How to Get More Reviews, and Why AI Reads Them Differently

Three weeks ago, Yelp signed a licensing deal with OpenAI that pipes its reviews, ratings and photos straight into ChatGPT. If you have been ignoring Yelp for a decade, that changed on July 23, 2026. And if you think 40 reviews is plenty, the number is not the thing you should be worrying about anyway.

Key takeaways

Ask for reviews that describe the specific job, not the star rating. A review that says "replaced our water heater the same day we called" is a sentence ChatGPT can quote back. Five stars with no text is not.
Yelp signed a licensing deal with OpenAI on July 23, 2026, adding reviews, ratings and photos to ChatGPT. Reservations and waitlist followed on August 10, 2026. A Yelp page you have ignored for years is now feeding an engine your customers use.
In our own research on 9,454 question-by-business checks, the business was not among the sources the answer used in 84.3% of cases. Google presence for the exact question was the strongest predictor of clearing that stage (odds ratio 12.74). This is an association, not a promise.
Each engine looks somewhere different. ChatGPT leans on Foursquare Places and now Yelp. Gemini is grounded in Google Maps. Perplexity leans on Yelp and industry directories. Claude leans on Brave Search.
Word "citation" means two things in this post. In local SEO it means a directory listing of your name, address and phone. In AI search it means an engine naming your business in its answer. When both come up, we say which one.
The "you need 150+ reviews before AI cites you" number is repeated everywhere. It has no published methodology. Ignore it.

The fastest way to get more Google reviews

Ask right after the job, by text, with a direct link to your Google review page, and prompt the customer to describe what you did. That is the whole tactic. What matters more than the count is what those reviews actually say.

Here is the ask script we would give a plumbing company owner. Send it from the tech's phone the same afternoon:

"Hey, thanks for having us out today. If we did right by you, would you leave a quick review? Google link here. If you can mention what we fixed (water heater, main line, whatever it was), that helps other neighbors find us. Takes 60 seconds."

Two things are doing the work in that script. It goes out same day, when the memory is fresh. And it asks for the job in words, not the star count. That second part is why a plumber with 40 specific reviews often beats a competitor with 200 generic ones in AI answers.

The old advice was volume. The new reality is that an engine can only quote a review if the review has words in it worth quoting.

Why the Yelp and OpenAI deal changes what you should do this month

On July 23, 2026, Yelp announced a licensing agreement that brings its reviews, ratings and photos into ChatGPT. On August 10, 2026, reservations and waitlist joined. If you are a plumber, a roofer, a dentist, a med spa owner, this matters more than any other local news in the last two years.

Two reasons. First, ChatGPT is the AI tool consumers actually use to find local businesses. In BrightLocal's Local Consumer Review Survey published February 11, 2026 (n=1,002 US adults), the share of consumers using AI to find local businesses jumped from 6% to 45% year over year. ChatGPT was 31%. Google's AI Mode was 23%.

Second, Yelp was already the most-quoted local review source in AI answers before the deal. A Foundation and AirOps analysis of 28 million AI responses in Q4 2025 found Yelp received 512,680 citations, 3.4 times the next platform. Over half came from "near me" queries. Now Yelp reviews flow directly into the tool 31% of local searchers are using.

If you have not looked at your Yelp page since 2018, look at it this week. Check that your hours, phone number and service list are current. Ask two or three recent customers to post there too, using the same job-specific script. You do not need to become a Yelp power user. You need the listing to exist, be correct, and have real sentences about real jobs on it.

What AI actually pulls a review from (and it is not the same as Google)

Each AI engine looks in a different place for local businesses, and this is where most local advice goes wrong. There is no single "AI." Where you show up depends on which engine, which platform it trusts, and whether the sentence a customer wrote answers the question someone asked.

