Why AI Assistants Recommend Some Contractors

Homeowners in Albany, Syracuse, and Westchester are no longer starting every search on Google. They are opening ChatGPT and typing "who is the best plumber near me for a burst pipe" and then calling whichever two or three companies the assistant names. According to 2026 industry research, AI usage for local search jumped from roughly 6% in 2025 to about 45% in 2026. That is not a trend line anymore. That is nearly half your market. Here is the uncomfortable part. Only about 1.2% of local businesses currently get recommended by AI assistants at all. If you are a trades business owner and you have never seen your company name come out of ChatGPT, you are in the overwhelming majority. This article is not about what AEO and GEO mean in theory. It is about the specific 2026 thresholds that decide whether you get named, and a checklist you can run on your own business in the next ten minutes.

Why Only 1.2% of Local Businesses Get Recommended

AI assistants do not simply recommend whoever ranks first on Google. They lean on trusted third-party sources and only name a business when several of those sources agree on it. Current benchmarks show only about 1.2% of local businesses clear that bar, which means the filter is not ranking. It is corroboration.

Think about the difference in what the two systems are doing. Google is comfortable handing you ten blue links and letting you sort it out. An AI assistant has to commit. When it names one plumber instead of ten, it is putting its own credibility behind that answer, so it holds back until the signals are strong enough that the recommendation is defensible. Thin, unverified, or inconsistent businesses get skipped rather than risked. That is why a contractor who ranks third in the Google map pack can be completely invisible in ChatGPT while a competitor who ranks lower gets named every time. The competitor is not winning on rank. They are winning because their review volume, rating band, and listing data all line up across multiple sources the assistant already trusts.

The Review Count Threshold: Roughly 150 Per Location

2026 industry research points to roughly 150 reviews per location as the level below which AI assistants rarely name a business. Below that line you can have excellent reviews and still be invisible, because the assistant treats a small sample as insufficient evidence rather than as weak evidence.

The per-location detail matters more than most owners expect. If you run three branches across the Capital Region and you have 200 reviews total, you do not have a business that clears the threshold. You have three locations sitting well under it. Each address is evaluated on its own record, so a strong headquarters does not carry a satellite office with eleven reviews and a half-finished profile. The practical takeaway is that review generation is no longer a nice-to-have reputation exercise. It is the gate. Most trades businesses collect reviews passively, which produces a handful a year. Getting from 40 to 150 requires an actual system: an automated request sent by text the same day the job closes, a second nudge a few days later, and a link that takes one tap.

The Rating Band: Why 4.3 Beats a Perfect 5.0

Current benchmarks put the businesses AI assistants recommend in a 4.2 to 4.7 average star range, with ChatGPT's picks clustering around 4.3. The counterintuitive result is that a perfect 5.0 built on a small number of reviews performs worse than a 4.3 built on many. Volume and believability beat perfection.

This makes sense once you look at it the way the model does. A 5.0 with nineteen reviews is statistically indistinguishable from a business that asked only its happiest customers, or from a profile that has been gamed. A 4.3 with four hundred reviews is what a real company with real crews and real weather delays actually looks like over several years. One of those patterns is easy to fake and the other is not. So stop treating the occasional four-star or three-star review as a crisis to be scrubbed. It is part of what makes the rest of your record credible. What you should worry about is the opposite failure: sliding below roughly 4.2, where the assistant starts reading the average as a genuine quality signal rather than as normal variance. The target is a wide, active, slightly imperfect record, not a spotless thin one.

AI Assistants Read Bing Places, Apple Maps, and Yelp Too

Bing Places feeds ChatGPT directly. Apple Maps, Yelp, and Google Business Profile also feed AI assistants. Most New York contractors have spent years maintaining exactly one of those four listings, which means the assistant checking for corroboration finds one source that knows them and three that do not.

Bing Places is the single biggest blind spot in the trades. Almost nobody claims it, because almost nobody searches on Bing, so it never felt worth the twenty minutes. That calculation changed the moment Bing became a direct pipe into the assistant nearly half of local searchers are now using. An unclaimed or auto-generated Bing listing with a stale phone number is an active liability, not a neutral absence. Consistency across those four sources matters as much as presence. If your name, address, and phone number differ even slightly between platforms, with a suite number on one and not another, or an old tracking number lingering on Yelp, the assistant sees conflicting records about a business it was already uncertain about. Conflicting data is a reason to skip you and name the competitor whose details match everywhere.

The 10-Minute Audit You Can Run on Your Own Business Today

You do not need a tool or an agency to find out whether you have a visibility problem. Open ChatGPT, search your own category and city the way a homeowner would, then check your review count, your star average, your Bing Places claim status, and your listing consistency. Ten minutes gives you a clear yes or no.

Step one: ask ChatGPT "who are the best [your trade] in [your city]" and then ask two follow-ups the way a real customer would, such as "which one is best for emergency work" and "who has the best reviews." Write down every company named. Step two: count your reviews on your primary location profile. If you are under 150, that is almost certainly your ceiling and nothing else on this list will matter until you fix it. Step three: check your star average against the 4.2 to 4.7 band and note which side of it you are on.

Step four: go to Bing Places and search your business name. Confirm whether the listing exists, whether it is claimed by you, and whether the phone number and hours are current. Step five: open Apple Maps, Yelp, and your Google Business Profile side by side and compare the name, address, and phone number character for character. Note every mismatch, including abbreviations, suite numbers, and old numbers. Finally, check whether your most recent review is from this month or from last spring, and whether you have replied to any of the last ten.

What to Fix First

Fix in this order: claim and correct Bing Places, since it feeds ChatGPT directly and takes under an hour. Then make your name, address, and phone number identical across all four sources. Then build review volume toward 150 per location with an automated same-day request. Everything else is secondary.

The reason for that order is time to effect. Listing claims and data corrections are one-afternoon jobs that remove hard blockers immediately. Review volume is the long project, which is exactly why it should start today rather than after the other work is finished. Review freshness and responding to reviews both matter as well, and they are the cheapest habits on this list: a steady trickle of recent reviews signals an operating business, and replies signal one that is paying attention. A profile whose newest review is fourteen months old reads as dormant no matter how good the average is.

Frequently Asked Questions (FAQ)

Q: How do I get recommended by ChatGPT as a local business? A: Clear the thresholds AI assistants use. 2026 benchmarks point to roughly 150 or more reviews per location, an average in the 4.2 to 4.7 range, a claimed and accurate Bing Places listing since it feeds ChatGPT directly, and identical business details across Bing, Apple Maps, Yelp, and Google Business Profile.

Q: I rank first on Google. Why does ChatGPT never mention me? A: Because AI assistants do not simply recommend whoever ranks first on Google. They lean on trusted third-party sources and look for agreement among them. If your Bing listing is unclaimed or your review count sits below the threshold, high Google rank alone will not get you named.

Q: Is a 5.0 star rating hurting my chances? A: If it comes with few reviews, yes. Current benchmarks show ChatGPT's local picks clustering around 4.3, and a perfect 5.0 on a small sample performs worse than a 4.3 with high volume. Focus on getting more reviews rather than protecting a spotless average.

Q: How long does it take to reach 150 reviews? A: That depends entirely on your job volume and whether you ask systematically. Do the arithmetic on your own business: if you complete forty jobs a month and a third of customers leave a review when prompted, you clear 150 inside a year. Without a system that asks after every job, collection is passive and the count barely moves.