GEO Fundamentals
Why your competitors show up
in AI, and you don't.
You ask ChatGPT who the leading vendors are, and a competitor's name comes up instead of yours. It stings, and it's rarely a verdict on your product. AI names whoever it can read, quote, and corroborate on the buyer's question, and that's a gap you can close.
GEO Fundamentals · Last updated
For the marketer watching a competitor get recommended in answers they should be in.
If AI keeps naming a competitor and not you, it isn't because they're the better company. AI names whoever it can read, quote, and corroborate on the exact question your buyer asked, and a competitor ahead on any of those wins the slot. Resonate Labs finds which queries you're losing and closes the gap.
Not better, more citable
AI doesn't rank company quality. It names what it can read, quote, and corroborate on the question.
They can trail you on Google
Only about 12% of AI citations overlap Google's top ten. A competitor below you in search can still win the answer.
The gap is closable
AI citations churn 40 to 60% a month. Mentions are earned per question, so slots change hands.
AI doesn’t rank vendors by quality or reputation.
It names whoever it can read, quote, and corroborate on the exact question your buyer asked. Which means a competitor’s lead is work you can take back, not a verdict you have to live with.
Why AI names a competitor over you
Start by setting aside the obvious explanation. When an assistant recommends a competitor instead of you, the instinct is to assume they're the better-known or better-regarded company, and to read the omission as a verdict. Usually that's not what's happening. AI doesn't rank vendors by quality or reputation. It assembles an answer from the sources it can reach and read, and names the companies it can quote and corroborate on the specific question in front of it.
That reframe matters because it changes what you do next. If the assistant were judging your product, the fix would be to build a better product. Since it's judging what it can find and lift about you, the fix is to change what it finds, which is work you control. The competitor isn't winning a popularity contest. They're clearing a few practical bars on that query that you haven't, and those bars are the subject of the next section.
The three reasons they get the mention
A competitor gets named over you when it clears three bars you may not, on the question a buyer asked. They stack in order, and being behind on any one is usually enough to lose the slot.
1. An AI crawler can read their site. Many AI crawlers don't run JavaScript, so a site that assembles itself in the browser serves them a near-empty page. If a competitor's pages render as plain HTML and yours don't, the model can read their answer and not yours, no matter how good yours is. This is the quiet, common cause, and it's covered in why a site that ranks on Google can be invisible to AI.
2. Their content answers the exact question, extractably. Models lift specific, self-contained passages. If a competitor has a page that answers the buyer's question directly, with the answer stated cleanly up top, it gives the model something to quote, while a broader or more meandering page of yours gives it nothing to pull. How to structure content AI will cite is that discipline.
3. Third parties corroborate them. Assistants lean on what the wider web says about a vendor, not just the vendor's own site. Mentions, reviews, and coverage across independent sources predict AI visibility more strongly than backlinks do, so a competitor with a stronger third-party footprint gets trusted and named more often. Earned media for GEO covers those brand signals, and what drives AI citation is the full model behind all three bars.
One consequence worth naming: because AI answers and Google draw on different sources, this can all be true while the competitor ranks below you in search. Ahrefs found only about 12% of the sources AI assistants cite overlap with Google's top ten, so roughly 88% of citation opportunities are decided on a different field than the one you may have been winning.
Who you’re actually being compared to
There is a step before any of that, and most companies skip it: knowing who is in the comparison at all. The three-to-five list is what the battlecard is built from, what the sales team is trained against, and what an AI visibility tool will track for you if you supply it.
Across 18 Foundation Reviews, we named 209 vendors an engine could surface instead of our client — a median of eleven per company, ranging from nine to eighteen. That is already two to three times what most companies would name unprompted. The more useful number is the second one: only about half of those compete for the same decision. The rest overlap, sit adjacent, or solve the problem a different way.
And in 14 of the 18, the set included at least one option that is not a vendor in the category at all. Two kinds kept recurring.
The buyer builds it instead. For a data-redaction company, the set included Microsoft Presidio, the open-source build-it-yourself baseline that nearly every technical evaluator POCs before buying anything, and spaCy, the engine behind most homegrown redaction scripts. For an identity platform, Amazon Cognito, the build-it-yourself default for AWS-native engineering teams and the cheap baseline every business case is measured against, and Keycloak, which anchors the do-nothing option. For a practice-management tool, Notion — free or near-free, infinitely flexible, and already installed.
Someone the buyer already pays extends into your category. A mobile security vendor was weighed against Veracode, which already owns the application security budget line at most large enterprises. A warehouse software firm against SAP, the default incumbent-ERP answer, because buyers already on S/4HANA face internal pressure to take the module rather than buy a specialist. A streaming analytics company against Datadog, an incumbent observability platform already deployed in most enterprise stacks.
