Google Keyword Planner and KWFinder tied at the very top of the keyword research category: each was named in all 6 of the 6 answers we collected. We asked ChatGPT and Gemini six versions of the same buyer question, "which keyword research tool is best," and those two names came back in every single answer. No other tool was unanimous.
The next tier drops off fast. Allintitle KW Explorer, AnswerThePublic and Semrush were each named in 4 of the 6 answers (67%). Ubersuggest landed in 3 of 6 (50%). After that it thins out: Google Search Console and Google Trends appeared in 2 of 6 each (33%), and eleven more tools were named just once (17% each). Out of 19 distinct tools mentioned across all six questions, only two showed up every time.
Google's own tools quietly took three of the top eight slots: Keyword Planner (6 of 6), Search Console (2 of 6) and Trends (2 of 6). That is worth sitting with if you sell a paid tool in this space, because a third of the durable shortlist is free Google product.
Here is what we did, so you can repeat it. On 2026-08-04 we ran six buyer-style questions through ChatGPT and Gemini, each with one live web search, and recorded every product they named and every source domain they cited. Six questions, six answers. Every number on this page is a raw count out of those 6 answers. We weighted nothing.
How the shortlist changed across the six questions
The single most useful thing we found is that there is no one shortlist. The list rearranges depending on how the buyer phrases the question, and all 11 of the tools named just once appeared in exactly one of the six questions. None of them survived a rewording.
The bare "best keyword research tools" question returned seven names: Google Keyword Planner, Semrush, KWFinder, AnswerThePublic, Allintitle KW Explorer, Ubersuggest and Mangools. Mangools showed up here and nowhere else, so it lives entirely inside the plain "best" framing (1 of 6 answers).
Switch the buyer to "best free keyword research tools" and the list tilts toward Google's own stack: Google Keyword Planner, KWFinder, AnswerThePublic, Allintitle KW Explorer, Google Search Console, Google Trends and WordStream's Free Keyword Tool. WordStream's Free Keyword Tool appeared only in this free-focused answer (1 of 6). Semrush, named in 4 of 6 answers overall, dropped out completely the moment the word "free" entered the question.
"Most affordable keyword research tools" rearranged the board more than any other question. It was the only one of the six where Google Keyword Planner was not on top: Keysearch led that answer, alongside KWFinder, Ubersuggest, Google Keyword Planner, Google Search Console, Ahrefs Keyword Generator and LowFruits. Three of those, Keysearch, Ahrefs Keyword Generator and LowFruits, appeared in the affordability answer and nowhere else (1 of 6 each). Semrush and AnswerThePublic both vanished from this list too.
The "best for small teams" question pulled in the most one-off names of all. Five tools showed up only there: WordStream, RyRob's Free Keyword Tool, Ahrefs, Keywordtool.io and Keyword Explorer, each named in 1 of 6 answers. The "what should I use" question added its own single exclusive, Keyword Tool (1 of 6). If you sell to small teams, the competitive set the AI draws from is wider and shakier than the headline list suggests.
Two things stayed constant. Google Keyword Planner and KWFinder held at 6 of 6 through every rewording, and the generic "keyword research tools recommendations" question introduced zero new names: every tool it returned had already appeared in another answer. That is what durable AI visibility looks like. Everyone else was conditional on how the question was asked.
Where the answers came from
The answers did not come from review directories. Across all six questions the assistants cited 18 distinct domains, and not one of them was G2, Capterra, TrustRadius or Software Advice. If you have been grooming your directory profiles to influence AI answers, this category suggests that is not where the models are reading.
The single most-cited source was Zapier's blog: zapier.com fed 6 of the 6 answers (100%). Community threads carried real weight too: reddit.com was cited in 5 of 6 answers, quora.com in 3 of 6, and community.hubspot.com in 2 of 6. That matches what we keep seeing across categories, a busy forum thread beats a polished landing page for getting read by the model. Independent roundups were the other big feed: zapier.com (6 of 6) and marketermilk.com (3 of 6), plus a long tail of single-cite roundups like adai.news, yourcreativecontent.com, blog.hubspot.com and moz.com (1 of 6 each).
