The single domain every AI assistant leaned on, and the tools they named
We put six questions to ChatGPT, Gemini, and Perplexity, and every single answer came back drawing from the same site: onelittleweb.com. All six prompts produced a readable response, and across 18 total assistant responses, that domain fed the recommendations without exception. That uniformity is far more striking than any individual tool ranking, because it means the assistants’ view of the AI SEO tools category is, right now, filtered almost entirely through a single publisher’s content.
When we asked things like “best ai seo tools,” “best ai seo tools for small teams,” and “ai seo tools recommendations,” Clearscope was the only product to appear in four of the six answers, making it the most consistently named tool. Ahrefs, Rankability, Semrush, and Surfer SEO each appeared in exactly half the answers (three of six). No other tool crossed that threshold. Frase, Mangools, SE Ranking, Semrush One, Surfer, Ubersuggest, and Writesonic managed two mentions; another 15 products, including Ahrefs AI Content Helper, AirOps, and seoClarity, got called out only once.
What surprised us was the assistants’ willingness to recommend themselves, though it happened rarely. ChatGPT and Perplexity each appeared in a single answer, both times in the response to “best free ai seo tools.” That question produced an entirely different set of names from the other five: eesel AI blog writer, AIOSEO Analyzer, Google Search Console, Frase GEO, and the two assistants. The list for “most affordable ai seo tools” was distinct again, leaning on Frase, Surfer SEO, Clearscope, Rankability, Ahrefs, and Keysearch. The assistants clearly use query intent to switch between tool sets, but the underlying source pool remains narrow.
Beyond onelittleweb.com, a small cluster of domains appeared in four of the six answers: freddiechatt.com, rankability.com, and whatagraph.com. eesel.ai and seo.ai fed three; behindrankings.com, gomega.ai, quattr.com, reddit.com, semrush.com, and youtube.com fed two. The rest, including ahrefs.com, zapier.com, and getaira.io, showed up once. This concentration means a handful of publishers effectively dictate which tools the assistants surface when someone asks for AI SEO recommendations. For any tool not named on those sites, the chance of being recommended by ChatGPT, Gemini, or Perplexity today is near zero.
Clearscope was the closest thing to a default pick, but no tool appeared everywhere
Clearscope was the most frequently named product across the six questions, appearing in 4 of the 6 answers. That makes it the only tool the assistants recommended more often than not, but it is still two answers short of universal. No product appeared in all six; the assistants never converged on a single must-have, even for a tightly defined category like AI SEO tools. The gap between 4 mentions and 6 means a searcher asking “what ai seo tools should I use” might get Clearscope one day and miss it entirely the next, depending on how the question is phrased.
Five products sit at 3 mentions apiece: Ahrefs, Rankability, Semrush, and Surfer SEO. (Surfer SEO is listed separately from Surfer, which the assistants sometimes used as shorthand; more on that in a moment.) This group mixes content optimization platforms with broader SEO suites, and it feels like the second ring of the assistants’ attention. Ahrefs and Semrush are the names a shopper would expect to see, while Rankability and Surfer SEO represent the AI content side more squarely. The even split suggests the assistants are not distinguishing between tool categories as a human buyer might; they treat a backlink checker and a content optimizer as interchangeable answers to “best ai seo tools.”
The next cluster, named in 2 of the 6 answers, includes Frase, Mangools, SE Ranking, Semrush One, Surfer (the short form), Ubersuggest, and Writesonic. The Surfer/Surfer SEO split is a measurement artifact worth flagging: if you combine the two name variants, Surfer would have been named in 5 of the 6 answers, matching Clearscope’s reach. But we measured what the assistants actually wrote, and the inconsistency is real. A product team reading this should care, because a brand that lets its name fracture across “Surfer” and “Surfer SEO” in the same category of prompts is leaving recommendations on the table. Semrush shows a similar naming diffusion: Semrush (3 mentions), SEMrush (1 mention), and Semrush One (2 mentions) all appear. The assistants treat these as separate entities, and a buyer scanning a list might too.
