Engine Land, which spurred more articles from smaller brands like this agency who straight up called this a “black hat SEO” tactic (eye roll).

We also heard about it directly from clients who sent us the article or told us other advisors told them these product category list posts weren’t helping. 

We saw Lily’s article when it came out and discussed it in depth internally. Lily Ray is a well-respected voice in the SEO community, including us, but in this case, our evidence strongly suggests her takeaway is wrong.

Across dozens of clients and hundreds of articles, we see no evidence of Google cracking down on self-promotional listicles. We also think the evidence in her article itself doesn’t support this thesis, as we explain below. 

What we think is actually happening is simpler: Google is cracking down on bad content. AI slop or human slop. Bad content absolutely hurts your SEO long term, whether it comes in the form of self-promotional listicles or anything else. And frankly, this isn’t new. It’s been true for decades.

In our client data, we’re seeing self-promotional listicles be just as effective now as they were years ago when we first started talking about the importance of this kind of bottom-of-funnel content. The key is that, like all content, they need to be written well. They need to be honest, and they need to be specific and detailed enough to be genuinely useful. 

We discuss all of this with data and examples below. 

Issue #1: Product List Posts Don’t Have to Be ‘Self-Promotional Listicles’

Before we get into the data, there’s an important detail that dictates whether your category list post is a misleading “self-promotional listicle” or not: force ranking your list and calling yourself the best. 

This is how Lily defines a “self-promotional listicle”:  “ranking the best companies or products in their niche and placing themselves in the #1 spot. The common thread is that the company publishing the blog post ranks themselves, and/or their own products, in the top position.”

Even worse is if you then pretend like the list is based on some objective analysis where you evaluated all the options and ended up “objectively” concluding your product is the best. 

We absolutely agree with Lily that this type of content is misleading. It’s also unnecessary.  

We routinely get all the benefits of ranking for these valuable “best X software” type keywords without positioning our clients’ articles as a ranked list of the best products. Searchers of these keywords are just looking for a list of good options, with clear explanations of which product is good for which situations. That doesn’t need to come with a ranking. 

In fact, we think not force ranking your list better fits the search intent of these keywords because not every product is good for all situations. You don’t know the specifics of every reader’s situation, so it’s presumptuous to call any product “the best”.

Discussing Your Product First Doesn’t Have to Mean Saying It’s The Best

Now, as a marketer, you should discuss your product first, simply because eyeballs drop the further down the page you go. You’re just giving yourself less exposure if you put yourself last. 

But as our examples below show, you can do that honestly and transparently. Many of our introductions include some version of “We’ll start with a deep dive into our own product and how we designed it to fulfill the criteria above.”

That’s simply telling the reader: “We’re proud of our product. Here are the choices we made when building it, who it’s a fit for, and why. And if it’s not right for you, here are other major players in the space.” What’s misleading about that? 

We’ve ranked for hundreds (maybe even a thousand or more) keywords like this over the years for 100+ clients. We never position them as force-ranked lists, and our articles rank for the keywords and convert extremely well (examples below).

Issue #2: There’s No Evidence That List Posts Are Being Dinged by Google

The core evidence in Lily’s article for a possible Google “crackdown” on list posts is showing declining traffic graphs of SaaS company blogs and noting that those blogs happen to have hundreds of “self-promotional listicles”.

But the self-promotional listicles only make up a tiny fraction of the total content on those blogs. She never isolates and examines the rankings and traffic of just those listicles. 

Here’s a table summarizing the traffic decline, but importantly, the number of self-promotional listicles versus total posts of every example in her article: 

Look at the last column. In all of the examples, the self-promotional listicles make up less than 10% of the total posts; for most, less than 5%. For some, less than 1%. So if the overall blog traffic is declining, how do we know it’s because of these listicles? 

We don’t.

To Lily’s credit, she never directly says any of this data proves anything is wrong with self-promotional listicles. It’s just insinuated. But the insinuation is strong. 

I mean, just look at the title and subtitle:

It’s not ambiguous what this is implying.

And as I mentioned in the intro, even though her post is implying and not proving, it got picked up by a bunch of other outlets. Many of those outlets went further than her to conclude outrageous things like saying these listicles are “black hat SEO”. The insinuation is strong enough to get people to jump to that conclusion. That matters because people will start changing their marketing strategy based on these insinuations. 

So what data would we need to see to really prove this? 

Well, to start, if Google is indeed cracking down on self-promotional listicles, you should see a decline in their rankings. Otherwise, what does a “crackdown” by Google even mean if those pieces aren’t being pushed down or off the SERP? 

We have that data from dozens of clients, and we’re simply not seeing any kind of widespread rankings hit on these articles. We’ll show a few examples to illustrate.

Our Data: Multiple Examples of Self-Promotional Listicles Doing Just Fine

Here are multiple examples of actual rankings of product category list posts from many of our clients. The data is from Ahrefs rank tracker, which we use internally. It’s never perfect, but directionally good enough. I’m showing a graph of the ranking position over the past year or so, including the late Jan 2026 period where Lily is saying all of the declines in her examples occurred. 

B2B tour operator software

This one has been ranking number one since the spring of 2025. You can see that over the course of a year, it’s been sitting at #1 almost the entire time. 

It’s also performing well on AI visibility, being cited and getting our client a brand mention for various variations of that term: 

Trucking management software

Also published in the spring of last year (about a year before I’m writing this) and after the first month, it consistently hovered in the top 10 for the entire year. No signs of any cracking down. 

Retainer management software

Steady for Nov 2025 to now. Not a sign of a dip in late January 2026. If anything, it seemed to move up a couple of spots then. 

AQM software 

Totally steady. I included this to show an example of natural ranking fluctuation, which you can see in April 2026 here. Google briefly removes the blog post, but after a few weeks, it moves right back to its #1 spot. This is normal fluctuation. 

Best private school for autism

Here’s a non-SaaS example. Also a natural ranking fluctuation in October 2025, but it goes right back to #1. No sign of any crackdown in late January. 

Best concussion clinics

Here’s one in healthcare, a field where Google is notoriously strict. This one has a bunch of other URLs Google tests for this keyword, and an entire month in Oct 2025 where the blog post fell off the SERP. But once again, it came right back. No late January dip. 

There are more…

These are just a few examples. We have tons of others. We debated how to show this data and whether we needed some large statistical analysis of hundreds of these posts in case people wondered if we cherry-picked the few that happened to maintain their rankings. We ultimately decided it wasn’t worth the effort or the downside of exposing a massive amount of client content strategy just to prove the point.

We think these six examples across different industries are enough to show that no widespread crackdown on these types of posts appears to be happening. You can absolutely still achieve strong, stable rankings for valuable category keywords without seeing signs of decline.

If you don’t trust our data, that’s fine; we urge you to just look at your own. Either you’ve never published product category list posts, in which case, why are you even reading this far? Or you have, which means you can look at your own rankings; you don’t have to trust us. 

What If It’s Just AI-Generated Content That’s Getting Dinged?

One of the most notable parts of Lily’s article is at the end, when she notes a couple of other common characteristics of the blogs she examined. One key pattern is that she ran them through an AI writing detector, and it said all had a 100% confidence that the text was AI-generated.

That’s a massive red flag. Folks have published studies showing ample evidence that Google is cracking down hard on sites with a heavy amount of low-quality AI-generated text. 

That alone could explain the blog-wide traffic declines in her examples. 

But I also want to highlight qualitative differences in writing quality between her examples and the posts we write for clients. This could also be hurting long-term ranking ability, whether or not a human or AI wrote them. 

Most ‘Self-Promotional Listicles’ Are Just Bad Writing (Examples)

Lily shares a few examples of specific posts in her article, which show clear examples of how not to write these bottom-of-funnel product category posts. 

First, there’s a post on top-rated test management tools. 

Look at the opening sentence: 

This is corporate marketing gobbledegook that doesn’t say anything useful. It also has questionable claims and unnecessary insertions of “2025”:

• No evidence or reasoning is given to support the claim that software delivery cycles started accelerating in 2025. It’s almost certainly not true. 
• In fact, test management tools have been a thing for a long time. Why are we even talking about 2025?
• Is “software delivery cycles accelerating” really the reason people need these tools? Software testing tools have been around for decades. 

The first two H2s continue this largely non-informative, corporate jargon writing:

Issues: 

• No one Googling “top test management tools” needs to be told what they are.
• Even more inane is asking “Why use them?” The searcher is already looking for them; they know why they’ll use them. 
• Once again, nothing magical happened in 2025. They’re just blatantly forcing the date into it because someone must have told them “it helps SEO”.
• So much corporate jargon: “organizations are adopting…to meet growing demands.” No one talks like this. 

