Local SEO in 2026: What Changed When AI Started Answering

Local SEO did not stop working. It stopped being the whole job. A second front opened where the buyer never sees a results page at all, and most local businesses have no idea whether they show up on it or not.

Key takeaways

BrightLocal's February 2026 survey (n=1,002 US adults) found 45% of consumers now ask AI to find a local business, up from 6% a year earlier. ChatGPT leads at 31%, Google's AI Mode at 23%.
In our own dataset of 9,454 question-by-business checks across US local businesses, the business was not among the sources the AI answer used 84.3% of the time. About 94% of failure happens before the engine picks anyone at all.
The single strongest predictor of clearing that stage was showing up in Google organic results for the same question (odds ratio 12.74). Nine on-page technical fixes we tested did not survive statistical correction.
We can explain why a business is missing from an answer. We cannot promise you that any specific action will get you cited, and anyone selling you that formula is guessing.
The six local signals AI answers actually pull from are a map listing, real recent reviews, presence in the listing data engines query first, matching name and address and phone across directories, being named in local best-of roundups, and one clean page per service and town.

What actually changed in local search in 2026

The short version: a second search happened, and your business either shows up on it or it does not. BrightLocal's Local Consumer Review Survey (published February 11, 2026, n=1,002 US adults) found that the share of consumers using AI tools to find a local business rose from 6% to 45% in twelve months. ChatGPT was the most-used tool at 31%, with Google's AI Mode at 23%.

That number is the whole reason this post exists. A year ago, "how is my landscaping company doing in AI search" was a weird question. Today, close to half the people looking for a landscaper in your town at least try ChatGPT or Google's AI Mode before they open a map. If they do not get your name back, you never even had a chance to compete.

The traditional local SEO stuff still matters. Google Maps still runs the world for "landscaper near me". But there is now a parallel channel where the buyer types a full sentence like "who does small backyard drainage jobs in Rochester" and reads a paragraph answer instead of scrolling ten listings. The mechanics of showing up in that paragraph are different, and that is what this post is really about.

Where each AI engine actually looks for a local business

Every AI answer engine has a favourite place it goes first when someone asks about a local business. If your business is not in that place, or the info there is wrong, you are already out of the running before the engine even writes its answer.

Here is what we know so far, based on public partnerships and observable behaviour:

ChatGPT leans on Foursquare's Places API for local business data. OpenAI announced that partnership on December 16, 2024, and Foursquare's dataset covers more than 100 million points of interest across 200+ countries. On July 23, 2026, OpenAI added Yelp: reviews, ratings and photos now flow into ChatGPT, with reservations and waitlist added on August 10, 2026. When those come up dry, ChatGPT falls back to Bing Places.
Google's AI Mode and AI Overviews are grounded in Google Maps. Your Google Business Profile is not just a listing here. It is the primary source.
Perplexity leans heavily on Yelp, TripAdvisor and industry-specific directories.
Claude searches through Brave Search, which pulls from its own independent index.

You do not need to memorise this. You do need to notice that no two engines look in exactly the same place, and none of them look at your website first. Your website is what they check LAST, once they have a shortlist of names from somewhere else. This is the biggest single reversal from ten years of local SEO advice, and it is the reason a lot of the old checklist stopped moving the needle.

A Foundation and AirOps analysis of 28 million AI responses in Q4 2025 found that Yelp alone received 512,680 citations, 3.4 times the next platform on the list. The same study found that 91% to 97% of AI citations came from category questions ("best plumber in Cleveland") rather than from someone typing a specific business name. Category questions are how strangers find you. And strangers are most of your growth.

The one finding that matters most

We ran a study of 9,454 question-by-business observations across US local businesses. For each observation, we asked an AI engine a real local question, then checked whether the business appeared among the sources the answer used.

Two numbers matter.

First, the business was not among the sources the answer used 84.3% of the time. When you sort out where failure actually happens along the path from question to citation, roughly 94% of failure happens before the engine even picks a shortlist. The business is not in the pool being considered. Everything else, page quality, wording, review sentiment, is competing for the last 6%.

Second, we tested a long list of things you might think would fix that. Nine common technical fixes on the website itself: none survived Bonferroni correction, which is the statistical bar for "this is a real effect and not noise." The one thing that stood out was whether the business showed up in Google's regular organic results for the same question. That relationship was strong: odds ratio 12.74, with a McFadden R-squared of 0.24 for the model overall.

Here is the important part, and I am going to be careful with the words because most of the local-AI industry is not: this is an association, not a task. Businesses that already have Google presence for a question are far more likely to be among the sources an AI answer used. We cannot tell you that fixing your Google presence will cause AI to cite you. Nobody can tell you that yet. What we can tell you is that when a business is missing from an AI answer, Google presence is the strongest place to look first.