Sixteen products were named exactly once. The list is long: Ahrefs AI Content Helper, AIOSEO Analyzer, Aira, AirOps, ChatGPT, eesel AI blog writer, Frase GEO, Google Search Console, Keysearch, Perplexity, Rankscale.ai, SE Ranking AI Overview, SE Ranking AI Overview Tracker, SEO.AI, seo.ai, seoClarity, and Whatagraph. A few patterns jump out. Two AI assistants, ChatGPT and Perplexity, recommended themselves, which is a recursion worth noting but not actionable for a tool vendor. The SE Ranking ecosystem shows up in multiple guises: the core tool got 2 mentions, while the AI Overview and AI Overview Tracker each got 1. That kind of fragmentation tells a vendor something about how their subproducts surface (or don’t) in conversational answers. Frase GEO and Ahrefs AI Content Helper are narrow feature names, not the parent suites, and they each appeared exactly once. The assistants are pulling very specific, sometimes obscure, product names from their source material and treating them as standalone recommendations.
What is missing from the list is as instructive as what is on it. Moz is nowhere. Screaming Frog is nowhere. Ahrefs’ core tool is present at 3 mentions, but the broader Ahrefs suite is reduced to an AI Content Helper named once. Google Search Console got a single mention, despite being free and universally used. These omissions are not evidence of irrelevance; they are evidence of how the assistants built their answers. Every response drew from a small set of domains, with onelittleweb.com feeding all six. The tools that benefited most are the ones that publisher chose to feature. A tool that was not mentioned in those source articles had almost no path into the assistant responses, regardless of its market share or user base.
The shape of the named-product list also reveals how the assistants interpret the category. The top names are overwhelmingly content optimization and AI writing tools, not rank trackers, technical SEO crawlers, or analytics platforms. Clearscope, Surfer SEO, Frase, Rankability: these are tools for creating and refining content against search intent. When the assistants hear “AI SEO tools,” they reach for content assistance first, and full-stack SEO platforms second. A vendor on the content side should see this as validation that the assistants map the category to their product. A vendor of technical SEO AI should see it as a visibility problem that six diverse prompts barely surfaced their subcategory at all, unless it was in a single-mention niche name.
The takeaway for a product named once is straightforward: a single mention across six questions means the tool appeared in one prompt’s answer and vanished from the other five. That is noise. A buyer cannot rely on it showing up again. For the tools at 3 or 4 mentions, the signal is stronger, but still not a lock. The fact that no tool hit 6 of 6, and that the source influence was so concentrated, means the assistants’ recommendations are a narrow reflection of one editorial viewpoint, captured by the measurement.
The full list, counted
Every product serper+or:chatgpt,gemini,perplexity named across 6 answers, with how many of those answers named it. The share column is out of the answers we read, not out of the category.
| Product | Named in | Share | Read as |
|---|---|---|---|
| Clearscope | 4 of 6 | 67% | 4 answers |
| Ahrefs | 3 of 6 | 50% | 3 answers |
| Rankability | 3 of 6 | 50% | 3 answers |
| Semrush | 3 of 6 | 50% | 3 answers |
| Surfer SEO | 3 of 6 | 50% | 3 answers |
| Frase | 2 of 6 | 33% | 2 answers |
| Mangools | 2 of 6 | 33% | 2 answers |
| SE Ranking | 2 of 6 | 33% | 2 answers |
| Semrush One | 2 of 6 | 33% | 2 answers |
| Surfer | 2 of 6 | 33% | 2 answers |
The source map: where the assistants found their recommendations
Every recommendation these assistants made traces back to a URL. Six domains fed four or more of the six answers, and onelittleweb.com fed all six. That is not a pattern; it is a dependency. Across 18 assistant responses, not a single answer failed to cite it. The next tier, freddiechatt.com, rankability.com, and whatagraph.com, each appeared in four answers, enough to shape a consensus but not enough to counterbalance the first source. The gap between six and four is the difference between a filter everything must pass through and a supporting voice.
Fewer than a quarter of the domains we saw are forums or community platforms. Reddit appeared twice, Quora once. YouTube,sometimes treated as a research layer,showed up in two answers, always alongside written sources rather than as a primary pull. The recommendations did not emerge from scattered user opinions or video deep dives. They came from a small set of editorial pages that round up and rank tools. onelittleweb.com, freddiechatt.com, rankability.com, and whatagraph.com all fit that mould: they publish listicles, comparisons, and “best of” guides. When ChatGPT, Gemini, or Perplexity needed an answer for “best ai seo tools for small teams” or “most affordable ai seo tools,” those were the pages it reached for, and it rarely looked past them.
