Data · Local Search
Demand moved. Supply did not follow. The gap between those two numbers is an opening with a closing date.
By Izzy Gregorio · Updated August 2026 · 8 min read
In short
The local shortlist got shorter and almost nobody is on it. The old map pack showed ten businesses and let the customer choose. An AI answer names one to three and does the choosing. That is a structural change in how local demand gets allocated, and the businesses that fix their machine-readable signals first will take a share that is currently unclaimed.
The numbers
BrightLocal surveyed 1,002 US adults for its 2026 Local Consumer Review Survey and found 45 percent now use AI tools to find local businesses. The prior year the same measure sat at six percent. That is a seven-fold move in twelve months, and consumer behavior rarely moves that fast in any category. AI is now the third most-used tool for local recommendations, behind Google and Facebook and ahead of Yelp.
SOCi measured the other side of the exchange. Its 2026 Local Visibility Index covered more than 350,000 business locations across 2,751 brands and asked a simpler question: when someone asks, who actually gets named?
| Surface | How often a local business gets surfaced |
|---|---|
| Google three-pack | 35.9% |
| Gemini | 11% |
| Perplexity | 7.4% |
| ChatGPT | 1.2% |
Read those two studies together and the picture is uncomfortable in a useful way. Customer demand for AI-assisted local discovery grew seven-fold. The engines answering that demand name almost nobody.
That gap is not a market failure. It is an opening with a closing date.
The mechanism
The map pack was a list. You appeared alongside nine competitors and the customer compared you. Proximity, reviews, photos, hours, and the customer made a judgment.
An AI answer is not a list. It is a recommendation, usually naming one to three businesses, delivered as a sentence rather than a grid.
Look at what disappeared. The comparison step is gone. The customer is not weighing you against nine others, because they never see nine others. They see the ones the assistant decided were defensible picks, and everyone else is not in second place. They are absent.
The collapse has been measured directly. Joy Hawkins at Sterling Sky tracked AI local packs surfacing 5,943 unique businesses across a query set where the legacy three-pack surfaced 18,330.
Roughly 32 percent of the previous coverage. Two out of three businesses that used to appear somewhere no longer appear anywhere.
AI Overviews appear on roughly 68 percent of local queries against 39 percent for the traditional local pack, per Whitespark's 2026 study of 540 manual queries across six local industries. That average hides a split you need to see, because it decides which half of your funnel is disappearing.
| Query type | What it sounds like | AI Overview appears |
|---|---|---|
| Simple transactional | Tacos San Francisco. Plumber near me | 15%, and the local pack still appears on more than 90% |
| Informational | How long does an eye exam take near me | 92% |
| Hybrid | Average cost of dental implants in Phoenix | 97% |
If somebody already knows they want you, the map pack still delivers them. If they are still deciding who to call, the AI answer has almost entirely replaced it. The half you are losing is the half where the shortlist gets made.
Translate it
The failure mode differs by what you sell. Find your row.
| If you run a | What a customer is actually asking AI | Where you get filtered out |
|---|---|---|
| Home services company | Who can fix this, how fast, and do they cover my city | Service area and turnaround time are not published anywhere a machine can read. Vagueness reads as risk |
| MedSpa or clinic | Who does this procedure near me, and are they credentialed | Credentials sit in an image or a PDF instead of text. Outcome claims with no substantiation get dropped |
| Mortgage or real estate office | Who works with my situation, and are they licensed here | License and jurisdiction not stated in text. Inconsistent business name across listings splits the entity |
| Restaurant | Who is open now, what is the food actually like, can I get a table | Hours conflict across listings. Menu locked inside a photo or a script the crawler cannot read |
| Church or nonprofit | Which congregation near me teaches what I believe | Beliefs, service times, and who you serve are implied through design rather than stated in words |
The pattern is the same in every row. The information exists. It is just not in a form a machine can lift.