None of those appear on a battlecard. No vendor writes competitive copy against a free library, a tool the buyer already has open, or a budget line someone else owns. But that is the comparison the buyer is actually making, and an assistant assembling an answer has no reason to respect the boundary of your category. It is answering a question about a problem, and some of the ways to solve that problem involve buying nothing.
The practical consequence is narrow and worth stating plainly. If a tool reports your share of voice against a competitor list you supplied, it is measuring the race you believe you are in. Whether that is the race the model is running is a separate question, and it has to be answered before the number means anything. Mapping the set is the first section of a Foundation, and it is free, because nothing downstream of it can be trusted until it is right.
One limit worth being straight about. This is the set an engine could reasonably draw from, built from public evidence before any crawl runs. It is not a measurement of who the engines actually name on a given question — that is what the crawl answers. The two are different claims, and we would rather separate them than blur them.
Why the gap is closable
Here's the part that should change how you feel about it: a competitor's lead in AI answers is far less fixed than a lead in search rankings. AI citations move. Profound found that 40 to 60% of the domains cited in AI answers are different from one month to the next, so being named is earned per question and re-earned continuously, rather than locked behind years of accrued authority. A competitor who owns a query today can lose it to a better-structured, better-corroborated answer.
The honest caveat keeps this grounded: not every slot is winnable. Ahrefs found that only about a third of ChatGPT's top cited pages are the kind of off-site content anyone can influence, with the rest being Wikipedia entries, homepages, and reference sites. You won't take every mention from a competitor. But that influenceable third is exactly where competitive visibility is decided, and where the month-to-month movement happens. The work is to win the queries that are winnable, systematically, before your competitor does.
Where Resonate Labs fits
The first move is to find out exactly which queries your competitors are winning. You can do it by hand, ask the assistants what your buyers ask and note who gets named, which is the method in is your brand visible in AI search. For the full and stable version, Resonate Labs runs a fixed set of buyer-intent queries across ChatGPT, Claude, Gemini, and Perplexity and scores where you and your competitive set are each named, cited, or absent, so the guesswork becomes a ranked list of the exact conversations you're losing.
From there the work follows the three bars: make sure the engines can read you, build the extractable pages that answer the questions your competitors currently own, and earn the third-party presence that gets you corroborated. A free Foundation maps your buyers, your competitors, and what AI can read on your site, and the full AI Visibility Crawl turns it into the prioritized plan to take the winnable slots back. If you'd rather talk it through first, book a call.
Frequently asked questions
Why does ChatGPT recommend my competitor and not me?
Usually not because the competitor is the better company. AI assistants don't rank vendors by quality or reputation; they name the ones they can read, quote, and corroborate on the exact question a buyer asked. A competitor gets the mention when it clears three things you may not: an AI crawler can actually read its site, its content answers the buyer's question in a clean, liftable passage, and it has enough third-party presence, mentions and coverage across the web, for the model to trust it. Being behind on any one of those, even while you have the stronger product, is usually why the assistant names them.
My competitor ranks below me on Google but shows up in AI. How is that possible?
Because AI answers and Google rankings draw on different sources. Ahrefs found only about 12% of the sources AI assistants cite overlap with Google's top ten results, so roughly 88% of citation opportunities are platform-specific. A competitor can sit below you in search and still be the vendor an assistant names, if its content is more extractable or its third-party presence is stronger on that question. Ranking well helps, but it doesn't automatically transfer into being cited, which is earned separately and per question.
Can I actually displace a competitor in AI answers?
Yes, more than in traditional search, because AI citations move. Profound found that 40 to 60% of the domains cited in AI answers change from one month to the next, so being named is earned per question rather than locked in. The honest caveat is that some slots are structural: Ahrefs found only about a third of ChatGPT's top cited pages are the kind of off-site content you can influence, with the rest being Wikipedia, homepages, and reference sites. You won't win every mention, but that influenceable share is exactly where competitive visibility is decided, and where the movement happens.
How do I find which queries my competitors are winning?
Ask the assistants what your buyers ask, and write down who gets named. In ChatGPT, Claude, Gemini, and Perplexity, run the phrasings a buyer would use, like "best [category] for [use case]" or "top [category] vendors," and note which competitors appear when you don't. That list of competitor-won queries is your target list. A single manual pass is a smoke test because answers vary, so for the full, stable picture, start with a free Foundation from Resonate Labs, the map the AI Visibility Crawl scores you and your competitive set against.
No card required
See which queries your competitors win.
Start with a free Foundation, the map of your buyers, your competitors, and what AI can read on your site. The full AI Visibility Crawl scores your competitive set across every engine and hands you the plan:
- The buyer questions where a competitor is named and you aren't
- Where you're visible, cited, or absent across the four engines
- The winnable slots to take back first