This “Key Features” section is similar to something we’ve included in our bottom-of-funnel pieces for a long time: help the reader understand what they should look for when evaluating options.

But the above list is not useful. The descriptions are too short and generic. You can contrast this with how we do it in our examples below. 

Overall, this is just bad content, and bad content hurting your SEO long term is a tale as old as time.

This kind of writing exists throughout the examples Lily shares in her post. We won’t critique all of them. You can go find those articles if you want to read more. 

Our Examples: How We Write These Bottom-of-Funnel List Posts

Here’s an example of one of our bottom-of-funnel list posts, in this case targeting the keyword “trucking management software” (full article if you want to read it).

Here’s the full intro and the start of our client’s section:

A few important things to note in how we position this piece in the intro: 

1. We never say this is a ranked list
2. We never say Toro’s software is “the best” 
3. We never say we’ve “tested” all the others
4. We never allude to any sense that this is an objective third-party evaluation

Instead, we just describe what we know about the space (trucking management software, TMS) and steer the discussion to the needs of Toro’s target customer: bulk haulers, a subset of truckers. 

You can see this directness in the first H2. It doesn’t say “Toro is the best!” It says it’s software for bulk haulers, which is both true and notable. Most trucking management software is generic, so our discussion of what bulk haulers need is genuinely valuable to the reader. Their needs are different, and that discussion doesn’t exist online.

If you are a bulk hauler looking for a TMS, this is genuinely helpful content, even though it’s clearly selling Toro. 

See how the “self-promotion” isn’t, by itself, a negative? It’s if the company lies or pretends to be objective, or pretends they “tested” all the options, that it comes across as cringey (at best) or dishonest (at worst).  

In fact, for these bottom-of-funnel queries (like “trucking management software”), it’s our strong belief that the searcher is looking to be sold to. When you’re researching products, in particular complex or expensive options like many B2B purchases are, you want to understand the features, benefits, and differentiators of each option at a deep level. So if one of the brands has an article that basically lays out:

1. Here’s our view on what matters in this space
2. Here’s how we designed our product to align with that view 

That’s genuinely useful. It doesn’t matter that the content lives on the brand’s site. The reader knows where you are. They know the brand is selling their product, and that’s fine; they’re expecting a sales pitch and will use that pitch to help their decision-making. 

You can see this in action later in the piece when we introduce Toro:

We list what features you need as a bulk hauler, with real explanations, not just AI-sounding generic descriptions (see the second bullet as an example), and say how Toro meets those criteria. 

Then, throughout that section, we go into extreme detail about exact features and how they help bulk haulers:

This level of detail helps AI search in particular because AI recommendations rely on matching detailed pain points with solutions.

This is genuinely useful content for the target audience without any of the unnecessary pretending that you won some contest and ended up first.

We think content like this is aligned with Google’s long-term goals (provide truly valuable content to users for each keyword), and we’ve written content for these bottom-of-funnel category keywords in this way for 8+ years. All of them have survived countless Google updates.

And even though this piece isn’t a listicle, it wouldn’t feel complete without a little self-promotion. So if you want to talk about publishing honest, valuable content targeting high-converting keywords for your brand, you can reach out here. 

For example, if you’re a software company that helps small business owners track employee time off, then a blog post or landing page written to rank for the keyword “best employee time off tracking software” is BOFU content. People searching for queries like that are ready to make a purchase.

That conversion potential is BOFU content’s primary selling point, and we’ve been preaching the benefits of bottom of the funnel content for years.

We think too many brands overfocus on top of the funnel content and would be better off prioritizing BOFU instead. We have plenty of data to back that up, including a conversion analysis of 95 articles we’ve written for clients.

We’ve noticed, however, that even when marketers publish BOFU content, they make several mistakes:

• Using the wrong criteria to pick topics. In our experience, marketers pick keywords using one of two frameworks: (a) researching what the competition is ranking for and (b) prioritizing keywords by search volume. Both criteria are worth considering as you develop your strategy, but neither is a strong enough indicator on their own when choosing which topics to write about. 
  
  • First, just because a competitor ranks for a keyword doesn’t mean that keyword is driving leads, or that it would drive leads for you. A competitor might be ranking for topics with no buying intent. They also might be ranking for BOFU topics where you cannot meaningfully differentiate yourself (i.e. topics where you’re not going to connect with your target audience).
    
  • Second, higher keyword volume doesn’t translate to more conversions. In fact, we have seen organic leads grow for our clients when targeting mini-volume or even zero-volume keywords. (Keyword volume becomes even less meaningful when your goal is to show up in AI search. No one knows what queries are being typed into AI tools like ChatGPT and Gemini, but we do know that ultra-specific writing is showing up in AI search results.)

• Writing bad content. This can negate the benefits of publishing BOFU content. For example, the content may not match search intent and fail to show up in search engines, or it may not address specific customer pain points and struggle to convert. 
  
  At Grow and Convert, we’ve always championed the importance of addressing the exact pain points your customers have, not surface-level overviews, in our content. This approach has paid off for our clients in the past, and is continuing to pay off as more customers use AI search. This is because AI search results tend to be very specific to the user, based on chat history. We wrote about this aspect in our piece on Invisible Prompts. The more specific and precise your content, the more likely LLMs and Google are going to connect you with your target audience.
  
• Not knowing how to make strategic choices based on what’s working. As you publish more content on your site, you’ll start tracking success through various metrics, such as page views, rankings, AI visibility, and new leads. But often companies focus on the wrong metrics or miss the full picture — like seeing declining Google traffic and concluding their BOFU content isn’t working, while not realizing that the same content is being recommended in AI search tools they aren’t tracking.

At Grow and Convert, we create organic channels that bring in qualified leads for brands. We do this by writing bottom of the funnel content that matches what your target customers are looking for.

We publish this content on a client’s blog so it can rank in search engines (both traditional and  AI search). This creates an organic channel that drives conversions. We’ve been doing this for a decade, across a wide variety of clients, and have a lot of data on what bottom of the funnel content types convert the most. 

In this article, we look at different types of BOFU content (with examples of content we’ve written for clients), how to find the best BOFU topics for your company, and how we write BOFU content.

Table of Contents

• 5 types of BOFU content. We cover five types of BOFU content and show actual data on how well they convert for our clients.
• How to come up with bottom of the funnel content ideas. We share how we ideate topics with new clients and why we don’t start with keywords, even though SEO is key to our process.
• How to write bottom of the funnel content. We discuss our research-based method, which lets us write high-quality, rankable content across various industries.
• Measuring your BOFU content and making data-informed decisions. We cover what metrics to track — including conversions, rankings, and AI visibility — and how to use that data to decide which content to double down on, re-optimize, or deprioritize.

5 Types of Bottom of the Funnel Content Shown to Drive the Most Leads

To help us find out what type of BOFU content converts the most, we analyzed 95 articles across clients, totaling over 123,000 organic pageviews and 4,687 conversions. 

Across that dataset, we identified five types of content: 

1. Category keywords
2. Side category keywords
3. Category keywords with a layer of specificity 
4. Comparison and alternatives keywords
5. Jobs to be done keywords

(Read more about the conversion rates across these keywords.)

1. Category Keywords (4.85% Average Conversion Rate)

These are queries that describe exactly what you offer. Think of topics like “best accounting software” or “men’s running shoes.” For Grow and Convert, a category keyword would be “content marketing agency,” because that’s what we are.

Most people don’t expect blog posts to convert at multiple percentage points, but our data shows they can when they rank for the right keywords and are written to sell the product. For example, Geekbot, a past client and a solution for running online standup meetings, ranks #2 for “Slack standup bot” via a post we wrote for them, and that single blog post has a lifetime conversion rate of 8.36%.

To be clear, this is extremely high, and essentially unheard of for a typical blog content strategy (which heavily skews to top of funnel content). For example, in one of our clients we compared the conversion rate of BOTF vs. TOF content. 

Their TOF content converted at just .19%, while their BOTF content had a 4.78% conversion rate.

We’ve seen this pattern across clients: when you target variations of your core product keywords, most variations convert well. A video marketing client we ranked on page one for app, software, and service variations of their category saw conversion rates of 5.73%, 3.31%, and 3.00%, respectively. Not every variation will convert equally, but category keywords consistently convert at 3–5% or higher across our clients.

When you’re developing your BOFU content strategy, you want to identify your main category keywords.

2. Side Category Keywords (1.94% Average Conversion Rate)

These are queries that show buying intent for a secondary feature or use case — not your core offering. Many companies build content to promote these offerings, and while these keywords don’t usually convert as well as your main category keywords, they still can drive leads.