The other thing I want to say clearly: we did not see AI "not find" anyone. What we saw is which sources an engine's returned answer used. Anyone who tells you that "AI could not find your business" is overreaching. The honest sentence is "your business was not among the sources this answer used." That distinction sounds pedantic. It is not. It is the difference between a real diagnosis and a made-up story.

We wrote up the whole finding in more detail in our post on where local businesses fall out of the answer if you want the numbers behind the numbers.

The six signals AI answers actually pull from

For a local business, there are six inputs that consistently show up in the sources AI answers use. This is the spine of the local track, and every specific tactic below fits under one of these.

1. A clean, complete map listing

Your Google Business Profile is not a formality. It is a database entry that Google's AI Mode and AI Overviews read directly, and that Bing Places (and by extension ChatGPT's fallback) reads too. Missing hours, no service area, no service list, an old address, a category that does not match what you actually do: any of these can be the reason your business is not in the pool the engine looks at.

A landscaping company I looked at recently had "Landscaper" set as its primary category but did nothing to indicate they mainly did small-job drainage and grading work. When the local question was "who fixes soggy backyards in Rochester", they were not in the shortlist. Their competitor, who had drainage listed as a service on the profile, was. Same size company. Two trucks each. One was in the answer. One was not. Details on this in the Google Business Profile deep dive.

2. Real reviews that are recent and specific

Reviews do two jobs for AI answers. They tell the engine you are a real going concern (recency signal), and they give the engine specific language to match against the question. A review that says "Mark did a great job on our French drain" is worth ten reviews that say "great service."

Here is where the Yelp deal matters. Since July 2026, ChatGPT reads Yelp reviews directly. So does Perplexity. If your Yelp presence is a wasteland, that is a hole. Google reviews still do the heavy lifting for Google's AI Mode. The count number that gets thrown around in local-AI circles ("you need 150 reviews before AI will cite you") has no published methodology behind it, so I will not repeat it. What is true is that recency and specificity both matter more than raw count. More on the mechanics in how to get more reviews and why AI reads them differently.

3. Presence in the listing data AI queries first

This is the one most owners have no visibility into. Every engine has a place it looks first, and that place is not your website. For ChatGPT it is Foursquare, then Yelp, then Bing Places. For Perplexity it is Yelp and TripAdvisor and industry directories. For Google's AI Mode it is Maps.

If Foursquare has you listed at your old address, that is what ChatGPT sees. If your industry directory (like Angi for contractors, or Zocdoc for medical) has an incomplete profile, that is what Perplexity sees. Our breakdown of where each AI actually pulls local business names from covers the specific data sources for the main engines.

4. Matching name, address and phone across every directory

Quick note on wording, because it trips people up. In local SEO, a "citation" is a directory listing of your name, address and phone. In AI search, a "citation" is when an engine names your business and points at your page. Same word, different meanings. From here on I will say "directory listing" when I mean the local-SEO kind.

Consistency across directory listings is not glamorous work. It is also one of the six signals. If your business is "Rochester Grading and Drainage" on Google, "Rochester Grading & Drainage LLC" on Foursquare, and "R.G.D. Landscaping" on Yelp, the engines can't confidently link those to one entity. The pool of things they might name shrinks. Not because they cannot find you, but because they cannot confidently say all three records are the same business. Local directory listings versus AI citations has more on why these two things get mixed up.

5. Being named in local best-of roundups

This is where most local businesses are quietly losing. When someone asks ChatGPT "who are the best landscapers in Rochester", the engine often reaches for local roundups: neighbourhood publications, HomeAdvisor top-10 lists, local Reddit threads, regional trade association pages, city subreddits, the local paper's "best of" issue. These roundups get cited disproportionately.

The Foundation/AirOps study I mentioned earlier found that 91% to 97% of AI local citations come from category-style questions, and those questions are exactly what best-of lists answer. If you are not named in any of them, the engine has no third-party source to point at. If you are named in three, it does. More on which lists actually get quoted in the local directories and best-of lists AI quotes back.

6. One clean page per service and per town

This is the one exception to the rule that local businesses do not need to write content. You do not need a blog. You do need one clear page for each service you offer, in each town you serve. A landscaping company that does drainage, grading, and sod installation across three towns should have nine pages, each one specific.

Not junk pages. Not the same page with the town swapped. Real pages with the actual streets you have worked on, actual jobs, actual before-and-after photos, actual pricing bands if you can share them. The reason: when the engine has narrowed to a shortlist and needs to write a paragraph, it grabs specifics from the pages of the names on its shortlist. If your page for "grading in Pittsford" is empty, the engine gives that space to a competitor. How to build service area pages without making 40 junk ones walks through the sensible version.

What this looks like for a two-truck landscaping company

Say you run a landscaping business with two trucks. You cover three towns. You have a Google Business Profile with 43 reviews averaging 4.7 stars. Your website is on a template from 2019. You have never touched Foursquare. Your Yelp page exists but has three reviews. You have never been in a local best-of list.

