90% of Local Businesses Are Invisible to ChatGPT. Here's Why That's Not About Your Website
The Ranking You Cannot See
Is your business ranked number one on Google and still invisible everywhere else? That's the situation most local business owners are sitting in right now without knowing it.
The SOCi Local Visibility Index looked at this gap directly in March 2026. Google's local 3-pack recommended businesses at a rate of 35.9%. Perplexity recommended them at 7.4%. ChatGPT recommended them at 1.2%.
That's not a small dip. That's a different game with different rules, and most businesses are still playing the old one.
An April 2026 audit of more than 60 businesses found the same pattern at a larger scale. Over 90% had zero visibility across ChatGPT, Perplexity, and Google AI Overviews combined. Not low visibility. Zero.
These were not businesses doing nothing right. Many had solid Google rankings, decent websites, active social profiles. The Google-era checklist was complete. AI still had nothing to say about them.
This matters because of where customers are actually starting their search now. A 2026 BrightLocal consumer survey found that 58% of consumers have already used AI tools to get local business recommendations. That number is only moving in one direction.
So the ranking you have spent years building on Google is real. It is also not the ranking that decides whether ChatGPT mentions you at all.
Where ChatGPT Actually Gets Its Answers
So where is ChatGPT actually looking when it answers a question about a business near you?
Not your website. Not your Google Business Profile, not directly anyway. April 2026 data pegged Foursquare as the source behind roughly 70% of ChatGPT's local business data. That is a wild number if you have never once thought about Foursquare since it stopped being a check-in app people used to compete for mayorships.
Perplexity leans somewhere else. Measurements from July 2026 put Yelp at about 62% of Perplexity's local citation share. Different model, different backbone, same underlying idea: these tools are not crawling the web fresh every time someone asks about a plumber in their zip code. They are pulling structured data from a small number of directories that have already done the work of organizing business information.
This is the part that trips people up. You can have a beautifully optimized website, a strong local pack ranking, years of SEO work behind you, and still be nowhere in these answers, because the model was never looking at your website to begin with. It was looking at Foursquare's data feed, or Yelp's, and if your listing there is thin, outdated, or missing entirely, you are invisible no matter how good the site is.
Reviews Beat Ratings, Every Time
a business with a 4.9 average across 12 reviews loses to a business with a 4.3 average across 400 reviews, most of the time, when an AI model decides who to mention.
That is the finding from Uberall's September 2026 analysis, which looked at more than 120,000 AI mentions across 3,793 locations. Review volume predicted whether a business showed up in AI answers more strongly than star rating did. Not by a small margin either. The pattern held across the dataset.
Think about why that might be true. A model pulling from Foursquare or Yelp is not evaluating your business the way a human weighing a five-star review against a four-star one would. It is looking for enough signal to be confident you exist, that you are active, that people are actually walking through your door and talking about it. Twelve reviews, even glowing ones, do not give a language model much to work with. Four hundred reviews, even mediocre ones, tell it you are a real, ongoing concern worth mentioning.
This flips a lot of local marketing advice on its head. Chasing a perfect rating while ignoring volume was a reasonable strategy for the Google era. It is a losing one for the AI era.
What to Fix This Month
Start with your listings, not your website. Pull up your business name, address, and phone number on Google, Yelp, and Foursquare side by side. If they do not match exactly, fix that first. A model treating Foursquare as a primary source has no way to reconcile "123 Main St" with "123 Main Street Suite 4" as the same business.
Next, check your schema. LocalBusiness schema on your site will not override what Foursquare or Yelp have on file, but it still matters for Google AI Overviews, which pulls from a wider net than ChatGPT does. Ask an LLM to review your page's code and flag what schema is missing if you are not going to do it by hand.
Then go get reviews, specifically on Yelp and Google, and specifically in volume. Given what the Uberall data showed about review count outweighing star rating, a business with 15 reviews should be worried less about protecting a perfect average and more about getting to 50, then 100.
Last, chase third-party mentions. A local blog post, a chamber of commerce listing, a news mention. These are the citations that build the kind of confidence a model needs before it will say your name.