For example, one of our clients sells digital asset management (DAM) software. Based on our conversations with their sales team, we wrote a BOFU post on storing video assets, a specific use case within their broader product. That post converts at 1.45%. That’s still far higher than typical top of funnel content, but their posts targeting general DAM keywords convert at 3%, 6%, and even 15%. The difference comes down to fit: users who only need video storage could opt for something simpler, whereas users searching for general DAM software need exactly what this client offers.

3. Category Keywords + Layer of Specificity (2.96% Average Conversion Rate)

These are queries that add a word or two that make them more specific to a particular feature, use case, or customer type. For example, adding “for small business” to “best accounting tools,” or adding “for trail running” to “men’s running shoes.” You’re targeting searchers who already have some knowledge of the category and are searching for a very particular solution.

Here’s a recent example from our client that works in the private education space. Their main category keywords are topics like “best private schools in california” or “best private schools in the US.” But a key part of their offering is working with students that have learning differences, such as dyslexia, ADHD, and autism.

For this client, we targeted the topic “dyslexia private schools.” That’s a more specific query than “best private schools.” It has a smaller audience and smaller search volume, but it’s been a major driver of conversions for our client. It has an average conversion rate of 1.11%.

Taking this learning, we also targeted “private schools for autism,” which while bringing in less traffic, has brought in more conversions with a higher average conversion rate of 2.32%.

Note this kind of specificity has an additional benefit for AI search, where LLMs recommend products that are highly specific solutions to the users situation or problems. More on this below.   

4. Comparison and Alternatives Keywords (8.43% Average Conversion Rate)

These are queries that mention competitor names, indicating the searcher is aware of other brands in your industry, is looking for alternatives, and could be a good candidate for your solution. This includes keywords like “salesforce alternative” or “Salesforce vs. Pipedrive.”

These were the highest converting keyword types in our analysis. This makes sense: these searchers already know about your industry, are familiar with competing products, and are actively evaluating options with a serious intent to buy.

But you want to be strategic about which competitor keywords you target. From our data, keywords mentioning direct competitors convert significantly better than those featuring loosely related competitors. Although the average conversion rate is 8.43%, the majority of the 23 competitor keywords we analyzed actually converted at less than 4%. The closer the competitor is to your product, the higher the conversion rate.

That said, even conversion rates between 1–4% are far higher than top of funnel content. And many comparison and alternative keywords have very low search volume — but as we’ve seen with our analysis of mini-volume keywords, the high conversion rates make them well worth targeting.

VS Keywords

These are a specific flavor of comparison keywords — queries comparing two solutions head-to-head, like [Brand #1] vs. [Brand #2]. The searcher is trying to decide between two specific options they’re already considering, which is why these converted at 5.45% in our analysis.

5. Jobs to Be Done Keywords (2.44% Average Conversion Rate)

These are queries that describe a job to be done (which is a problem to be solved) that your product offers the best solution for. These are mostly “how to” queries: how to organize design files, how to do a poll in Slack, how to get video testimonials from customers, etc. Although these users may not be aware of your brand or competitors, they still have the potential to buy because they have a job or problem that you solve.

While these types of keywords have a lower conversion rate than the above buckets, 2.44%, they still convert leagues higher than typical top of funnel blog content. Also these JTBD keywords often have higher search volume, which means more traffic. 

And when the job in the query maps directly to what your product does, these keywords can convert incredibly well. For example, “collect video testimonials” — a keyword we targeted for a video marketing client — converts at 9.71%, because the searcher needs software to do that exact task and our client’s product is built for it. 

The takeaway is that looking at the highest average conversion rate and focusing only on those keyword types is not a strategy. There is a large variety in what converts well, based on the client and the types of customers they serve.

How to Come Up with Bottom of the Funnel Content Ideas for Your Business

Often agencies approach content ideation in one of two ways: they open a keyword tool and target whatever keywords have the highest search volume, or they look at what competitors are ranking for and try to cover the same topics. 

Neither approach is wrong on its own, but when your goal is to build an organic channel that generates qualified leads, they’re not the best starting points. High-volume keywords may not align with what your ideal customers are actually searching for when they’re ready to buy. And just because a topic works for a competitor doesn’t mean it’s the right fit for your business — their customers, positioning, and strengths may be completely different from yours.

Instead, the more effective approach is to start by asking: what problems do our customers have that our product solves?

At Grow and Convert, we begin every new client engagement with a series of interviews with people from different departments,  sales, customer success, founders, anyone who has direct interaction with customers. The goal is to understand the details of the product, the ideal customer personas, and most importantly, the specific pain points that the product solves.

Outside of speaking directly with customers, this is the fastest way to identify topics that resonate with potential buyers and naturally position your product as the solution.

These interviews are what allow us to choose topics based on how your customers actually search, rather than defaulting to whatever has the highest volume, mimicking what competitors are doing, or even relying on average conversion rates amongst types of BOFU content.

For example, one of our clients made a straightforward employee time off tracking tool for small business owners. Through our interviews with their team, we learned that most of their customers had been managing time off in spreadsheets and physical wall calendars. They weren’t coming from competing software. That insight changed the entire content strategy. Instead of prioritizing comparison and alternative keywords, which have the highest average conversion rate in our data, we focused on JTBD keywords like “how to track employee time off,” because that matched how their customers were actually searching. It worked: we grew a profitable organic channel for them by targeting keywords that matched their specific customers’ pain points.

Find Keywords That Correspond to Your Topics

Once you have topic ideas grounded in customer pain points, you can plug them into keyword research tools like Ahrefs or Semrush to check search volume and keyword difficulty. The role of keyword tools in our process is to confirm that people are actually searching for the topics you’ve identified, not to generate the topics in the first place.

The most important thing to resist here is letting search volume override buying intent. A keyword with 200 searches per month and high buying intent will almost always outperform a keyword with 10,000 searches per month that attracts people who are just learning about a topic. We’ve seen this repeatedly in client data: the posts that drive the most leads are rarely the ones with the highest traffic.

That said, not every topic you identify needs a Google keyword to be worth pursuing. With more customers using AI tools like ChatGPT and Gemini to find products and services, there’s demand happening in conversations that no keyword tool will ever show you. 

As we mentioned above, LLMs recommend products based on the specificity and relevance of your content to a user’s situation, not based on whether you targeted a high-volume keyword (though we do see a positive correlation between ranking #1 for a keyword and showing up in AI results). So if your interviews surface a pain point that your product solves well but there’s no obvious keyword to match it to, it can still be worth producing content around that topic.

We’ve seen this with clients whose content is now showing up in AI search results for prompts we never could have predicted. We call this new type of search query answering Invisible Prompts, and it’s a big part of why we’ve developed a topic-based approach to GEO that doesn’t rely on keyword volume.

How to Write Bottom of the Funnel Content That Ranks and Converts

Choosing good high-intent keywords to target is the most important step but a close second in importance is how these pieces are written. If the content itself is generic or surface level (for example a very basic listicle of top software in a category) it risks you (a) not ranking and (b) not impressing or converting visitors that do land. 

We’ve noticed this is actually extremely common, and we think it happens because most writers — and now most AI tools — don’t have deep enough knowledge of the product or customer to write anything original. Without that knowledge, they default to copying what’s already ranking in the SERP, which produces content that says the same thing everyone else is saying.

At Grow and Convert, we use a content writing process that helps us overcome these challenges.

• We interview subject matter experts. For every piece we write, we interview subject matter experts from within our client’s organization — sales leads, founders, product managers, whoever knows the product and customer best. This is the step most content teams skip. Without it, writers default to “Google research papers” — reading what competitors have published and repackaging it into a new article that says nothing original. The interview surfaces the specific details, positioning arguments, and customer insights that make a piece actually useful to a buyer. (Learn more about our content writing process here.)

• We conduct a SERP analysis. Before writing, we analyze what’s currently ranking on Google for the target keyword. We review the types of content on page one (listicles, how-to guides, landing pages), identify the subtopics and themes that come up repeatedly, and assess what existing results do well and where they fall short. This tells us what our piece needs to cover to compete, and where there’s an opportunity to differentiate. (Learn more about our SERP analysis process in our article on SEO Content Writing.)

• We use the Grow and Convert Questionnaire, not a formulaic template, to help us outline the piece. Most content briefs focus on logistics — word count, how many internal links to include, which heading tags to use, what Grammarly score to hit. But that kind of information doesn’t help a writer produce great content that will rank and convert.
  
  Our questionnaire is fundamentally different. After the interview, writers answer a series of questions designed to help them synthesize what they learned from the interview and their SERP analysis. Our Questionnaire focuses on understanding the target audience at a level deeper than “our audience is a startup looking for accounting software” and understanding the SERP. 
  