Here is what an audit tends to find, in order of what actually matters:

Your Google Business Profile has "Landscaper" set, but not the specific services (drainage, grading, sod, tree work). AI Mode has no way to match "who does grading in Pittsford" to you. Fix in an hour.
Foursquare has you at your address from three years ago, and lists you as closed. ChatGPT is answering local questions with your competitor because your Foursquare record disqualified you. Fix in a week (you have to claim the listing).
Your Yelp page has three reviews from 2021. Since July 2026, ChatGPT reads Yelp directly. Two dozen fresh reviews would change your shortlist odds in ChatGPT more than anything you could do on the website itself.
There is a Rochester-area home-services roundup on a local blog. You are not on it. Your competitor is. Reach out to the writer.
Your website has one "Services" page listing everything. Splitting that into a page per service per town is a weekend of work.

Compare that to what most agencies would tell this owner: rewrite the homepage, add schema, start a blog, publish weekly, build backlinks. None of those five things is on the list above, because none of them addresses where the engines actually look first. If you have burned money on that kind of program before, this post on why past SEO work stopped paying off is worth a read.

What we do not know yet, and won't pretend to

A few things you will see other people claim confidently. I want to be direct about which ones are guesses.

"Foursquare drives 60 to 70% of ChatGPT's local results." This number is everywhere. There is no published methodology for it. We know Foursquare is the announced partner. We do not know the share. I will not repeat a specific percentage.
"You need 150 reviews to get cited by AI." Same problem. No methodology, no dataset, wide repetition. Recency and specificity matter. A hard threshold is not something I can defend.
"Do X on your website and AI will start citing you." We did not find any single on-page fix that survived statistical correction in our 9,454-observation dataset. Something on your page might help. We cannot promise which thing, or by how much.

The honest version: get the six signals in order, and your business is far more likely to be among the sources the engine reaches for. That is the strongest claim the data supports.

How to start without doing everything at once

If you own the business and you have thirty minutes this week, do this:

Search ChatGPT for "best [your trade] in [your main town]". Note whether you appear. Note who does.
Do the same in Google's AI Mode.
Look up your business on Foursquare and Yelp. Check that address, phone, hours and categories match Google.
Note which of the six signals above you are weakest on.

That is more diagnostic work than most of your competitors have done. If you would rather have someone else run the check across all the main engines in one pass, that is what the free AI Visibility Check does. Details below in the FAQ.

If you know your trade, there are also per-trade guides in this cluster: dentists, roofing companies, plumbers, HVAC, contractors, and med spas. Each one walks through the same six signals for that specific trade's answer patterns and directory sources.

Frequently asked questions

Is local SEO dead in 2026?

No. Google Maps is still where most of your near-me traffic comes from, and Google's AI Mode is grounded in Maps data. What has changed is that Maps is no longer the whole picture. About 45% of consumers now also ask an AI tool when looking for a local business, per BrightLocal's February 2026 survey. If your Google Business Profile is your whole strategy, you are covering less of the market than you used to.

Does my website matter for AI search anymore?

Yes, but not first. The engines pull a shortlist from directories, map listings and review platforms, then look at the pages of the businesses on that shortlist to write their answer. If you are not on the shortlist, the website could not save you. If you are, your service pages are what fills in the paragraph. So it matters, at the last mile.

Why do reviews on Yelp matter now if my customers only use Google?

Because ChatGPT reads Yelp directly since July 2026, and Perplexity has always leaned on Yelp. The customer you already have may only use Google. The next customer, who is currently typing into ChatGPT, is a Yelp-informed answer whether they visit Yelp themselves or not. It is not about where your customers post reviews. It is about where the AI reads them.

How do I know if I am being recommended by AI right now?

Search the questions your customers actually type, in the engines they actually use, from your local IP. That means asking ChatGPT, Google's AI Mode, Perplexity and Claude the same category question ("best [trade] in [town]") and reading who each one names. How to check if AI recommends your business walks through a seven-minute version. The free AI Visibility Check does it across engines automatically.

What is the $27 Local AI Visibility Playbook and how do I get to it?

It is the follow-on to the free AI Visibility Check. When the free check identifies you as a local business, it routes you to a version of the playbook built for local, with the six-signal audit, the directory list per engine, and the service-area page template. You reach it through the free check, not through a separate page. Start with the check and it will offer the playbook if it fits.

Where to start this week

Run the AI Visibility Check on your business, in the towns you actually serve, using the questions your customers actually ask. Read who each engine names instead of you. Cross-check those names against Google Maps, Foursquare, Yelp and any local best-of list you know. The gap will show you which of the six signals is your weakest, and that is the one to fix first. Do not try to do all six at once. Fixing the weakest one usually pulls the others along.