Here's the uncomfortable part: it's almost never a reputation problem. A clinic with 60 reviews can be getting named over one with 144. The AI isn't judging who's best — it's reading a short list of machine signals about each business and recommending the one whose signals are clearest. Fix the signals and you change the answer.
First — what "showing up in ChatGPT" actually means
The blue links aren't the front door anymore. When a buyer asks an AI assistant — ChatGPT, Google's AI Overview, Gemini, Perplexity — for a recommendation, the model doesn't show ten ranked results. It writes a two-or-three-sentence answer that namesa small handful of businesses, and most buyers stop there. They don't scroll. They don't compare ten options. They take the recommendation.
Getting named in that answer is a different game from classic SEO, and it has a name: GEO — Generative Engine Optimisation.SEO wins a click. GEO wins the recommendation. They share plumbing, but the outcome you're optimising for is "is the AI confident enough to say your name," not "did you rank on page one."
Why the AI goes missing on you
An AI assistant can't visit your shop, read your Google reviews like a human, or get a feel for the room. It assembles a recommendation from structured, checkable signals — and it leans toward the business whose signals leave the least doubt. You go missing when those signals are absent, thin or contradictory. The three failure modes we see most:
- The profile is half-built. A Google Business Profile with the wrong category, missing services, stale photos and a review count that hasn't moved in months tells the model you're not the active, established option in your suburb.
- The site says nothing the machine can read. No structured data (schema) means the AI has to guess what you are, where you are and what you're rated — and it would rather name a competitor it doesn't have to guess about.
- Your identity is blurry across the web. Your name, address and phone appear three slightly different ways across Google, directories and your own footer. The model can't be sure you're one established business, so it discounts you.
The signals that decide who gets named
If you fix one thing, fix the signals below — in roughly this order of impact. None of them are tricks. They're just the inputs these models actually read.
1. Google Business Profile
The single biggest input into who the local pack and Google's AI Overview name. Rating, review count and recency, correct primary and secondary categories, a complete services list, real photos, and regular posts. Most businesses leave half of this blank. Completing it properly is the highest-leverage hour you'll spend.
2. Machine-readable schema
Schema is structured data — invisible to humans, decisive to the AI. The right LocalBusiness, Serviceand review markup tells ChatGPT and Google exactly what you are, where, what you offer and how you're rated, with no guessing. It's the difference between the model inferring your business and the model knowing it.
3. Entity clarity & NAP consistency
"NAP" is name, address, phone. When those are identical everywhere you appear online, the model is confident you're one established, trusted entity. When they conflict — an old suite number here, a different phone format there — that confidence drops, and confidence is exactly what the AI is rewarding.
4. Reviews — volume, velocity and the words inside
Review count matters, but so does the rate they arrive and what they actually say. A steady flow of recent reviews that mention the service and the suburb in plain language gives the model both the trust signal and the keywords it's matching against. A pile of five-year-old reviews reads as a business that may not even be open.
5. Third-party corroboration
Directory listings and "best [category] in [suburb]" listicles act as independent confirmation. When other sites name you in your category, the model treats that as corroboration that you belong on the shortlist — which is why being mentioned in the right third-party places quietly moves the needle.
How to check where you stand — in ten minutes
You don't need a tool to get a first read. Ask the engines the way a buyer would:
- Open ChatGPT (with browsing), Google's AI Overview, Gemini and Perplexity, and type your real buyer prompts: "best [your category] in [your suburb]", "who should I book for [service] near [suburb]", and a few niche variants.
- Note who gets named — and whether you're one of them. Run each prompt two or three times, because the answers are non-deterministic and the named businesses can change call to call.
- Then check your foundations: is your Google Business Profile complete and recent? Does your site carry
LocalBusinessschema? Is your name, address and phone identical everywhere it appears?
If you're not getting named and your foundations are thin, you've found your problem — and the good news is that the fixes are concrete, not mystical.
What to do about it
In order: complete the Google Business Profile properly; add the right schema to your site; clean up NAP consistency across Google, directories and your own pages; build a steady review habit; and earn a few third-party mentions in your category. The foundational fixes land in days and Google's index usually reflects them within weeks. The rest — reviews, corroboration — compounds over weeks to months. It's closer to ongoing maintenance than a one-time switch, because the engines re-crawl and re-answer continuously and competitors keep moving.
The honest bit: likelihood, not guarantees
Plenty of people will sell you "we'll get you #1 in ChatGPT." That's misleading, and it's also not how these models work — ask the same question twice and you can get different businesses named. So we sell citation likelihood, never a guarantee. The fixes above genuinely raise the odds the AI names you, because schema, reviews and entity clarity are how these models decide who to cite. Anyone promising a fixed ranking is promising something the technology can't deliver.
See exactly who the AI names instead of you
You can do all of this yourself with the checklist above. If you'd rather have it measured and prioritised, that's exactly what our GEO Auditdoes: we run your real buyer prompts live across the engines, show you precisely who's being named instead of you and why, grade every signal, and hand back a prioritised fix pack — with every line labelled measured or inferred. There's a free one-page mini-audit if you just want the single most damaging finding first.
Further reading on the foundations behind this: our production + marketing retainer (the video and pages the AI ends up quoting), and our pricing.