• We draft the piece. The draft is written with the product woven throughout, not just mentioned in a CTA at the end. This means opening with the customer’s pain points, explaining the product’s positioning early to grab the right reader’s attention, and highlighting differentiating features as the foundation of the article. For example, when we wrote a piece for TapClicks targeting “paid search dashboard,” we structured the entire post as a walkthrough of how to create a dashboard using their product, which let us naturally showcase the features that made them different from competitors. (Learn more about our writing style in our post on How to Improve Your Content Writing.)

We’ve also built an AI writing tool, Wave Writer, that helps with this process. Unlike other AI writing tools that only learn your brand voice, Wave Writer ingests your business’s products, features, positioning, and customer pain points. It analyzes a SERP for any given keyword and suggests angles tailored to your business, produces outlines grounded in your actual arguments, and drafts content that says something substantive.

We used Wave Writer to do an SEO analysis for this very post, which you can view here, along with my comments on the output.

Measuring Your BOFU Content and Making Data-Informed Decisions 

As discussed in this article, writing BOFU content is a good strategy to grow your organic channel. When done correctly — that’s writing about the right topics and writing content that ranks — your BOFU content can lead to new leads, free trial signups, demo requests, new customers. 

But a key aspect of developing your BOFU content strategy is knowing what types of content to double down on, and what to deprioritize. At Grow and Convert, we track three things to create a full picture of your organic channel: conversions (our primary metric), keyword rankings and traffic, and AI visibility. 

• Conversions. This is our primary metric. Traffic and rankings matter, but only because they lead to conversions — new leads, signups, demo requests. We track conversions from each piece of content so we can see which posts are actually driving business results, not just pageviews. (Learn how we track conversions in our article on measuring conversions in GA4.)

• Keyword rankings and traffic. Rankings and traffic tell you where you’re positioned, what’s growing, and where you’re slipping. For example, when one client saw a dip in conversions despite still ranking for their target keywords, we discovered they had lost rankings for secondary keywords — related keywords their posts had been ranking for that were quietly driving conversions. Re-optimizing 22 posts brought monthly conversions from 171 to 258. (Read the full process in our article on how to find and use secondary keywords to increase conversions.)

• AI visibility. Traditional search metrics don’t capture whether LLMs are recommending your brand. We track AI visibility at the topic level, not the individual prompt level. Our client Constitution Lending now appears in AI search results for 50+ bottom of the funnel prompts across ChatGPT, Perplexity, and Google AI Overviews — competing with lenders that have been in the market for decades. There’s still a gap in attribution here — it’s more difficult to track customers who found you through AI search — but visibility is the leading indicator. (Read more about our topic-based approach to GEO.)

Over the past eight years, we’ve built organic channels for clients across dozens of industries — from SaaS to healthcare to private education — using the process described in this article. Our BOFU content has driven thousands of conversions for clients, with conversion rates routinely hitting 3-10% on individual posts.

Our clients are now showing up in AI search results alongside competitors with decades of industry presence. If you want to build an organic channel that drives real leads, learn more about working with us.

Their goal was simple: generate qualified leads from both traditional SEO and AI search. 

To achieve this, we: 

• Applied Pain Point SEO to rank for high-buying-intent keywords in Google
• Built a GEO Topic Map that teaches LLMs when to recommend them in AI search

So whether prospects search on Google or ask LLMs (like ChatGPT or Claude) about survey software, they have a strong chance of being recommended as a solution.

Below, we explain the principles behind our GEO and SEO strategy and share examples of our results.

Our Strategy for Gaining Both Traditional SEO & AI Search Visibility for Our Client

The core principle of Pain Point SEO is prioritizing high-buying-intent keywords over the top-of-funnel informational content most brands focus on. Our data shows these bottom-of-funnel keywords convert 10x to 25x better than low-intent queries.

For example, instead of targeting keywords like “What is a net promoter score” or “What is a focus group,” where the reader still has a lot of learning to do before buying, we prioritize keywords like “best survey software for enterprises” or “best customer research tools.”

The same principles apply to AI search visibility, since LLMs rely on web results when users ask for product or service recommendations. But there’s an important distinction.

Unlike Google, where people type short, predictable queries, LLM conversations are longer, more detailed, and often informed by prior interactions. In many cases, these systems have weeks or months of context about a user, which they factor into their recommendations.

As a result, LLM recommendations are more personalized than Google search results.

Take two users evaluating our client’s survey software: a startup running feedback loops with early users and an enterprise looking for affordable scale. If both search “best survey software” on Google, they’ll see similar results. But the same query in an LLM would factor in previous conversations, and the recommendations would differ based on each user’s context.

In AI search, it’s less about ranking for well-defined keywords and more about being a brand the model recommends when a relevant topic comes up. 

That’s a fundamentally different challenge than ranking for short keywords. LLMs need enough context to make a compelling case for your product — and that context has to be relevant to each user’s situation.

We call this approach Topic-Based GEO.

Topic-Based GEO argues that to maximize AI search visibility, you need published content that gives LLMs what they need to position your product as the solution to specific use cases and situations (“topics”).

We visualize the entire universe of topics relevant to your business as a map and call it your GEO Topic Map. 

To build it, we start by brainstorming every use case and pain point someone might ask an LLM about survey software.

The result is our Topic Map: a body of content that addresses every conversation or situation in which an LLM might reasonably recommend our client’s survey software.

For example, our client works with:

• Startups using survey software to gather early customer feedback
• Enterprises with strict security requirements
• DIY market researchers looking for affordable tools

We then created highly specific content for each of these use cases. Below, we share our results.

Examples of Keywords We’re Ranking For & Our AI Search Visibility

Example 1: “DIY market research tools”

One of the first articles we published for our client targets conversations where a user is asking about DIY market research tools, a common use case among their ideal customers.

If we look inside Traqer.ai — the tool we developed to track AI search visibility — we’re showing up in the top 5 positions in ChatGPT, Perplexity, and Google AIO for the topic “DIY market research tools.”

In Traqer, we also see that whenever someone enters a prompt about DIY tools for market research, our article is cited more than any other.

As a bonus, we’re ranking 4th for “DIY market research tools” on Google and generating leads from organic search.

Example 2: “CPG market research companies”

We also want LLMs to recommend our client’s survey software when someone asks for CPG market research companies, since CPG (consumer packaged goods) brands are among the heaviest users of market research.

So, we published a highly differentiated article on “CPG market research companies” that shows how our client’s survey software addresses the specific pain points of CPG brands.

In Traqer, we can see that we’re in the top 3 recommendations in ChatGPT, Perplexity, and Google AIO for prompts related to CPG market research companies.

LLMs also pull from our article more than any other piece on this topic.

We’re also currently ranking 1st for the keyword “CPG market research companies” on traditional Google search.

Example 3: “Enterprise survey software”

A large percentage of our client’s customer base is enterprise clients looking for software that scales well and doesn’t become increasingly expensive as they add users. So, we knew that enterprise survey software is an important topic to show up for on LLMs.

In Traqer, you can see that we consistently rank in the top 5 for queries related to enterprise survey software.

Our article is also the source AI cites most often when users ask about enterprise survey software.

On top of that, the article we published targeting this keyword is ranking 2nd on Google.

These are just a few examples, since we can’t realistically cover every topic we have visibility for in AI search.

Overall, the topic map we built gives our client AI search visibility across 100+ high-buying-intent topics.

Our Client’s AI Search Visibility Versus Competitors

We also think it’s worth comparing our number of AI citations with those of competitors, as this helps put the effectiveness of our AI visibility strategy into context.

As shown in the screenshot below, our article is cited nearly twice as often as SurveyMonkey, Involve.me, and Quantilope on the topic of “AI survey tools.”

On the topic of “enterprise survey software,” AI tools are citing our client more than competitors like SurveyLab and Zonka Feedback.

The same is true for “conjoint analysis software.” We have more citations than Conjointly, OpinionX, and Qualtrics.

These are just a few examples, but the pattern holds across most topics. For most high-buying-intent prompts, AI tools cite our articles more frequently than those of our competitors.

Key Takeaways

This case study highlights two key takeaways for brands looking to appear in LLMs:

1. Prioritize high-buying-intent terms your target audience is likely to use in LLMs. This increases the chances your content is surfaced and cited when generating responses.
   
2. When writing articles targeting high-buying-intent keywords, be extremely detailed about your value proposition, differentiators, and why someone should choose you. This gives LLMs the context they need to present your solution clearly and persuasively.

Improve Your AI Search Visibility with Grow and Convert

Book a call with us to learn more about how we can help your brand show up for high-buying-intent queries in LLMs.

