I've been auditing contractor websites for a while now. A pattern that comes up constantly: the business with 300 Google reviews gets passed over by ChatGPT, and their competitor with 40 reviews gets recommended instead.
Contractors are confused. They feel like they're being ignored despite doing everything right. The truth is they've been optimizing for the wrong thing.
Here's what's actually happening.
ChatGPT Doesn't Read Google Reviews
This is the core misunderstanding. When a homeowner asks ChatGPT "who's the best plumber in Akron, Ohio," ChatGPT is not polling Google reviews. It's not looking at your star rating. It doesn't have access to your GBP profile in the way that matters.
What ChatGPT actually does in web search mode is run a live search, pull content from websites and directories, synthesize an answer from what it finds, and recommend whoever shows up in those results with the most specific, credible content.
Your 300 Google reviews live on Google's platform. They're not indexed as crawlable content that AI tools can synthesize from. So as far as ChatGPT is concerned, those reviews might as well not exist.
What AI Tools Actually Pull From
When ChatGPT or Perplexity recommends a local contractor, the content comes from:
Your service pages. Not your homepage. Not your "about" page. Your actual service pages — the ones that say "AC repair in Akron, Ohio" or "kitchen remodel in Copley" in the H1, describe the service in 300+ words, and explain your process, pricing range, and what a typical job looks like.
Directory listings. Yelp especially. Perplexity pulls heavily from Yelp. ChatGPT's web search surfaces Angi, HomeAdvisor, and Yelp results. If your competitor has 40 Google reviews but a detailed, keyword-rich Yelp profile with photos and service descriptions, they're in the game. You're not.
FAQ content. When someone asks ChatGPT "how much does it cost to replace a water heater in Ohio," it looks for pages that answer that question directly. Not pages that say "call for a free quote." Pages that say "A standard 40-gallon water heater replacement in the Akron area typically runs $800 to $1,400 including parts and labor."
Structured data. LocalBusiness schema, Service schema, FAQ schema. AI tools use these to categorize what you do, where you serve, and what questions you answer. A site with proper schema gets parsed cleanly. A site without it requires the AI to guess — and it usually guesses wrong or skips you entirely.
The Specific Content Your Competitor Has That You Don't
Let me make this concrete. Your competitor with 40 Google reviews and strong AI visibility probably has:
A service page for every major service they offer, each targeting a specific city. Not "Plumbing Services" — "Emergency Plumber in Akron, Ohio." Not "HVAC" — "AC Installation in Copley, OH" and "Furnace Repair in Fairlawn, OH" as separate pages.
A cost and process FAQ on every service page. Something like: "How long does a bathroom remodel take? What's included in a drain cleaning service? Do you charge for estimates?" These aren't just good for humans — they're the exact content AI tools grab when answering homeowner questions.
A Yelp profile with actual service descriptions. Not just a list of services. Paragraphs describing what they do, where they work, and what makes a job go well. Yelp reviews help, but the business description itself is what AI tools index.
Consistent NAP across directories. Name, address, phone — identical across Google, Yelp, Angi, BBB, and your own site. When AI tools cross-reference sources and find inconsistencies, they lower confidence and may skip the recommendation.
Why Your Star Rating Doesn't Help Here
I know this feels backwards. You've spent years earning 300 five-star reviews. It feels like that should count for something with AI.
It does count for human trust when someone finds your GBP through Google Maps. It counts for your local SEO ranking in Google's map pack. Those things matter.
But AI recommendation systems — ChatGPT web search, Perplexity, AI Overviews, voice assistants — are making decisions based on indexed text content, not aggregate ratings from a single platform. They're asking "does this business have credible, specific content that matches what this person is asking?" not "does this business have more stars than the competition?"
A business with 40 reviews and detailed, FAQ-rich service pages in every city they serve looks more credible to an AI than a business with 300 reviews and a homepage that says "we do it all, call us."
The Fix Is Not Complicated
You don't need to throw out everything you've built. You need to add what's missing:
Service pages that describe, not list. For every service you offer in every city you serve, a page that explains the service, your process, what a job typically costs, how long it takes, and what customers should expect. Not a paragraph — a page.
FAQ content that sounds like real questions. "How much does a roof replacement cost in Canton, Ohio?" not "Our services." Contractors who answer the actual questions people type into AI tools are the ones who get cited.
LocalBusiness schema with your full service area. If you serve 8 cities, list all 8 in your schema. AI tools use this to match you to geographic searches.
A Yelp profile that describes your work. Fill out the business description. Add service categories. Upload photos. Yelp's weight in AI recommendations is disproportionate to how much attention most contractors give it.
An llms.txt file. This is new and worth doing. It's a plain text file at /llms.txt on your site that tells AI tools what your business is, what you do, where you work, and what makes you different. ChatGPT and Claude index these. It takes 30 minutes to write and has asymmetric upside — most contractors don't have one, so yours stands out.
How Long This Takes
Content indexed by Google and Bing gets into ChatGPT's web search results within weeks once it's crawled. If you publish solid service pages today and get them indexed, you can start showing up in AI recommendations within a month.
Training data is slower — 6 to 18 months for a model update cycle. But web search mode is where the action is for local contractors right now. Someone asking ChatGPT "who should I call for a burst pipe in Medina" is getting a live search result, not a training data answer. That's the channel you need to be in today.
The contractors who are doing well in AI search right now are not the ones who started 18 months ago. They're the ones who figured this out 3 months ago and built the pages. You can still catch up. The gap is not closed yet.
We do this for contractors at VibeTokens. Run a free brand audit at vibetokens.io/start — you'll see your AI visibility score, what AI tools currently know about your business, and exactly what's missing.
