Are Your Competitors Showing Up in Your Customers’ ChatGPT Conversations?

Somewhere today, a prospect who'd be a perfect fit for your business opened ChatGPT and typed "who are the best companies for this?" — and got a shortlist that named three of your competitors and not you. You'll never see that conversation. There's no impression in your analytics, no line in a rank tracker, no notification. Yet a buying decision quietly narrowed, and you weren't in the room. This article is about that invisible layer: whether your competitors are showing up in your customers' ChatGPT conversations, why it happens, and what you can do to earn a place in those answers.

I've spent the last several years helping clients get found. For most of that time, that meant ranking on a page of blue links you could inspect and argue about. That page is now only part of the story. A growing share of research happens inside AI assistants — ChatGPT, Gemini, Perplexity, and Google's AI Overviews — where the tool doesn't hand back ten options to weigh. It composes an answer, and often that answer already names a recommendation.

The uncomfortable part is that this shortlisting happens in private. You can't watch it or bid on it. But you're not powerless: the brands that surface aren't chosen at random, and the signals behind them are ones you can influence. Let me walk through why an AI names one brand over another, and the work that improves your odds of being the recommendation.

The Conversation You're Not Invited To

Picture the old buying journey: someone searched, scanned a results page, clicked, compared, and decided. Every step left a measurable trace, so you could see the shape of demand even if you didn't win it.

The AI-assisted journey collapses those steps into one exchange. A buyer asks an assistant to recommend options or compare providers, and the model answers in a paragraph or a list. If a competitor gets named and you don't, the prospect may never run the search that would have surfaced your website. The shortlist formed before you could compete for the click — and some of your most important marketing moments now happen where you have no seat.

How Buyers Actually Use AI Assistants to Shortlist

To defend your presence, it helps to understand what people are actually asking. In my client work, the prompts that decide shortlists fall into a handful of patterns, each a moment where a brand can be named or left out.

Notice how few of these look like a traditional keyword. They're conversational, comparative, and intent-rich — a prospect isn't typing a keyword and browsing; they're asking an assistant to think for them. Being visible there is a different discipline from ranking a page, because the assistant filters before anyone reaches your site.

Why This Is a Discovery Layer, Not a Gimmick

It's tempting to dismiss this as a novelty used by early adopters. The usage numbers argue otherwise. Most credible estimates put ChatGPT's weekly active users in the hundreds of millions, with the other assistants adding sizeable audiences of their own — I break the figures down in my piece on ChatGPT usage statistics for 2026. Even a fraction of those sessions involving a purchase decision means an enormous amount of research now flows through answer engines instead of a results page.

What makes this a genuine discovery layer is where it sits in the funnel. AI assistants are increasingly the first stop for "help me figure out my options," the top-of-funnel moment that used to belong to search. When a tool shapes the initial consideration set, it holds real power over who gets considered at all. Being left out early costs more than ranking a few positions lower: you lose not a click but the awareness that would have led to one.

To be measured about it, AI recommendations aren't yet the only path buyers take, and plenty still cross-check with a conventional search. But the direction of travel is clear enough that treating your presence here as optional is a bet against where attention is moving.

How to Audit What AI Says About You vs. Competitors

You can't manage what you don't measure, so the first move is to stop guessing and go look. You can run this audit yourself in an afternoon: become the customer and observe what the assistants say. Approach it methodically rather than asking one flattering question and calling it done.

Run this the way you'd run a competitive review, because that's what it is. The discipline in my guide to competitor analysis for SEO in 2026 applies directly: you're mapping where rivals are strong, where the model trusts them, and where a gap exists that you can close. Keep a simple spreadsheet of prompts, brands that surfaced, and sources cited — that record becomes the brief for everything you do next.

Why AI Recommends Who It Recommends

Once you've seen who gets named, the natural question is why. No one outside these companies has the full recipe, but across many clients and industries the recommendations consistently favor brands that are easy for a model to understand and comfortable for it to vouch for. A few signals do most of the heavy lifting.

