The Local AI Visibility Audit You Can Run Yourself

Quick read: A local GEO (Generative Engine Optimization) baseline audit is a repeatable way to check how AI systems like ChatGPT, Gemini, Perplexity, and Google AI Overviews describe, recommend, or ignore a specific business, before spending money to change any of it. Search Engine Land outlined the methodology on July 16, 2026, citing SOCi's 2026 Local Visibility Index: across roughly 350,000 business locations analyzed, ChatGPT recommended just 1.2% of them, Gemini recommended 11%, and Perplexity recommended 7.4% — compared to a 35.9% appearance rate for the same businesses in Google's local 3-pack.

Someone in your town types a question into ChatGPT right now: "who's a good [your kind of business] near me." There's a very good chance you're not in the answer. Not because you're bad at what you do. Because, per the numbers above, almost nobody is in that answer yet — and most local business owners have no idea whether they're one of the rare ones or not.

That's the actual problem with AI visibility right now. Not that it's hard to fix. That almost nobody's checking.

What a GEO baseline audit actually is

Search Engine Land published a walkthrough on July 16, 2026 of something the AI-visibility research world has been circling for a while: before you spend a dollar on "AI search optimization," you should know exactly where you stand today. Not guess. Check.

The framework is simple enough to steal, and I mean that as a compliment to whoever wrote it, not a knock. It comes down to four moves:

1. Benchmark first. Ask AI tools the kinds of questions a real customer would, and write down what comes back.
2. Fix eligibility and trust issues before touching content. If your Google Business Profile, hours, or reviews are wrong or thin, no amount of blog content fixes that.
3. Structure your information so AI has a reason to cite you. Clear, specific, accurate information beats clever copywriting here.
4. Repeat it on a schedule. A one-time check tells you where you stood on the day you checked. These systems change fast enough that a single snapshot goes stale.

That's the audit. It's not a tool you buy. It's a habit, done with the AI chat windows you already have open — the practical, DIY version of what Answer Engine Optimization is actually trying to measure.

Why almost nobody passes this yet

The stat set behind this matters more than the methodology, honestly. The same Search Engine Land piece cites SOCi's 2026 Local Visibility Index, which analyzed roughly 350,000 business locations and found ChatGPT recommended only 1.2% of them. Gemini did better, recommending 11%. Perplexity landed at 7.4%. Compare any of those to the 35.9% appearance rate those same businesses got in Google's ordinary local 3-pack, and the gap is enormous — AI engines are recommending a far smaller slice of local businesses than a plain Google search already surfaces.

Here's the part that should actually change how you think about this: that gap isn't really about competition. Traditional local SEO is genuinely zero-sum — three spots in the map pack, everyone fighting over them. AI recommendation isn't like that in the same way. It's closer to a pass/fail eligibility check than a ranking fight. Get the fundamentals right — accurate, specific, consistent information the AI can actually cite with confidence — and you're not competing against every other business in town for one of three slots. You're just trying to clear a bar most of them haven't cleared yet, because most of them haven't checked.

I ran a dog grooming salon for years before any of this existed. If someone had told me "there's a free, fifteen-minute check that puts you ahead of almost everyone else in your category who hasn't run it," I'd have found the fifteen minutes between the last dog of the day and closing up — because that's the trade this actually is. It's not a big technical project. It's an overlooked one.

This is the same gap our guide to local SEO for small business describes for traditional search, just wider, because almost nobody's doing the work yet.

How to run this yourself, this week

You don't need a subscription to a tracking platform to get a real baseline. You need four kinds of questions, four AI tools, and about fifteen minutes.

The four question types:

• Discovery — "Who's a good [your service] in [your town]?" This tells you if you show up at all.
• Comparison — "[Your business] vs. [a specific competitor], which is better?" This tells you how you're framed when you're already in the conversation.
• Trust — "Is [your business] legit / reliable / worth using?" This tells you what the AI has absorbed about your reputation.
• Logistics — "What are [your business]'s hours, and do they accept [a specific service or payment method]?" This tells you whether the basic facts on file are even correct.