The presence of vendor domains inside that source list is harder to ignore. eesel.ai and seo.ai each fed three of the six answers, meaning the assistants cited the website of a tool as a source for comparing tools. In the “best free ai seo tools” response, eesel.ai appeared as both a recommended product and a source for the answer. The line between what counts as third-party validation and what is vendor collateral blurred completely. semrush.com fed two answers, ahrefs.com fed one. In a measurement where independent review sites (seoreviewtools.com appeared only once) barely registered, vendor domains acting as source material is less a flaw and more a structural feature of how the category is assembled.
What is missing matters as much as what is present. No large affiliate-heavy comparison site showed up; the assistant did not surface pages from the kind of SEO tool round-up that dominates organic search results for these queries. Instead, a handful of smaller publisher domains, some of them unfamiliar to many SEO buyers, defined the competitive set. behindrankings.com, gomega.ai, and quattr.com each fed two answers. marketermilk.com, wordlift.io, and zapier.com showed up once. The list has no obvious centre of gravity beyond onelittleweb.com. It is not that the same few pages were cited across every question; the specific sources shifted per query, but the pool they were drawn from remained tiny.
That narrow source pool means a single publisher’s editorial judgment about which tools to include, and which to omit, propagates through every assistant. If onelittleweb.com decides a tool is not worth a mention, it does not appear in any answer, because no other source gets a chance to compensate. The category’s shape is not an average of independent assessments; it is an echo of a few pages that happened to train well, rank well, or get indexed in the exact way these retrieval systems prefer. A product manager selling an AI SEO tool would need to be on that short list of domains to have any hope of being surfaced, and the list changes slowly, one editorial update at a time.
Clearscope’s four appearances all share the same four source domains
Clearscope appeared in 4 of the 6 answers, and in every one of those answers the assistant’s source list included onelittleweb.com, rankability.com, whatagraph.com, and freddiechatt.com. Remove any of those four domains from the mix, and Clearscope vanishes: the two answers where it is absent, “best ai seo tools for small teams” and “best free ai seo tools,” pull from source clusters that lack at least one of them. The “small teams” question drew on eight sources but skipped rankability.com entirely; the “free” question added reddit.com, quora.com, and niche blogs, and used none of the four. This is not a subtle correlation. Clearscope’s visibility depends on the presence of a specific quartet of publisher domains, not on any universal judgment about the tool.
Self-referencing sources put their own products into the answers
Two products in the list are almost certainly there because the domains that named them are also the products they sell. Rankability was named in 3 answers, and rankability.com fed 4 of the 6 source lists. The three answers that named Rankability (“what ai seo tools should I use,” “most affordable ai seo tools,” “ai seo tools recommendations”) all include rankability.com as a source. Whatagraph appears in exactly one answer, the “should I use” question, and whatagraph.com is a source for that answer. There is no way to distinguish a product recommendation from a self-citation here; the assistant is repeating the name of a site that site itself is promoting. For a product like Rankability, whose domain appears almost as frequently as the big aggregators, that alone could explain a third of its mentions.
The outlier question surfaces community sources and a different product set
The “best free ai seo tools” answer looks nothing like the others. Its source list pulls from reddit.com, quora.com, eesel.ai, wordlift.io, and marketermilk.com, alongside onelittleweb.com. The tools named there (eesel AI blog writer, AIOSEO Analyzer, Frase GEO, Perplexity, ChatGPT) appear nowhere else in the measurement. That is a direct consequence of the source shift. The assistant did not decide that ChatGPT is a free AI SEO tool; it found a source, likely a forum thread or a blog post, that mentioned it. Once the source fell outside the four-domain cluster, the recommendation list reset. This shows that the assistants do not carry a stable, internal catalog of the category. They read whatever the search step returns, and the product names follow the domains.