We started working with Level AI in 2024 with the goal of ranking for high buying-intent SEO keywords like “call center customer analytics tools.” In 2025, as AI search exploded in popularity, that goal expanded to include getting LLMs to recommend Level AI when users ask for customer intelligence solutions. 

Across both traditional and AI search, the goal is always the same: drive conversions, not just traffic. That principle guided both Level AI’s SEO and AI search (AEO/GEO) strategy. 

This case study focuses on the AI search component. We’ll show how we earned Level AI brand mentions (not just citations) in 100+ product-centric prompts using our Prioritized GEO strategy. 

If you’d like help improving your AI search visibility, get in touch with us.

How We Achieved AI Search Visibility for Level AI

Prioritized GEO is based on the fact that LLMs almost always search the web to help formulate responses when users ask for product recommendations. This process is known as grounding. Their training data alone is not always up to date or sufficient for these prompts, so they use search to supplement it. 

So, if we want LLMs to recommend Level AI when someone asks for product recommendations in their space, traditional search is the most efficient way to make that happen. For example, if a user asks ChatGPT for voice of customer platforms, it will search the web — and if our article discussing Level AI appears in those results, it significantly increases the chances of being recommended in the response. 

Learn more about Prioritized GEO.

What Keywords We Rank for and Where We Stand in LLMs

Let’s look at a few examples of high buying-intent keywords that are important for Level AI, the content we published to rank for each, where we rank, and the AI prompts we appear in.

Example #1: AI call center monitoring

One of the first keywords we targeted for Level AI was “AI call center monitoring,” and we now rank on the first page for that term.

Our AI visibility tool, Traqer, shows that we consistently appear in the top 5 positions in LLMs when users ask about AI call monitoring solutions. 

Note how Traqer asks about the topic in multiple ways. This is because AI search is not as reproducible as traditional SEO — meaning even slight variations in prompt wording produce different results, and even the same prompt asked at different times can yield different results. 

Traqer accounts for this by asking about a single topic in multiple different ways. We explain this in detail in Topic-Based GEO. 

Side note: We’ve noticed across dozens of brands that visibility on Perplexity and Google AIO often increases before ChatGPT. We hypothesize this is because Perplexity and Google AIO are more heavily influenced by traditional web search results. They function more like search summarizers, so traditional SEO rankings have a greater impact on their output. ChatGPT, on the other hand, blends more of its training data into its responses, which naturally biases it toward larger, more well-known brands. 

Beyond brand mentions of Level AI, we also see LLMs using our article as a source in responses to these prompts.

Here’s an example of Google citing our article and listing Level AI as a solution in its AI Overview for “AI call monitoring solutions.”

Example #2: Call center real time reporting

Another relevant topic for Level AI is “call center real time reporting” because it’s one of the few QA tools that analyzes customer conversations in real time, not after the fact.

Ahrefs shows we rank first for “call center real time reporting” on Google. 

Meanwhile, Traqer shows that we have good AI search visibility across ChatGPT, Perplexity, and Google AIO when someone asks for software suggestions for real-time call center reporting.

We also see that LLMs cite our article more than any other source for topics related to “call center real time reporting.” In fact, across the 6 prompts above, Google AIO, Perplexity, and ChatGPT cite our article a combined 15 times, almost twice as often as any other article. 

This is how you influence LLM responses: create content they discover through web search and use to generate their answers. 

Here’s Google listing Level AI as a popular solution in this space and citing our article first in its AI Overview:

Example #3: Call center analytics software

Another broad topic area where Level AI has a strong value proposition is “call center analytics software.” It’s a popular product category, and appearing in those discussions has long-term value for Level AI.  

So, we wrote an article targeting that keyword, which at the time of writing ranks first on Google.

In Traqer, we see strong visibility for prompts related to call center analytics software on ChatGPT, Perplexity, and Google AIO.

Perplexity recommends Level AI when it’s prompted with questions asking for call center analytics software options, and cites Level AI as one of the main sources it uses to formulate responses.

Example #4: Call center quality assurance tools

The fourth keyword we’d like to cover is “call center quality assurance tools,” where we rank third on Google.

In Traqer, Level AI has strong visibility on ChatGPT, Perplexity, and Google AIO when users ask about call center QA tools.

Although we’re not the top-cited article, LLMs consistently reference us across multiple prompts related to that topic.

We can’t cover every keyword we’ve targeted for Level AI, but these 4 examples demonstrate a clear pattern: ranking well for a keyword on Google gives us AI search visibility for prompts related to that keyword. Keywords we aren’t yet ranking for have little to no visibility in LLMs.

All in all, we’ve targeted 50+ keywords for Level AI so far, resulting in visibility in LLMs for 100+ related high buying-intent prompts.

Examples of AI Using Our Content to Respond to Prompts

Another core principle of our AI search visibility strategy is that the level of detail in your public content helps dictate how LLMs talk about your products and services (full article). 

This means LLMs act like your first salesperson — summarizing your product’s differentiators, value propositions, and arguments, and using them to educate users on what makes you different from competitors and why they should choose you.

As a result, brands need to produce thorough, detailed content that articulates their solution in extreme detail: who it’s for, use cases, features and benefits, pain points it solves, and differentiators versus competing solutions. This way, you feed LLMs your exact value propositions and differentiators so they can make a compelling case for your solution. 

If you publish generic content that echoes what others are already saying, even ranking in LLMs won’t help because it will describe you the same way it describes everyone else, making readers far less likely to convert.

The differentiators and value propositions we focused on for Level AI were:

• Level AI uses semantic intelligence to understand the full context of conversations, making its results more accurate than many alternatives. Most software simply looks for keywords.
• Level AI can detect emotions in customer conversations. Most call center QA software struggles with this.
• Level AI can auto-score all conversations based on the client’s QA rubrics. Teams don’t have to listen to hours of customer conversations and score agents manually.
• Level AI’s deep analysis of conversations uncovers hidden issues and revenue opportunities without surveys, sampling bias, or manual analysis.

As a result, we’d like to share screenshots of LLMs using almost the exact same text in their responses as we use in our articles.

When we ask ChatGPT for accurate AQM software, it emphasizes that Level AI doesn’t just look for keywords but can understand the full context of customer interactions. 

This is known as semantic understanding.

Semantic understanding is the main value proposition we focus on in our articles.

Here, we can see how ChatGPT says Level AI can detect, label, and filter specific emotions like anger, disappointment, worry, and happiness:

We use almost the exact same phrase in our articles:

These are just a few examples, but generally, when ChatGPT recommends Level AI for customer intelligence solutions, it mirrors the same language, value propositions, and differentiators we use in our content.

What This Means for Your Brand: 2 Key Takeaways

1. Target high buying-intent keywords and prompts your target audience is likely to enter into LLMs. LLMs search the web when users ask for product recommendations, so if you rank well for a keyword in Google search, they’re more likely to recommend your brand when users enter prompts related to that keyword.

2. Write highly specific, differentiated content. This gives LLMs the context to position your product or service clearly. Avoid publishing generic content, or your LLM pitch for your brand will sound the same as everyone else’s. Learn more about writing specific, high-quality content.

How to Take This Strategy Further in Your Own Organization

Ranking for high buying-intent keywords with highly differentiated content should give you a strong foundation for LLM search visibility and SEO rankings.

The next step in our Prioritized GEO strategy is Tier 2: citation outreach. This involves identifying the sources AI systems reference most for a given prompt or topic (which you can do in Traqer) and reaching out to those websites to request a mention of your brand. Sites like G2 or Capterra generally require payment, while competitors are often willing to trade mentions.

This works through the same underlying logic as Tier 1: owned content. When LLMs search the web to respond to user prompts, they see your brand mentioned across multiple sources (not just your own website), increasing the likelihood that they mention you more prominently.

That said, there are two mistakes to avoid when doing citation outreach:

1. ​Asking any website in your industry for mentions: We recommend being strategic and focusing on earning mentions in articles that LLMs are already citing.

2. Not controlling the narrative on other websites: We strongly advise writing the product snippet yourself and being specific about your value propositions and differentiators. This way, your messaging is consistent across sources, giving LLMs the information needed to position your solution in a clear and differentiated way.

Increase Your Brand’s LLM Visibility with Grow and Convert

If you’d like us to improve your brand’s LLM visibility and drive conversions, get in touch for a quick call.

In the marketing community, there’s a growing belief that if you rank for these fan-out queries, you’ll consistently get cited by LLMs.

But we wanted to verify and quantify this: How much overlap is there between fan-out query rankings and the sources ChatGPT actually cites? Do the sources cited in a ChatGPT response rank on Google for the fan-out queries? If so, how highly do they rank?