The through-line is trust and legibility. A model recommending a business puts a piece of its credibility on the line, so it favors brands whose quality and identity are well-evidenced across the open web. This is why the crawlable, well-marked-up foundation I build through technical SEO matters so much — if a model can't read your facts cleanly or reconcile conflicting information, it plays it safe and names someone clearer.

Improving Your Share of Voice: Generative Engine Optimization

The practice of earning a better place in AI answers has a name: generative engine optimization, or GEO. It's less a replacement for search optimization than an extension of it toward a machine audience, and I compare the two mindsets in my article on generative engine optimization versus traditional SEO. The goal shifts from ranking a URL to becoming an entity a model understands, trusts, and reaches for when it composes a recommendation.

Almost none of GEO is a trick. The work maps to a handful of pillars: a strong entity and About presence reinforced with schema; citations and mentions earned from reputable sources; a steady flow of authentic reviews; honest comparison and "alternatives to" content that gives the model your framing; structured data that types your facts; and consistency of your core information everywhere the model might encounter you. None of it is wasted effort, because the same work makes you stronger to human customers too.

Content is the engine underneath most of those pillars. Thin, generic pages give a model almost nothing to quote or trust, whereas substantive, expertise-rich material gives it clear evidence to lean on — the point of the expert content writing I produce for clients. To be described accurately, you first have to supply the raw material that lets a machine describe you at all.

A Concrete Action Plan to Get Named

Auditing tells you where you stand; this is how you improve it. I run the following sequence with clients, ordered so foundational work comes first. You don't need to finish everything before you see movement — even the first few steps often change how assistants describe a brand.

If sequencing all that feels heavy, that's why I package it as a coordinated program rather than a pile of disconnected fixes. You can see how the technical, content, and reputation pieces reinforce one another in my complete SEO solutions, which order the work so each layer strengthens the next instead of competing for attention.

Measuring Whether It's Actually Working

Measurement here is genuinely harder than the traditional kind, and I'd rather be honest than sell a false dashboard. Because these conversations are private and answers vary between users and even sessions, there's no clean "position" the way a rank tracker gives you. What you can build is a directional read on your brand's standing in AI answers over time.

The most reliable approach is disciplined, repeated prompt testing: run a fixed set of buyer questions across the major assistants on a cadence and log whether you appear, how you're described, and who appears alongside you. Pair that with your analytics, watching for referrals and inquiries from AI surfaces, which are increasingly identifiable. The metric that matters isn't a vanity number — it's whether the systems that shape shortlists understand you and put you forward.

What You Can't Control — and Shouldn't Try to Fake

Let me set expectations plainly, because this field attracts overclaiming. You do not control what an AI says about you. There's no setting to edit your entry, no way to guarantee you'll be named, no reliable trick to force a model's hand. Anyone promising certainty is selling something. What you influence is the public evidence the model reads — and you influence it by being clear, consistent, and credible, not by gaming it.

Two temptations are worth resisting. The first is manufacturing fake reviews or planting self-serving mentions; these systems are increasingly good at discounting inauthentic signals, and getting caught damages the reputation you're building. The second is chasing rumored "prompt hacks" that supposedly manipulate outputs, which are unstable and vanish with the next model update. The durable strategy is unglamorous on purpose: build a real business, document it well, earn honest trust, and make it easy for a machine to read. That's a position no competitor can quietly take from you inside a conversation you can't see.

Conclusion: Earn the Recommendation Before You Need It

The most important shift here isn't technical, it's mental. For years we optimized for a page everyone could see and audit. Now a meaningful slice of buying decisions is shaped inside private AI conversations where competitors get named, compared, and recommended without your knowledge. You can't watch those exchanges or buy your way in, and that can feel like a loss of control. I read it differently: it pushes every business back to fundamentals that were always the real work — a clear identity, a consistent story, an honest reputation, and content that proves you know your field.

Start by looking. Run the audit, see who's recommended in your customers' conversations, and note where you're absent or misdescribed. Then close the gaps methodically: define your entity, fix your inconsistencies, build authentic reviews, publish content worth citing, and earn mentions from sources the models trust. Do that patiently and you stop being the brand nobody thought to mention — you become the one an assistant is comfortable putting forward to a customer you've never met, which is where the next wave of growth will be decided.