What the source list cannot tell us
We have no data on whether a domain mentioned a product favorably or in passing. The measurement only records that a named product appears in an answer and that a domain is among the sources. We cannot say if semrush.com actively recommended Semrush One when it appeared in two answers, or if an aggregator like onelittleweb.com cited it neutrally alongside competitors. We also cannot separate genuine consensus from circular referencing: rankability.com and whatagraph.com show up in each other’s source lists, so the same small cluster of sites is cross-validating itself. Further, we do not know how the assistant’s language model selected which product to name from a source page. A long article listing twenty tools might be reduced to just the first three the assistant extracts, and we see no record of that triage. Finally, slight name variations (Surfer vs Surfer SEO, SEMrush vs Semrush) were split in the count, which understates the combined reach of that tool and obscures whether the same source used one spelling or both. Without that extraction-level detail, the product names are downstream effects of editorial decisions made by sites we cannot audit.
Your spot in the answer belongs to the sources, not to you
The pattern that opens the last section, where Clearscope appeared in every answer whose source list contained onelittleweb.com, rankability.com, whatagraph.com, and freddiechatt.com, is not a subtle correlation. It is a wiring diagram. Clearscope holds its 4-of-6 lead only because those four domains published content that named it, and those four were simultaneously pulled by the retriever. Remove one, and Clearscope drops: the “best ai seo tools for small teams” answer skipped rankability.com and named Semrush, Surfer SEO, Mangools, and Ahrefs instead. The “best free ai seo tools” answer drew from a completely different set that excluded all four, and Clearscope is absent there too. If you sell Clearscope, you are not defending a brand preference the assistants have developed. You are defending the continued alignment of four third-party editorial choices, none of which you control.
That fragility tightens as the mention count falls. Ahrefs, Rankability, Semrush, and Surfer SEO each appeared in 3 of the 6 answers. A three-appearance position feels solid, but inspect the source clusters behind those answers: the “most affordable ai seo tools” response, which named Ahrefs and SEMrush, ran on onelittleweb.com, whatagraph.com, yourcreativecontent.com, rankability.com, freddiechatt.com, reddit.com, gomega.ai, and eesel.ai. Swap one of the listicle domains for a source that rates a different set, and a vendor could lose that entire answer pocket overnight. The assistant is not doing competitive analysis. It is echoing whichever pages the retriever served. So a product at 3 mentions is defending a position that is thinner than a single editorial revision on whatagraph.com or rankability.com.
Tools sitting at two mentions, like Frase, Mangools, Ubersuggest, and Writesonic, are living on an even narrower ledge. Frase appeared when the questions pulled in lists that happened to include it, not because the assistants consistently valued it. One rewrite of a roundup on onelittleweb.com or freddiechatt.com, and two answers vanish. For the free-tool sellers mentioned only in the “best free ai seo tools” response (eesel AI blog writer, AIOSEO Analyzer, Google Search Console, and even ChatGPT and Perplexity as tools in that context), the exposure is singular: that answer leaned on reddit.com, eesel.ai, wordlift.io, marketermilk.com, and quora.com. Those are not the same set of domains that power the other five questions. If a future retriever favors a different mix for a free-tools query, or if one of those community threads drifts, that single mention disappears and the product’s assistant visibility goes to zero.
The cost of losing a recommendation is rarely a slow decline. It is a domain update. The moment onelittleweb.com republishes its list and drops a tool, that tool’s count across all six questions could fall by two or three immediately. Because that domain fed every answer, it acts as a master switch for a large slice of the category. A vendor who builds a go-to-market strategy around “ChatGPT always mentions us” is betting that a publisher they do not employ will keep their name on a page. The assistants have no memory of what they recommended last week; they only have the next retrieval. If the core publisher set shrinks or changes ranking criteria, the assistant’s answer shifts with it, and no amount of product truth matters.
If you sell in this category and were not named, the retrieval chain already decided
Every tool that appears in these answers has one thing in common: it is mentioned on the pages the assistants fetched. The products that were named, from Clearscope and Ahrefs down to a single-appearance tool like Rankscale.ai, all show up in the source set we captured. If your tool is not in that list at all, it is because the content the retriever pulled, across six deliberately varied prompts, simply does not contain your product’s name. There is no hidden evaluation, no assistant-level judgment overwriting the sources. The absence is a sourcing gap, and the fix is getting into those pages.