The framework
Five checks. Score each honestly out of five, then fix the lowest first. This costs nothing but an hour.
Named. Ask ChatGPT, Gemini, and Perplexity best [your service] in [your city] and write down every business named. Are you in any of the three answers? Then ask the research version, how much does [your service] cost in [your city], since that is where the AI answer has almost fully replaced the map.
Consistent. Is your business name, address, and phone identical across your website, Google Business Profile, Yelp, and every directory that carries you? Not similar. Identical.
Stated. Are your service area, hours, pricing range, and turnaround written in plain text on your site, not implied and not trapped in an image?
Corroborated. Does anyone other than you say you are good at this, on a page a crawler can reach? Reviews, coverage, roundups, listings.
Readable. Open yourdomain.com/robots.txt. Are GPTBot, OAI-SearchBot, Google-Extended, PerplexityBot, ClaudeBot, and Bingbot allowed?
Item five is the one that produces the most surprise. A robots.txt copied from a template can remove a business from AI answers entirely, and nothing in any dashboard reports it.
Keep going
Local AI answers change monthly and no dashboard tells you when your name stopped appearing. The prompt sets being run, which engines shifted, and the specific fixes that put a business back into the shortlist. Written for owners who need a decision, not a debate.
No spam. Unsubscribe anytime.
The honest read
Because being named requires evidence, and most local businesses have built for humans who were willing to do the evaluating themselves.
A human visitor will squint at a photo of your menu. A human will infer your service area from the towns in your gallery. A human will assume you are licensed because your truck looks professional. A retrieval system does none of that. It reads what is written and it declines to guess, which is the correct behavior when the answer is going to be delivered as a recommendation.
The businesses getting named are not the biggest ones. They are the legible ones.
One question worth sitting with. If a neighbor two miles away asked an assistant who to call for what you do, and it named three businesses, would yours be one of them? And if it named one of your competitors instead, how would you currently find out?
Common questions
Yes, and the shift was fast. BrightLocal's 2026 Local Consumer Review Survey of 1,002 US adults found 45 percent now use AI tools to find local businesses, up from six percent the prior year. That makes AI the third most-used tool for local recommendations, behind Google and Facebook and ahead of Yelp.
Usually because your competitor's information is more complete and more consistent across sources, not because their marketing is better. Retrieval systems decline to guess. A business with published service areas, stated hours, and corroborating third-party listings is a safer recommendation than one requiring inference.
Mostly research queries rather than booking queries. Whitespark's 2026 study of 540 local queries found AI Overviews on 92 percent of informational searches and 97 percent of hybrid ones, but only 15 percent of simple transactional searches, where the local pack still appears more than 90 percent of the time. If somebody already knows they want you, the map still delivers them. If they are deciding who to call, the AI answer has largely replaced it.
It contributes, because profile data feeds listings and citations that engines read. It is not sufficient on its own. Gemini surfaces a local business 11 percent of the time and ChatGPT 1.2 percent, per SOCi's 2026 index, so profile completeness alone does not produce a recommendation.
Typically one to three, compared with about ten in a traditional map pack. Sterling Sky tracked AI local packs surfacing 5,943 unique businesses where the legacy three-pack surfaced 18,330 across the same query set, roughly 32 percent of the previous coverage. Businesses not named are not ranked lower. They are absent from the answer entirely.
Crawler access, then information consistency. Confirm GPTBot, OAI-SearchBot, Google-Extended, PerplexityBot, ClaudeBot, and Bingbot are not blocked in robots.txt. Then make your name, address, phone, hours, and service area identical everywhere they appear. Both are free and both resolve quickly.
Start here
Run the five checks yourself, and you should. Or have it run properly. The AI visibility audit from Conspicuouz Creative Group puts your category's real questions to ChatGPT, Claude, Gemini, and Perplexity, records every business named in your place, and identifies which of the five checks is keeping you out of the answer.
Free. No obligation.