We focused on ChatGPT because, across dozens of clients, it drives the most referred traffic. 

We looked at 100 different buying-intent prompts, each generating 2-4 fan-out queries, and compared the sources ChatGPT cited against the Google and Bing search results for each fan-out query to see how much overlap there actually was.

In total, only 27% of cited sources ranked on Google for the fan-out queries used by ChatGPT. Bing was slightly lower at 23%. Accounting for the ~10% overlap between Google and Bing results, only about 40% of cited sources appeared somewhere in the first 10 pages of either search engine for those queries. 

That means, on average, more than half (60%) of cited sources in ChatGPT answers don’t rank in the first 10 pages of Google or Bing for the fan-out queries.

We also observed a wide range: for some prompts, there was 100% overlap between citations and fan-out query rankings. For others, none of the cited sources ranked within the first 10 pages for those queries.

This data doesn’t invalidate the connection between traditional search rankings and AI search visibility. According to our study, ChatGPT still searches the web for 80%+ of product-related prompts. And as we’ve written about before, these bottom-of-funnel prompts are where AI visibility matters most. 

But it adds to a growing truth about AI search: LLMs are fundamentally less predictable and less “hackable” than traditional SEO.

More specifically, trying to produce single pieces of content to “rank” for specific AI search prompts, in this case by targeting fan-out queries, doesn’t make sense for AI search. It’s fuzzier and less direct than that.

As a result, brands are better off doing authentic marketing — including publishing extremely detailed and thorough content about their products, use cases, features, and benefits — rather than trying to “hack” their way into ChatGPT by targeting specific fan out queries, or as we’ve talked about in previous articles, with tactics like adding FAQ pages, bullet point summaries, or llms.txt files.

You don’t know the true prompts your users are entering. And even if you did, this study shows that targeting fan-out queries gets you cited less than half of the time. So instead of a content strategy aimed at specific prompts, focusing on product-centric content that covers relevant topic areas makes more sense.

Below, we’ll explore these implications further, along with the obvious next question: if only 40% of citations come from the fan-out query SERPs, where does ChatGPT get the remaining 60%?

A Note on Methodology

If you’ve been following us, you’ll know that we believe bottom-of-funnel, product-centric prompts are the only ones worth worrying about for AI search. Top-of-funnel queries are typically answered by LLMs without mentioning or linking to brands or websites. 

All 100 prompts we tested had buying intent (e.g., “best help desk software for small teams,” “top accounts payable automation tools,” or “best standing desk for home office”). The prompts covered B2B, B2C, software, services, and physical products. 

We used the browser UI, not the API, so the data would reflect what a real user sees. 

Of the 100 prompts, 83 triggered a web search, meaning 17% of the time ChatGPT answered without searching the web at all. When a web search was triggered, each prompt generated 2–4 fan-out queries.

To pull Google’s SERPs for the fan-out queries, we used SerpAPI. For Bing, we ran the searches manually because the API returned unreliable results, so the dataset for Bing is smaller.

Finally, we compared ChatGPT’s cited sources against Google and Bing SERPs at the prompt level, by category, and by domain match (vs exact URL match). 

We also analyzed the sources ChatGPT cited that didn’t appear in Google or Bing SERPs to see what we could learn about where those citations come from — and your chances of being one of them.

The Overlap Between ChatGPT Sources and Google’s SERP Varies Wildly from Prompt to Prompt

20 of the prompts had zero cited sources found in the fan-out SERPs — despite the fact that ChatGPT chose to search the web. Only one prompt out of the 100 we tested had 100% overlap. 

A logical follow up question is, are we seeing this wide range of overlap between cited sources and the SERP because ChatGPT relies more heavily on search for some types of prompts vs others (where its training data is enough)? 

So we also grouped them by business type:

While there is some variability, there’s less variation across categories than you might expect. Physical products had the lowest overlap, which is what we’d expect since ChatGPT is likely tapping into features like Google Shopping that weren’t considered in this study.

B2C SaaS also has a weaker overlap between cited sources and the SERP, but, again, this is to be expected because these types of products are typically pulled from app stores rather than their website directly.

This tells us that the variability comes more from how it weighs search results vs training data rather than from what the user is looking to buy. 

We Went 10 Pages Deep, But ChatGPT Doesn’t for Most of Its Sources

Of the sources in our study that did appear in Google’s results for the fan-out queries, page 1 alone accounted for a third of all matches. The top three pages of the SERP contained ~66% of the matches, with a significant decline after that.

This shows what SEOs (including us) have observed and frankly, wished for, which is that ranking higher improves your chances of getting cited.   

But the next important question is if ChatGPT is only pulling from the live web about 40% of the time on average, where is it getting the rest of the sources?

Where Does ChatGPT Get the Other 60% of Sources It Cites?

At the end of the day, where ChatGPT is pulling its sources from is a black box. But we do have a few theories based on what we observed in the data.

ChatGPT May be Running Additional Queries Behind the Scenes

We can only see the fan-out queries that run in our browser. It’s likely, however,  that ChatGPT is running additional searches on OpenAI’s servers that we never see. 

As QueryBurst has documented, it would make sense for OpenAI to distribute searches across multiple servers simultaneously — using different search engines, running different query variations, etc. If it did all of this sequentially in the browser, responses would be painfully slow.

Response Caching

ChatGPT may also be remembering or caching results from past users’ queries to save resources. OpenAI offers developers discounts if they used cache responses in their builds, so it makes sense that they would be using this approach too. If millions of users are asking similar buying-intent questions, it would be cheaper and faster to cache frequently retrieved sources rather than running fresh web searches every time.

Training Data

There’s evidence that ChatGPT retains specific details and facts verbatim from its training data. Research from NeurIPS has demonstrated that large language models can memorize and reproduce training data verbatim. If ChatGPT can remember exact text snippets, it’s conceivable it could also remember exact URLs.

In fact, we found several cases in our data where ChatGPT cited a real, live URL but gave it an outdated title — one that no longer matches the current page. For example:

In another example, ChatGPT cited the same exact URL twice but with different titles, in the same source list. 

Here are the two titles ChatGPT showed: 

The best video conferencing software in 2026 | Zapier

The best video conferencing software in 2025 | Zapier

And, here’s the title that appeared in Google’s SERP for the fan-out query which is also the actual title of the article as of this writing: 

The best video conferencing software for teams in 2026

One explanation for both cases is that ChatGPT is pulling the URL and its old title from training data, not from a live web search. In the second example, one explanation for why it appeared twice is that  it may have pulled the URL from its training data and from a live web search. 

Caching doesn’t explain this either, since OpenAI’s own documentation shows caches typically last 5–10 minutes, and sometimes up to an hour — not months or years, which is the gap between these outdated titles and the current ones.

Hallucinations

In our dataset, we found roughly 10% of citations led to error pages. Recent academic research puts the number higher. A February 2026 study found hallucination rates above 14% for cited sources, and an earlier 2023 study published in Nature documented similar patterns. These are citations that look real — plausible URLs with plausible titles — but point to pages that don’t exist.

So where does all of this leave us? 

GEO ≠ SEO…

The core takeaway is that you can’t use the same mindset or strategies for GEO or AIO that you do in traditional SEO. In SEO, we’re used to the repeatable and predictable idea of “you write a relevant, helpful, quality article targeting keyword X, get your on page SEO right, and build backlinks, and rank for X”. But, LLM answers are way too unpredictable for this same approach to be useful. 

SparkToro’s recent research reinforces this. They found that the brand names that LLMs recommend are wildly inconsistent. Different users can enter the same exact prompt and not only get a different list of recommendations, but also get those recommendations in different orders. 

Add to that the fact that in AI tools, unlike traditional Google, users ask essentially the same question in wildly different ways and the LLMs factor in a ton of deeply personal context with every users’ prompt, and you see how GEO is less controllable and predictable than SEO.

On top of that, you get different fan-out queries for the same exact prompt. The LLM won’t always search the web even for the same prompts. If it does search the web, the amount it relies on that web search versus training data can differ wildly. Also, the model’s capabilities change based on whether the user is paying or using a free account. I could go on, but you get the point.  

And, all of that unpredictability only accounts for roughly 40% of the cited sources. 

That is why a content strategy built around trying to rank for specific queries in order to be cited is impractical and ultimately will be ineffective.

… But SEO is Necessary for GEO

While the nebulous nature of LLMs is uncomfortable for most marketers and business owners, there are patterns to how LLMs operate and actions you can take to get mentioned more often by LLMs. 

Be Available for Training Data

OpenAI admits to using web crawlers both for live web search and for training data, which means you have a chance of getting into ChatGPT’s training data just by putting content on your website. (More on what this content should be below. Hint: It’s not just “easy to read content with digestible chunks”.)

