I build websites for local businesses. For most of my career that job had one scoreboard: rank on Google, make the phone ring. Then a homeowner told me she'd hired a contractor because "ChatGPT recommended them," and I realised the scoreboard had quietly been swapped out while I wasn't looking.
So I did the thing I do when something doesn't add up. I went and checked the numbers — first on my own clients, then across every study I could find.
The number that made me stop
Here is the stat that reframed the whole problem for me. In SOCi's 2026 Local Visibility Index, which looked at more than 350,000 business locations, ChatGPT named a specific local business in only 1.2% of category prompts — versus 35.9% visibility in Google's local 3-pack. Gemini sat at 11%, Perplexity at 7.4%. Being recommended by an AI engine is, on the current evidence, somewhere between 3x and 30x harder than ranking on Google.
That would be a footnote if nobody used these tools. But they do. ChatGPT reported 800 million weekly active users by October 2025. Google's AI Overviews crossed 2 billion monthly users by mid-2025. And BrightLocal's 2026 consumer survey found that 45% of consumers now use AI tools to find local businesses — up from 6% a year earlier, making AI the third most-used local discovery channel behind Google and Facebook.
Put those together and you get the problem in one sentence: a fast-growing share of buyers now ask a machine "who's good near me," and most local businesses are simply not in the answer.
Why a great Google rank doesn't carry over
The instinct — mine included — is "I rank #1, so the AI will pick me." It doesn't work that way, and the reason is structural.
Google hands you ten links and lets you choose. An answer engine chooses for you and names two or three businesses. Those are different jobs with different inputs. Google rewards links, page authority, and on-page SEO. An answer engine is trying to assemble a trustworthy recommendation from whatever it can verify about you — and ranking is only a weak input. Ahrefs found that the share of AI Overview citations coming from the top 10 organic results fell from about 76% in mid-2025 to roughly 38% by early 2026. Your rank gets you in the room. It no longer wins the recommendation.
The competitor who keeps getting named instead of you is often not outranking you on Google. They're winning a different game you didn't know you were playing.
The four questions I kept coming back to
After auditing enough sites by hand, the failures sorted themselves into four buckets. Every business that was invisible to AI was failing at least one of these:
- Recognize — Does the model know you exist as a specific, real local entity? Or does it confuse you with a similarly-named business two provinces over?
- Evidence — Is there enough third-party proof (reviews, named mentions, real service detail, pricing) to justify recommending you over the next option?
- Answer — When a customer asks a real question, does a page on your site answer it in plain, quotable words the engine can lift verbatim?
- Link — Can AI crawlers actually reach and parse your pages? Or are you blocked, JavaScript-walled, or invisible to the specific bots that feed citations?
Recognize, Evidence, Answer, Link. That's the REAL Method — the scorecard I now run every audit against. It is not an official ranking from Google, OpenAI, or anyone else; nobody publishes that. It's a practical model that tells me, on any given Tuesday, which lever to pull first.
What I built, and why I gave the diagnostics away
I started by scripting the checks for my own clients. It worked well enough that I turned it into RankingLocal.ai. It runs the REAL layers across seven engine profiles — ChatGPT, Google AI, Perplexity, Claude, Gemini, Copilot, and Grok — scores each one, and tells you in plain English which fix moves the number.
I made the diagnostic tools free and ungated on purpose. You can run the Robots Check, the GEO Grader, the Schema Generator, and the AI Visibility check with no signup and no card. My reasoning is the same one I use as a contractor: I don't ask for a deposit before I've measured the kitchen. If the free check shows you have a problem and the fix is worth paying to monitor, that's when Flare earns its keep.
The thing I wish someone had told me earlier
If you run a local service business, your next fifty customers are going to split. Some will Google you. A growing share will ask an AI. The ones who ask the AI never see your Google rank — and right now, for most local businesses, the AI doesn't have enough to go on.
The good news is that most of the fix is mechanical, not magical. It's schema, crawler access, real answers on real pages, and third-party evidence. You don't need an agency to start. You need to know which of the four layers is failing — and then you need to watch it, because AI answers drift week to week.
Run your own domain through the free AI Visibility check and you'll see your four REAL numbers in about a minute. If something looks off, hello@rankinglocal.ai reaches me directly.