Skip to main content

AI Search won’t recommend your business? Here’s how to fix it

By Jason Roy•
AI Search won’t recommend your business? Here’s how to fix it

A year ago, almost nobody asked a chatbot to find them a plumber. Now nearly half of consumers do.

BrightLocal's 2026 Local Consumer Review Survey of 1,002 US adults found that the share of people using AI tools for local business recommendations jumped from 6% in 2025 to 45% in 2026. ChatGPT was used by 31% for business recommendations and Google's AI Mode by 23%.

Here's the uncomfortable part. Those assistants don't agree with each other, and they don't agree with Google's map results either. A business can hold the top spot in the local pack and still never get named when a customer asks ChatGPT who to call.

Each assistant reads different sources

Steady Demand's AI Citation Ledger ran about 1,500 local-service queries across 50 US metros and 10 categories, then analyzed more than 10,000 citations. The engines barely overlapped. ChatGPT and Gemini named the same top business only 4.9% of the time.

The reason is where they look:

  • Gemini leaned heavily on businesses' own websites, which made up roughly half of its citations.
  • ChatGPT cited business websites just 10.5% of the time. General business directories made up 46.0% of its citations, compared with 17.7% for Gemini.

A separate BrightLocal study of 1,355 business locations and nearly two million citations found the same split from another angle. Google Business Profile data was the top source for AI Overviews and AI Mode, with Yelp second. In ChatGPT, Yelp was the top source and showed up in 80% of answers.

What it means

There's no single profile to fix. A plumber with a polished Google Business Profile but a thin website may do fine in AI Mode and disappear in Gemini. One with a great website but a neglected Yelp listing may vanish in ChatGPT.

The safest move is making your information complete and consistent everywhere an assistant might look. That's less glamorous than a new tactic, but it's where the citation data points.

Fix 1: Treat Google Business Profile as the foundation

Google's own help page on improving local ranking says local results are based mainly on relevance, distance and how well-known a business is. It also states that businesses with complete and accurate information are more likely to show up. Since Google's AI features pull directly from Business Profile data, the same logic carries over.

What to do

  • Pick the most specific primary category that fits, and add secondary categories only for services you actually offer.
  • List every service by name. "Water heater installation" and "sewer line repair" give an AI something to match. "Plumbing services" doesn't.
  • Set your service areas to the towns and neighborhoods you really cover.
  • Check hours, phone number and website link every quarter. Stale details are the fastest way to lose trust with both people and AI systems.

If you want to know which profile numbers to watch while you work on this, our list of Google Business Profile metrics covers them.

Fix 2: Build service pages that answer what customers ask

Gemini's reliance on business websites makes your site more important for local AI visibility than many owners assume. But a homepage with a list of services doesn't give it much to cite.

What to do

Create one page per core service, and make each page answer the questions a customer would ask before calling:

  • What the service involves and how long it usually takes
  • Which towns or neighborhoods you cover for it
  • Who it's for (homeowners, landlords, restaurants)
  • What makes your approach different, with specifics: licenses, warranties, years doing this exact job

Write these the way you'd answer on the phone. Then add LocalBusiness structured data so your name, address, hours and services are machine-readable. Our guide to how structured data shapes AI visibility explains which properties matter.

Fix 3: Clean up the directories ChatGPT actually reads

ChatGPT leans on directories more than any other engine in the Steady Demand data. If those listings are incomplete, outdated or inconsistent with your website, you've handed it a reason to recommend someone else.

What to do

Start with Yelp, since it was ChatGPT's top source in the BrightLocal study. Then cover the general directories and the ones specific to your industry: the trade association listing, the platform customers in your field actually use, your local chamber of commerce.

On every one, your business name, address and phone number should match your website exactly. Our NAP guide explains why small mismatches cause big problems, and our list of directories for local SEO is a good starting checklist.

Quick win: Search your business name in ChatGPT and ask it to list its sources. Any directory it cites with wrong or missing details goes to the top of your cleanup list.

Fix 4: Get reviews that describe the work

AI users don't take recommendations on faith. BrightLocal's follow-up report on AI trust found that 97% of people who use AI for local recommendations sometimes double-check them against real reviews, and 42% always check reviews on the native review platform.

So reviews do two jobs. They feed the AI's answer, and they confirm (or undercut) it when the customer goes to verify.

What to do

A review that says "great service!" helps a little. A review that says "they replaced our tankless water heater the same afternoon" helps far more, because it ties your business to a specific service in words an AI can match to a specific question.

When you ask for reviews, ask customers to mention the job you did for them. Don't script the wording. Just the reminder is enough. And spread your effort: Google usage for reviews fell from 83% to 71% in the same BrightLocal survey, which means the other platforms customers check matter more than they did. For a deeper process, see Review Building 101.

Earn mentions beyond your own listings

Your own website and listings are only part of what an assistant can find. A feature in the local paper, a listing on a supplier's "find an installer" page or a joint project with another local business all put your name on pages you don't control. They're also the kind of independent confirmation a customer finds when they go to double-check. Our guide to local business partnerships has ideas that work for small teams.

Check where you stand every month

You can't fix what you haven't measured, and AI answers move around. In the Steady Demand data, Gemini's cited sources overlapped only about 40% to 46% when the identical question was asked twice. One check tells you very little.

  1. Write down the 5 to 10 questions your customers actually ask, such as "best emergency plumber in [town]" or "who installs heat pumps near [neighborhood]."
  2. Ask each one in ChatGPT, Gemini and Google AI Mode.
  3. Record who gets named and which sources are cited.
  4. Repeat monthly and watch the trend, not any single answer.

The sources column is the useful part. It tells you which listing or page to fix next.

Say a plumber finds that ChatGPT keeps naming a competitor and citing Yelp, while Gemini names them but quotes an outdated service list from their own site. That's two specific jobs: update the Yelp listing and its photos, then rewrite the service page Gemini is reading. Neither would show up in a rankings report. Both show up in a 20-minute monthly check. For a broader approach to tracking, see how to track brand mentions across AI assistants.

See what AI already says about you

Start with a baseline. Run your business through the LLM Visibility Checker to see how AI assistants describe you today, then run a site audit to make sure the service pages you're building can be crawled and read cleanly.

Dominate search today on Google and AI Engines.

Join 85,000+ SaaS Marketers, Growth Agencies, Content-Led Companies and E-commerce Brands.

See Pricing
Dashboard preview showing SEO site checkup metrics, page group insights, and issue prioritization