Increase Your Topical Authority

In the SparkToro piece we mentioned earlier, they mention that certain brands are mentioned repeatedly across responses.The top 3 most mentioned brands were cited 64% of the time for the same prompt on ChatGPT.

And our own data suggests ChatGPT may be recognizing that certain domains have more topical authority than others.

If we count all sources that had some page on their domain rank for a fan-out query — instead of requiring the exact URL to match — the percentage of cited sources appearing in Google’s SERPs jumps from 27% to about 50%.

This suggests that there’s some correlation between what ChatGPT and Google consider to be an authoritative site. And, ChatGPT may be favoring domains it recognizes as topically authoritative, even when it doesn’t cite the exact page that ranks. Your domain’s broader presence across a topic may matter more than whether one specific page ranks for one specific query.

One of the most effective ways to build that kind of topical authority is through SEO. Not the “optimize your meta tags and add FAQ schema” kind of SEO, but the kind where you’re consistently publishing detailed, substantive content that addresses the real pain points your customers have. The more comprehensively your domain covers a topic area, the more likely ChatGPT is to recognize you as a relevant source when someone asks about that topic, regardless of which specific fan-out query it runs.

78% of Cited Sources Were Typical SEO-Style Content, Whether or Not They Ranked

One of the big questions floating around marketing circles right now in various forms is whether or not content type needs to change. Do LLM’s even cite traditional marketing or SEO style blog posts such as list posts? And, important for this study, if you produce a traditional style SEO post like a list post but don’t rank yet for your target keyword (or related keywords), do you still have a shot at being cited by ChatGPT for typically similar prompts? Our data says, yes.

For example, here’s a post by Calendly that was ranking for the fanout query “tools for scheduling meetings with clients without back and forth emails scheduling tool”.

It looks very much like a typical SEO-style post that we would write for that keyword. And they put themselves first on the list, which we always recommend. 

In fact, 78% of all cited sources for these high buying intent keywords were listicles, product pages, homepages, pricing pages, etc — essentially webpages selling your products/services, i.e., typical marketing or SEO content.

While this makes sense for the 40% of sources that were in the fan-out query SERPs, what may surprise some marketers is that it was also true of the 60% of sources not found in Google or Bing’s SERPs. The number only went down slightly to ~74%.

Said another way, even the majority of content cited by ChatGPT that were not ranking for fan-out queries looked and felt like content you would publish to rank for those queries: list posts, landing pages, etc.

This is aligned with the theories discussed above of where else ChatGPT could get these citations from, if not from the fan-out queries: caching past searches, or finding similar content by doing web searches when it was last trained. 

This reinforces what we’ve been saying: the best way to influence LLM responses is to produce bottom-of-funnel, product centric content aimed to rank for similar queries on traditional google search. 

What we can add to that understanding with this data is that you don’t need to target and rank for fan-out queries that you can’t even truly find. Instead, creating typical SEO content is the best way to be available for training data and build topical authority, which, again, are your best bets for showing up in LLM answers. 

So, focus on building domain-level topical authority over time. What we recommend and do for clients is to identify your strengths and find Google keywords that map to those strengths. Then, write helpful, relevant, detailed content for those keywords. 

That’s an actual content strategy.

How this Looks for a Real Client, Toro TMS

Toro TMS is a good example of how publishing content builds AI search visibility. They’re a relatively young brand  that hadn’t done much content marketing before we started working with them. Once we began producing product-centric content targeting relevant Google keywords, their ChatGPT visibility followed.

They published thorough, detailed content about their product and the problems it solves, targeted at real Google keywords where their product is a genuine answer. The AI visibility came as a byproduct of high-quality content marketing.

Traditional SEO is predictable. You pick a keyword, create content around it, build links, and over time, you rank. You know which keywords have volume. You know which pages rank. You can track exactly where you stand. It’s measurable and repeatable.

AI search is different. There’s uncertainty at every step. 

1. Prompts: You don’t know exactly what prompts users are entering, because users don’t type short, repeatable keywords. They have long conversations unique to them. 
   
2. Context: Even if you do know the prompts users are entering, you or your AI visibility tool isn’t going to get the same answer as a real user because LLMs factor in a massive amount of user history, memory, and context to generate a response that’s personalized to that specific person. (Read more: Invisible Prompts).
   
3. Repeatability: Even the same user, asking the same prompt repeatedly, will get different responses and citations each time.

So what do you do with all this uncertainty? How do you get those LLMs to recommend your brand and your product when users ask about solutions to their problems in your space? 

The answer relies on this key fact: LLMs still run on content. 

How Search Changes from SEO to AI Search

Here’s the key difference between traditional SEO and AI search.

In SEO, everyone searches with short keywords and largely sees the same results. Yes, some personalization is done based on your browsing history and location, but it’s mostly around the edges.

For example, say we both search for “project management software.” We’ll likely get similar results: 

Seems normal, right? 

But what if you’re at an enterprise company switching from Jira and you want something very specific, and I’m a small business owner currently using pen and paper looking for my first real project management tool? Those details absolutely affect what project management software makes sense for you versus me, but Google doesn’t have any idea. So it serves us the same ten blue links. 

The way we used to deal with this in SEO is going after a few longer tail variants when they exist, like “enterprise project management software” or “project management software for small business.” But even that is insufficient. That doesn’t capture the nuance I mentioned above: that you’re already on Jira and want something specific that Jira doesn’t offer, and that I’m looking for my very first project management software. 

Those are just two random examples. There are really an infinite number of possible specific requests and pain points for a customer looking for any solution. Traditional Google search simply didn’t have a good way of handling all this personal context. 

But AI search does. 

In AI search, each user has a unique, personal conversation with the LLM where they share a massive amount of context. The LLM understands their company size, existing tools, and highly specific pain points. This context comes from both the current thread and a large body of prior chat history, enabling the model to recommend products tailored to their exact situation.

Here is Benji asking Claude about the best project management software “for us.” Look at how Claude knows what our company is, how big, and more. Benji didn’t type any of that into that prompt, but it knows. We call this phenomenon Invisible Prompts. 

LLMs aren’t just surfacing articles from high-authority domains.

They understand what a user needs, search the web for relevant solutions, read those results, combine that with what they already know, and recommend the product or service that best fits the user’s specific pain points. 

That’s a fundamentally different dynamic. And it changes the kind of content you need to produce.

If you want LLMs to recommend your product, you need to give them the details they need to recommend you in the right scenarios. Not generic “what is project management” content. 

You need to publish specific content that covers:

• What types of customers your product is best for
• What specific use cases and scenarios it handles
• What pain points it solves and for whom
• The features that matter and when they matter
• The benefits and outcomes customers experience
• How you compare to competitors and what makes you different
• Case studies that prove you’ve delivered results

Introducing Your GEO Topic Map

Your GEO topic map is a collection of content that teaches LLMs everything they need to know about your product so they can recommend you in the right conversations.

Here’s how we build it for clients:

1. We map every product angle that matters. Your categories. Your competitors. Your use cases. Your target personas. The jobs your product does. The pain points it solves. This becomes your topic map: the full universe of conversations where an LLM could potentially recommend you. Most companies aren’t thinking this way, much less producing this content. 

2. We interview the people who know your customers best. We sit down with your sales team, your customer support team, your product team, your founders, all of the customer facing roles inside your organization. We ask them exactly what users are telling them are their pain points and use cases. Then we extract the specific positioning, features, and differentiators that make your product the right choice in each scenario. These details are not something you can get from Google research. AI generated copy will not by default produce this. The details required to produce this kind of content have to come from the people inside your company who talk to customers every day.

3. We produce content for each angle at the depth of a sales conversation, not an intro guide. There’s this long standing culture in content marketing where everything is written at an introductory level.  “Intro guide to this.” “Beginners guide to that.” That came from everyone chasing high volume SEO keywords that were by nature beginner-leaning. 

But beginner level content won’t help you when your customer asks advanced and specific questions to an LLM. It wants to recommend brands that talk at that customer’s level. So your content needs to meet the customer there. Think about the level of depth and detail that happens in a sales conversation: detailed discussion of features, screenshots, specific scenarios where your product applies, use cases, case studies. That’s how we write this content.

4. We target Google keywords if and when it makes sense. If some of the user pain points we uncover in our interviews with your team match a known Google search term, we’ll target those search terms with our content so you get two inbound channels (traditional search and AI search) in one go.

For most businesses, there are plenty of these product recommendation keywords (what we call bottom of funnel queries) on Google. Keywords like: “Best [category] software,” “[competitor] alternatives,” “how to [solve specific problem].” 

