Almost Half of Customers Now Ask AI for a Local Recommendation: The 30-Minute Check That Tells You If It Names You

What's happening. Google's AI Mode, the chat-style answer layer inside Search, has crossed a billion monthly users, and queries running through it keep climbing every quarter. BrightLocal's Local Consumer Review Survey 2026, published February 11 on a panel of 1,002 US adults, found 45% of consumers now use an AI tool to find a local business or service. A year earlier that number was 6%. Very few customer behaviors move that far that fast.

Now the other half of the picture. SOCi's 2026 Local Visibility Index scored more than 350,000 locations across 2,751 brands on 120-plus metrics, and found ChatGPT recommended 1.2% of the locations it studied and Perplexity 7.4%, against 35.9% of those same locations appearing in Google's local 3-pack. Read the fine print on that study: it only covers brands with 50 or more locations. Companies with real marketing departments and national budgets are mostly invisible inside AI answers. Your single location was never in the sample.

The so-what for an operator. Page one of Google handed a customer ten links and a map, and being sixth still got you a call sometimes. An AI answer names two or three businesses and stops. The shortlist got shorter, and nobody sent you a notice when you fell off it. Most owners have never once checked.

The move: the 30-minute AI visibility check. Run it today, then again in 60 days.

1. Write the questions a real customer types. Use the words a panicked homeowner or a busy office manager would say out loud, and drop the industry terms. The prompt for this is below.
2. Run them as a stranger. Log out of every AI account, use a private window or a temporary chat, and run the same list through four surfaces: ChatGPT, Google AI Mode, Gemini, and Perplexity. Logged in, the tools bend toward what they already know about you, and you will see a flattering answer no customer gets.
3. Score every question three ways: Named, Mentioned in passing, Absent. Write down which businesses got named instead. That list is your real competitive set, and it is usually not the competitor you have been watching.
4. Ask for the receipts. After each answer, ask: which sources did you use to pick these businesses? You will get back review sites, directories, roundup pages, and profiles. That is your fix list, written by the machine you are trying to influence.
5. Check what it says about you. Wrong hours, a price you raised last spring, a service you dropped, a permanently closed flag. Errors cost you more than rankings do, and they are the fastest thing on this list to fix.
6. Fix the three inputs that actually move AI recommendations. Your name, address, phone, category, and service list identical everywhere they appear. Reviews from the last 90 days, since recency carries more weight than a wall of five stars from 2023. A profile touched in the last 30 days with current services, photos, and plain answers to the questions customers ask.
7. Re-run the identical list in 60 days. Same questions, same wording, same logged-out setup. That turns a gut feeling into a measurement.

The prompt for step 1:

Role: You are a local search strategist working with a [industry] business in [city, state].

Context: My business: [what I sell, service area, price range, what makes people choose me]. My three closest competitors: [names].

Task: Write the 12 questions a customer near me would actually type or speak into an AI assistant when they need what I sell. Cover four kinds: urgent (someone with a problem right now), comparison (best or top-rated near me), price (what does this cost around here), and trust (who is reliable, licensed, honest). Use ordinary language, the way a person with no industry knowledge would ask. Then build me a blank scoring sheet: one row per question, with columns for ChatGPT, Google AI Mode, Gemini, and Perplexity, where I mark Named, Mentioned, or Absent, plus a column for which business got named instead.

Format: The 12 questions as a numbered list, then the blank scoring table.

Constraints: Invent nothing about my business. Keep each question under 15 words.

When the sheet is filled in, paste it back with this: here is my completed scoring sheet and the businesses that got named instead of me. Give me the three fixes most likely to change these results in 60 days, ranked by effort, and tell me which one I should do first.

What it's worth. Thirty minutes to find out whether the fastest-growing way customers pick a local business currently says your name. If it does not, the fixes are review recency and profile hygiene, which cost time rather than money.