That does not mean optimizing your own site. There is no indication that an assistant crawled anything on your domain, your pricing page or your blog, to reach these answers. The retrieval step in all 18 assistant responses pointed to third-party articles, listicles and a few community threads, and the domains that showed up most frequently were overwhelmingly publisher sites. If you are not on onelittleweb.com, which fed every single answer, or on freddiechatt.com, rankability.com, and whatagraph.com, which each fed four of the six, you are absent from the track that produces the highest-frequency recommendations. Clearscope appeared in exactly those four answers where all four of those domains were cited, and nowhere else. The wiring is that direct.
The practical step is publisher outreach, not AI optimisation. Being recommended by these assistants means being included in the specific roundups and comparison pieces hosted on those domains. That means identifying the exact articles those assistants are retrieving (the ones the serper call returned for each prompt), checking which tools they already list, and earning a place. It is the work of providing a reviewer with access, getting a product into a group test, or being the subject of a standalone write-up that those pages then link to. It is not a bidding war for ad slots and not a metadata trick on your own site.
Notice that even large, category-defining brands did not walk into every answer. Ahrefs appeared in three of the six, the same count as Rankability and Surfer SEO, and one behind Clearscope. That tells you reach outside the core source cluster is narrow: Ahrefs is missing from the three answers whose sources pulled from a different combination of domains. If a name as universal as Ahrefs cannot brute-force its way into all six answers, no amount of general authority or search volume does it. The variable that flips is the source list.
The free tools question makes this even sharper. Prompt “best free ai seo tools” and the sources shift: reddit.com, quora.com, eesel.ai, wordlift.io and marketermilk.com appear, while freddiechatt.com, rankability.com and whatagraph.com disappear entirely from that answer’s citations. A tool like ChatGPT or Perplexity gets named there because these community platforms discuss them as free options, not because the assistants independently assessed pricing. If your product is not being discussed in those threads, you are invisible in the only prompt variation that surfaces free tools at all.
This is not a warning about quality. It is a description of a pipeline. The assistants recommend what the retriever finds, and the retriever finds a tight set of domains that, in this measurement, never exceeded six sources per answer. The product that appears most often, Clearscope, shares its source cluster with the domains that fed four or more answers, and every product with three or more mentions can be found on at least two of onelittleweb.com, freddiechatt.com, rankability.com and whatagraph.com. If you are not on those pages, you are not part of the pattern that drives visibility.
How to run this for your own category, by hand, this afternoon
The only way to see exactly what the assistants are pulling is to ask them yourself, six times, and write down every product name and every source URL. Open a blank spreadsheet, give it columns for prompt, assistant, named tools, and source domains. Then type six questions that matter in your space: the obvious one (“best [category]”), plus variations like “for small teams,” “what [category] should I use,” “best free [category],” “most affordable [category],” and “[category] recommendations.” Go to the web interface of ChatGPT, Gemini, and Perplexity, paste each prompt, and copy the full answer text. For citations, hover on each linked reference and record the actual domain, not just the display text. That is the raw data, and it will take longer than you think.
The tedium sets in as soon as you start extracting. The same product may appear as “Surfer SEO” in one answer, “Surfer” in another, and “SEMrush” (capital M) in a third. You have to read every line and normalise the names manually, because a simple find-and-replace will miss variations or split a tool with a multi-word name. Gemini sometimes structures its answer with bold headings and no clickable sources; you need to scroll down to a separate “Sources” section and match them by number. Perplexity embeds citations inline while ChatGPT often lists them after the answer. You are doing this eighteen times, and the rhythm gets broken every time a source link leads to a redirect or a page that has since been taken down.
Tallying the counts is where the real labour sits. You go through each cell in your spreadsheet and mark a “1” for every product-domain pair, then sum across rows. It is monotonous, but it is also the moment you will spot the pattern: a domain that shows up in every answer, a product that appears only when a specific source set is present. By hand, you notice what a script would obscure: that “best free” pulls from reddit.com and quora.com while skipping the commercial review blogs entirely, or that one publisher’s content is the single thread holding a recommendation list together. That insight is the payoff, but you have to do the two hours of copying and counting to get it.
If a product you sell is not named, you will only learn that after you have already tallied every mention and stared at an empty row. The assistants did not make a judgement call; the sources they retrieved simply never contained your name. Doing this by hand forces you to trace the absence back to specific domains you need to appear in. The process is tedious, repeatable, and the cheapest competitive map you can build for the cost of a lunch break and a spreadsheet.