5. We produce case studies. LLMs are smart enough to connect the dots between a story of your product helping someone and their user who is asking them about a similar issue. So it’s our experience and belief that case studies help. They naturally have the elements of specificity and pain point/solution pair elaboration that we’re talking about here. 

6. We track your AI visibility at the topic level. To make this process work, you need to measure success the right way. That means tracking visibility at the topic level, not the individual prompt level. Unfortunately most AI visibility tools (like Peec, Profound, Scrunch) are default to measuring success on an individual prompt basis. But we built our tool, Traqer.ai to be consistent with this topic-based property of GEO. That’s how we track improvement in visibility for clients and ourselves: which topics are we visible in and how is that growing.

Case Study: Constitution Lending

Constitution Lending is a private lender competing against companies with 80+ domain authority, thousands of backlinks, and decades of industry presence.

We built their topic map by interviewing their lending experts, the people who talk to borrowers every day, and producing content covering every specific lending scenario their ideal customers face: LLC mortgages, DSCR loans for specific property types, bridge loans for particular situations.

The result: Constitution Lending now appears in AI search results for 50+ bottom-of-funnel prompts across ChatGPT, Perplexity, and Google AI Overviews — competing with and often outranking lenders that have been in the market for decades.

This didn’t happen because of an llms.txt file or restructured headings (these on-site hacks don’t help LLMs connect your brand to users’ pain points, read more why here). It happened because LLMs found detailed, specific content about Constitution Lending’s products and could match it to user queries.

[Read the full case study →]

Case Study: InnovationCast

InnovationCast makes innovation management software for enterprise companies. 

Innovation software has multiple use cases, from crowdsourcing ideas from a large employee base, to validating ideas, to scouting new technologies.

When we started working with them, they had a limited content and SEO presence. But by producing content in their various topic areas, we’ve built not only an SEO presence that generates qualified leads for them, but also AI visibility

For example, here is their topic based coverage for “innovation portfolio management software”:

And here it is for another, slightly different way customers talk about software like theirs: “technology scouting software”:

Next Steps

If you’d like to talk to us about building your brand’s topic-based GEO strategy, you can reach out here.

If you’d like to read more about our GEO experience, observations and research here is further reading that underpins this Topic-Based GEO approach: 

• Ultra-Specific Content: Why top-of-funnel content doesn’t matter for AI search and why ultra-specific content is necessary
• Prioritized GEO: Why on-page hacks like llms.txt don’t work and why content and web search is the foundation of GEO
• Invisible Prompts: Why you can’t see the full context behind users prompts 

If you’d like more articles like this as we write them, you can join our newsletter. 

Here’s an example: company leadership asks marketing, “How are we doing on GEO/AEO/AIO?” Marketing turns to a tool like Peec or Profound, which spits out a single brand-visibility percentage, usually shown in a graph. Marketing reports back to leadership, “We have 26% visibility!” 

If that number is higher than last week, everyone’s happy and thinks marketing is “doing their job.” If it’s lower than before, leadership assumes something is wrong and wants marketing to change what they’re doing.

Below, we argue that this approach doesn’t make sense because it’s fundamentally the wrong way to think about GEO performance. 

A brand’s “AI search visibility” isn’t a brand-level metric. It’s a function of the specific topics or prompts being measured.

Reporting visibility at the brand level—especially as a single percentage—is, at best, unhelpful and, at worst, outright misleading. 

As we explain below, this mindset can lead to several negative outcomes, including: 

• Manipulation of metrics to make GEO performance appear stronger than it actually is
• A focus on “ranking” for prompts that have little to no business value
• Marketing teams being disincentivized to push harder or pursue more ambitious GEO strategies
• Resources wasted on activities that have no real impact on lead generation or revenue

Visibility Happens at the Topic Level, Not the Brand Level

If you remove the AI visibility tools and their dashboards from the picture, the idea of a single “AI visibility score” falls apart. In the real world, no brand has one. 

AI search visibility isn’t like domain authority in SEO, which represents a real, meaningful concept: how much weight Google’s algorithm assigns to a domain when deciding what to rank. In SEO, a site with high domain authority will, on average, outrank a site with low domain authority, all else being equal. 

There’s no real-world equivalent to a single AI search visibility percentage. Brands have visibility—or don’t—within specific topics. And if a brand has strong visibility in one topic, that visibility isn’t correlated, based on any data we have seen to date, with its visibility in another. 

By its very nature, AI search visibility is a topic-specific concept. 

Topic Level Visibility Is Actionable

More practically, we believe viewing visibility at the topic-by-topic level is how marketing teams learn which actions actually improve AI search visibility. When teams rely on a single, overall visibility percentage instead, those details get buried and the signal becomes misleading.

Here’s a real example from one of our clients, Mirascope, which has strong visibility on the topic of “prompt engineering tools”:

By contrast, on the topic of “context engineering platform,” they have far less visibility:

What does averaging these two together tell the Mirascope team? Nothing useful.

Leadership should think about visibility in these two topic areas completely separately. The actions required to improve visibility for “context engineering platform”—publishing more content on the topic, clearly explaining how Mirascope supports context engineering, and conducting citation outreach on relevant articles—will have no impact on visibility for “prompt engineering tools.”

The SEO analogy applies here. If Mirascope ranked #1 for the keyword “prompt engineering tools” and #60 for “context engineering platform,” what use would it be to average the two and report on a blended rank? No one does this in SEO because it makes no sense. 

Yet this is exactly what happens in GEO, largely because popular AI search visibility tools default to showing a single, overall brand visibility percentage in their dashboards. 

Single Visibility Percentages Create the Wrong Incentives

A single AI visibility percentage often becomes misleading (at best) or downright manipulated (at worst).

Consider what happens when a marketing team starts tracking additional prompts—aspirational topics they aren’t yet visible for but want to be. Adding those prompts, by definition, drives the overall visibility percentage down. 

On the surface, it looks like AI search visibility declined. Leadership may even comment on it or assume marketing isn’t doing their job well. But nothing bad actually happened. Visibility across ChatGPT and other LLMs is exactly the same as it was the day before. The only change is that the team is tracking more topics and being more ambitious about its GEO goals.

Here’s a concrete example. On January 13, the visibility score in Peec for this brand was 26.7%: 

On January 14, we added two new prompts that were relevant to the brand but had no visibility. As a result, the visibility score dropped to 20%: 

In reality, the brand lost no visibility. The metric only looks worse because new prompts were added.

This creates terrible incentives. It encourages marketing teams to be less ambitious. Teams are incentivized to avoid tracking new prompts because doing so makes their visibility percentage look worse. 

That’s the exact opposite of good marketing. Teams should be encouraged to identify new, high-value topics they want to rank for—not discouraged from tracking them.

How Our Visibility Tool Traqer Solves This

This is exactly why we built Traqer to be topic-based.

Topic-level visibility sits front and center in our tool, and that’s how we report on GEO visibility to our clients.

This page is what marketing teams can share directly with executives or clients. At a glance, everyone sees visibility by topic and can quickly understand where their brand is performing well and where it isn’t.

In the Mirascope example above, it’s immediately clear that they perform well for “prompt engineering tools” but have work to do for “context engineering platform.” No one has to dig through reports or click elsewhere to find this information. 

Yes, Traqer also includes brand-level metrics at the top. But instead of collapsing everything into a single percentage, those metrics show the total number of topics where visibility is low, medium, or high. This distinction matters. Adding new topics doesn’t artificially reduce performance elsewhere. 

This approach does two important things:

• It aligns incentives correctly. You can add 10, 50, or 100 new topics or prompts without hurting your overall score.
• It keeps teams focused on what’s actionable: which topics already have strong visibility and which ones need work.

The Right Way to Think About GEO Visibility

Regardless of whether you use Traqer, another tool, or track visibility manually in a spreadsheet, our recommendation is to think about GEO visibility at the topic level. That’s the proper way to approach GEO strategy.

Overall brand visibility percentages are, in our opinion, a lazy way to assess GEO performance. They hide the details that should (1) guide your strategy and (2) show your team whether GEO efforts are actually working.

When you track topic-level visibility, you can:

• Identify which topics need immediate attention
• Allocate content creation resources to the right areas
• Measure progress on specific topics and initiatives
• Avoid false signals from percentage fluctuations
• Make data-driven decisions about where to focus your GEO efforts

It’s the difference between knowing you have “32% visibility” and knowing you dominate prompt engineering conversations but need work on context engineering topics.

One tells you nothing. The other tells you exactly where to focus next.

Further Reading

• You can read the full Prioritized GEO strategy we execute for clients.
• You can read more GEO strategy articles.
• You can check out Traqer for proper AI visibility tracking.
• You can reach out about working with us to improve your GEO.