ChatGPT names a specific local business in only about 1.2% of searches where a local recommendation would make sense. Compare that to Google's own Local Pack — the map and three listings you see under most "near me" searches — which shows up on roughly 36% of local searches. Perplexity sits in between at around 7.4%, and Google's Gemini around 11%.
Those aren't small differences. They're different games with different odds, and most businesses are only playing one of them.
Four judges, four sets of standards
Think of Google's Local Pack as a busy high street — lots of shops get a look-in, because the bar to appear is "you exist nearby and you're not terrible." AI assistants behave more like a single, trusted friend recommending one place. Fewer businesses get named, but the ones that do get taken seriously.
- ChatGPT — recommends a named business roughly 1.2% of the time, and typically only businesses averaging 4.3★ or higher.
- Perplexity — around 7.4% of the time, threshold closer to 4.1★, and leans heavily on how fresh and clearly structured your information is.
- Gemini — around 11% of the time, threshold around 3.9★.
- Google Maps / Local Pack — will still show businesses rated 3.5–4.0★ if they're close enough, because proximity does a lot of the work there.
They don't even agree with each other
Here's the bit that catches people out: getting cited by one AI assistant doesn't mean you'll show up on another. Analysis of over 680 million AI citations found only around 11% overlap between what ChatGPT and Perplexity actually cite. They're reading different signals, weighting them differently, and in some cases pulling from different sources entirely.
You're not optimising for "AI." You're auditioning in front of several separate judges, each with their own scorecard.
That's a genuinely different job to traditional SEO, where ranking well on Google was more or less the whole task.
What actually moves you across the line
None of this is mysterious, it's just more demanding than most businesses realise:
- Get your genuine rating above the AI thresholds — not just above 3.5, but comfortably into the low-to-mid 4s if ChatGPT specifically matters to you.
- Keep everything current — recent reviews and recent photos read as "still active." A profile that hasn't moved since 2023 gets quietly passed over.
- Be specific, not vague — "emergency callouts, same-day, covering Swindon and Wroughton" is something an AI model can lift cleanly. "General plumbing services" isn't.
- Match everywhere — your name, address and phone number need to say the same thing on every site that mentions you. See our NAP consistency guide if you're not sure yours does.
A 4.9★ profile that's been silent for two years will still lose to a 4.4★ profile that posted last week and replies to every review. Recency and specificity beat perfection.
Curious where you stand right now? FoundByAI's free checker asks AI about your business live, the same way a customer would, in about 20 seconds.
Run the free checkSee exactly where you stand with AI assistants
The £99 clean-up includes a full audit of what ChatGPT, Perplexity and Google AI currently say about your business — with screenshots — plus the fixes written out ready to action.
Get the £99 clean-up →