Here is the short version:

Engine	Primary sources for local
ChatGPT	Foursquare Places (partnership announced December 16, 2024, covering 100M+ places in 200+ countries) and now Yelp reviews, ratings and photos
Google AI Mode / Gemini	Google Maps, Google Business Profile, Google reviews
Perplexity	Yelp, TripAdvisor, industry directories
Claude	Brave Search results, which pull from a mix of directory and web sources

Notice what is not on that list. Your website. Your blog. Your service pages. Those may still matter, but they are not where these engines look FIRST for a local trade query. They look at listings and reviews. If your Google Business Profile is thin, your Yelp is abandoned and your Bing Places is empty, none of the four engines have much to work with.

Reviews on Google are still the biggest lever for Google's AI Mode. Reviews on Yelp are the biggest lever for ChatGPT. If you only ask for one, ask for Google. If you can spread the ask, ask half your customers to leave a Google review and half to leave a Yelp one. Do not send the same customer to both. That is how you get zero.

For the deeper breakdown of which engine draws from where, we wrote a whole piece on where each AI actually gets local business names.

Why what the review says matters more than how many you have

An AI engine answering "who's a good emergency plumber in [town]?" has to write a sentence. It builds that sentence out of the words available to it. Star ratings are not sentences. "Great service!!" is not a sentence an engine can use. "Came out at 10pm on a Sunday and replaced a burst pipe under the sink before it flooded the basement" is a sentence an engine can use.

This is why the 40-reviews plumber with specific stories often shows up in ChatGPT answers when the 200-reviews competitor with "5 stars, thanks" reviews does not. The quotable material is on your page, not theirs.

Categories of things to prompt customers to mention in reviews:

The specific job (water heater install, drain cleaning, slab leak, sewer line, sump pump)
The timing (same day, weekend, after hours, holiday)
The location within town (neighborhood, zip, "off Main St")
Any specific problem your customer thought was going to be worse than it was
Any specific brand or model you worked on (if you specialize)

You do not script the review. You suggest the topic. The customer's own words are what matters, because engines detect and reject anything that reads like copy-paste.

One more thing. In local SEO, "citation" means a directory listing of your name, address and phone. That is not what we mean here. In AI search, a citation means an engine naming your business in its answer, and often quoting a source. Both matter for a local business, and they matter for different reasons. Consistent name, address and phone across directories helps engines match up "the plumbing company in [town]" with "the one with these reviews." Real, quotable reviews help engines actually name you. If you want the deeper split, our piece on local citations vs AI citations breaks it down.

What our own research says about Google presence

We ran 9,454 checks that paired a specific local question with a specific local business, then looked at which businesses appeared among the sources an AI answer used. In 84.3% of those checks, the business was not among them. About 94% of the observed failure happens before an engine picks its final source list. If you are not in the sources it considers, you cannot be quoted.

The strongest predictor of clearing that stage was Google organic presence for the exact question. The odds ratio was 12.74 (McFadden R-squared 0.24). We tested nine technical page signals. None survived a Bonferroni correction. In plain English: whether your page had certain hidden tags or a particular structure did not predict much. Whether Google showed you for the question predicted a lot.

Two hard caveats, and this is where most competitor content lies to you:

This is an association. We can see that businesses showing up in Google for a question are more likely to appear in AI answer sources for that question. We cannot tell you fixing one causes the other. Anyone who tells you they can prove that is guessing.
We do not observe what an engine retrieved and rejected. We observe its returned sources. So we say "not among the sources this answer used." We do not say "AI didn't find you." That would be a claim about retrieval, and we cannot see retrieval.

The practical read is boring and unglamorous. Reviews with specific words. A Google Business Profile that is current. A Yelp listing that exists. A page on your site for each real service in each real town you cover. If you want a fuller picture of how local businesses fall through this funnel, we wrote up where local businesses actually drop out of the AI answer.

A 30-day plan for a plumbing company with 40 reviews

Say you run a plumbing company in a town of 60,000 people. You have 40 Google reviews averaging 4.7 stars. Your Yelp has 6 reviews from 2019. Your Bing Places listing was auto-created and you have never touched it. Here is what to do, in the order that matters:

Week 1. Text every customer from the last 30 days. Same-day where possible. Use the ask script above. Send them to Google. Do not send them to Yelp yet. Aim for 8 new Google reviews. Ask each one to mention the specific job.