Run each one in ChatGPT, Gemini, Perplexity, and a plain Google search with AI Overviews turned on. Sixteen questions total, all typed into free consumer tools.

For each answer, log four things:

• Did it mention you at all? (mention rate)
• What tone did it use — neutral, positive, skeptical? (sentiment)
• Was anything it said about you actually wrong? (factual accuracy)
• Where did it say it got the information, if it said anything? (cited source)

That log is your baseline. Not a score out of 100 from a vendor's proprietary model — which we've covered elsewhere, and it's worth understanding why those scores can't tell you much on their own — just a plain record of what four AI tools actually said about your business on a specific date.

Then fix what the log tells you to fix, roughly in this order: accuracy problems first (wrong hours, wrong services, wrong address — these are usually Google Business Profile issues), then trust signals (review recency and responses), then content depth (does your site actually answer the comparison and logistics questions the audit revealed you're weak on).

Then do it again in 60 to 90 days. Not because it's a ritual — because these systems retrain and re-crawl on their own timeline, and the only way to know if a fix worked is to ask the same questions again and compare.

What won't help

Buying an "AI visibility score" instead of running the check yourself. A proprietary score from a vendor tells you a number. It doesn't tell you what a real customer's actual question returns, in your own market, in plain language you can act on. Run the sixteen questions above first — it's free, and it's more specific to you than any benchmark averaged across thousands of other businesses.

Treating this as a one-time project. The whole value of a baseline is having something to compare against later. A single audit with no follow-up is a photograph, not a trend line.

Fixing content before fixing facts. If your hours are wrong or your Google Business Profile hasn't been touched in a year, new blog posts won't outrun that. Eligibility and trust come before content in the actual order these systems seem to weigh things.

Assuming a low mention rate means you're doing something wrong. A 1.2% recommendation rate from ChatGPT is the average across roughly 350,000 businesses, most of whom have never run this check. Showing up at all currently puts you ahead of most of your market — this isn't a sign you've failed, it's a sign almost nobody's tried yet.

FAQ

What is a GEO baseline audit?

A repeatable check of how AI tools like ChatGPT, Gemini, Perplexity, and Google AI Overviews currently describe, recommend, or ignore a specific business — done before spending money to change any of it, so you have something real to measure improvement against later.

How is this different from regular SEO tracking?

Regular SEO tracking watches rankings and clicks on a search engine results page. This tracks whether an AI system mentions your business at all when someone asks it a question conversationally, and whether what it says is accurate — a different, newer kind of visibility with no click-through data behind it.

Do I need special software to do this?

No. ChatGPT, Gemini, Perplexity, and Google's AI Overviews are all free to use directly. The audit is sixteen questions typed into tools you likely already have open, logged in a spreadsheet or even a notes app.

Why does ChatGPT recommend so few local businesses?

Per SOCi's 2026 Local Visibility Index, ChatGPT recommended only 1.2% of the roughly 350,000 locations analyzed — far below Gemini's 11%, Perplexity's 7.4%, or the 35.9% appearance rate those same businesses got in Google's local 3-pack. Nobody has a definitive answer for the exact cause, but the audit framework itself points at eligibility and trust signals — accuracy, consistency, review activity — as the likely gatekeepers, and most businesses simply haven't checked whether they clear that bar yet.

How often should I re-run the audit?

Every 60 to 90 days is a reasonable cadence for most small businesses — enough time for a fix to actually get picked up and re-surfaced by these systems, without checking so often that you can't tell a real change from normal noise.

What should I fix first if the audit turns up problems?

Basic accuracy — hours, address, services, and category, usually on your Google Business Profile — before anything else. Content and comparison-framing fixes matter, but they don't help much if the underlying facts an AI system can find about you are wrong.

Is a low score on this audit a bad sign?

Not on its own. Most businesses haven't run this check yet, so a low mention rate right now mostly reflects an unclaimed opportunity, not a mark against you specifically. What matters is whether it improves the next time you check.

---

Running this audit tells you where you stand. Acting on what it finds — cleaning up the facts, fixing the trust signals, making sure the right things get cited — is the ongoing work a Get Found on Google and AI engagement is built around, so the fifteen-minute check doesn't just sit in a spreadsheet.