The key insight is what happened on the vendor domains. semrush.com was cited in 5 of 6 answers, the most-used source after Zapier, and Semrush the product was named in 4 of 6. Its own published roundups helped put its own name in the answer. The same shape repeats smaller: wordstream.com fed 4 of 6 answers and a WordStream tool was named; keywordtool.io fed 3 of 6 and Keyword Tool was named; app.neilpatel.com fed 2 of 6 and Ubersuggest, Neil Patel's tool, was named in 3 of 6; mangools.com fed 2 of 6 and Mangools' own KWFinder was named in all 6; lowfruits.io fed 2 of 6 and LowFruits was named; keysearch.co fed the "most affordable" answer and Keysearch topped that very list. When a vendor publishes the comparison, the vendor tends to be in it.
It is not automatic, though. seranking.com was cited in 1 of 6 answers, but SE Ranking was never named as a tool. Being the source is not the same as being the pick. Still, across this category the vendors whose own domains got cited were far more likely to be named than the ones who stayed silent.
What this means if you sell a keyword research tool
If you want an AI assistant to name your tool here, the source list is the target, not your market share. The two unanimous tools were the ones sitting inside the pages the models actually read: Zapier's roundup (6 of 6), the Reddit and Quora threads (5 of 6 and 3 of 6), and a handful of vendor comparisons. Nobody was named for being "the market leader." They were named for being on those pages.
Concretely, from this measurement, that means three moves. Get into the Zapier-style roundups that were cited in all 6 answers. Get talked about in the Reddit and Quora threads that fed 5 of 6 and 3 of 6 answers. And publish your own honest comparison, because the vendor domains that did (semrush.com at 5 of 6, wordstream.com at 4 of 6) got their products named. Chasing a G2 badge did nothing in this category, because G2 was never cited once.
And watch the phrasing. Semrush was named in 4 of 6 answers but in zero of the "free" and "affordable" ones. If your buyers ask AI for the cheap option, a strong general presence will not save you. You have to be in the specific sources that specific question pulls from.
The full list, counted
| Product | Named in | Share |
|---|---|---|
| Google Keyword Planner | 6 of 6 | 100% |
| KWFinder | 6 of 6 | 100% |
| Allintitle KW Explorer | 4 of 6 | 67% |
| AnswerThePublic | 4 of 6 | 67% |
| Semrush | 4 of 6 | 67% |
| Ubersuggest | 3 of 6 | 50% |
| Google Search Console | 2 of 6 | 33% |
| Google Trends | 2 of 6 | 33% |
| Ahrefs | 1 of 6 | 17% |
Eleven tools were each named in 1 of 6 answers, and we left the labels exactly as the assistants said them rather than merging them. So Ahrefs (1 of 6) and Ahrefs Keyword Generator (1 of 6) sit as two separate rows for the same brand, WordStream (1 of 6) and WordStream's Free Keyword Tool (1 of 6) are a second pair, and Keyword Tool (1 of 6) and Keywordtool.io (1 of 6) are a third. We did not add those pairs into a single higher number, because the data does not show one. The remaining one-time names were Keysearch, Keyword Explorer, LowFruits, Mangools and RyRob's Free Keyword Tool, each at 1 of 6.
Why we ran this
To be straight about it: Glotier is not a keyword research tool, and you will not find us anywhere in these 6 answers. That is correct. We do not sell in this category, so ChatGPT and Gemini should not be naming us here, and they don't.
But this measurement is exactly the thing Glotier does, pointed at someone else's category. We asked the buyer questions, recorded who ChatGPT and Gemini named, counted it out of 6, and mapped the sources behind each answer. If you sell a keyword research tool, or anything else, the same run tells you whether you show up when your buyers ask, and which pages you would need to be on to change that. The check is free, no account, and takes about a minute.