Week 2. Claim and fix your Yelp listing. Update hours, phone, service list. Add 5 photos of actual jobs (not stock, not logos). Send your next 10 customers to Yelp instead of Google. Same script, different link. You want the deal-flow going both directions.

Week 3. Claim your Bing Places listing. This takes 20 minutes. Bing Places is where ChatGPT falls back when Foursquare and Yelp do not have data. Add the same hours, phone, and service list as your Google Business Profile. The three listings must match exactly, character for character on the address.

Week 4. Add a page to your website for each of the top three services (water heater replacement, drain cleaning, emergency repair) crossed with each of the top three towns you cover. That is nine pages. Real content, not junk. This is the one and only place we would send you to write content for a local business, and there is a right way to do it that avoids the trap of making 40 near-identical junk pages. Our piece on service area pages that AI reads covers the guardrails.

At the end of 30 days you will have roughly 60 Google reviews with actual job descriptions, a live Yelp page with 10+ recent reviews, a corrected Bing Places listing, and 9 real service-area pages. You still cannot promise yourself an AI citation. But you have moved from a state where none of the four engines have much to say about you to a state where three of them have quotable material.

If you want to see where you stand right now before doing any of this, the free AI Visibility Check runs the same set of local questions through ChatGPT and shows you what each engine currently says. Ninety seconds. No email needed.

Frequently asked questions

How many Google reviews do I need before AI will cite me?

There is no threshold anyone has proven. You will see numbers like 150+ repeated online, and none of them come with published methodology. What matters more than count is content. A business with 40 reviews that describe specific jobs, in specific parts of town, is often more quotable than a business with 300 reviews that all say "great service." Ask for the story, not the stars.

Should I ask for Google reviews or Yelp reviews?

Ask for both, but not from the same customer. Send roughly half your customers to Google and half to Yelp. Google reviews feed Google's AI Mode and Gemini. Yelp reviews feed ChatGPT (as of the July 23, 2026 licensing deal) and Perplexity. If you only ever ask for one, ask for Google, because Google organic presence is the biggest single predictor we have measured. But do not skip Yelp anymore.

Are fake or incentivized reviews worth the risk?

No. Both Google and Yelp actively remove reviews that look coordinated, and repeat offenders get their listings suppressed entirely. That is a cliff, not a slope. An engine cannot quote you if your listing is not visible in the source it pulls from. The upside of a fake review is a fraction of a star. The downside is being wiped off the map source that ChatGPT or Gemini is looking at.

What if a customer refuses to leave a written review and just gives me 5 stars?

Thank them, ship them the star rating, and move on. Do not push. What you want is a mix: some star-only ratings (they help your average) and enough written reviews with real job language that an engine has something to quote. If one in three customers writes a real sentence, you are doing fine. Trying to force text out of every customer produces the fake-sounding reviews that platforms filter out.

Does what I put on my website matter at all for AI answers about local businesses?

It matters, but not the way it does for a content site. For a trade owner, the two things on your site that matter most are your service-and-town pages (a real page for each service in each town you cover) and your contact info matching what is on your Google Business Profile, Yelp and Bing Places. Everything else, blog posts, "ultimate guides" to plumbing, is the wrong lever. The right lever is listings and reviews. Our local SEO in the AI era piece goes deeper on where the two overlap and where they do not.

What to do this week

Pick one thing from the 30-day plan and do it before Friday. If you had to pick one, it would be texting your last 30 customers with the job-specific ask script. That single move produces the quotable material every engine needs, and it costs you nothing. Then run the free AI Visibility Check so you have a baseline of what each engine says about your business today, and you can see it move over the next few months. The businesses that adjust to the Yelp and OpenAI deal in the first quarter after it launched are the ones ChatGPT will be quoting a year from now. The rest will still be arguing about review count.