When an AI assistant decides whether to recommend a local business, review substance matters about six times more than review count. The model reads the reviews. A review that says "replaced our water heater the same day, explained the two options, cleaned up after" tells it what you do, how fast, and how you treat people. A review that says "great service, five stars" tells it almost nothing, and fifty of them tell it almost nothing fifty times.
This is a genuine change from how Google's own local ranking has worked, where volume and rating carried most of the weight. ChatGPT-recommended businesses average 4.3 stars — high, but not perfect — and what separates them from the 4.6-star business that never gets mentioned is usually the text.
What a substantive review contains
- The specific service. Not "they helped us" but "they did our kitchen remodel".
- Something about the process. Timing, communication, how a problem was handled.
- The outcome. What is different now.
- Ideally, the location. A mention of the neighbourhood or city reinforces the geographic signal AI uses to match you to a query.
How to get them without begging
Change the question. "Would you leave us a review?" produces a rating. "What did we do for you, and how did it go?" produces a paragraph. Send it automatically — a text or email within a day of the job, with a link that opens the review form in one tap. Businesses that do this consistently go from single digits to dozens of detailed reviews within a few months, and the reviews read like evidence rather than applause.
Respond to every one
Responding to reviews increases consumer trust by about 45%, and your responses are text the model reads too. A specific, calm reply to a negative review — what happened, what you did about it — often does more for your profile than the review does against it. Silence reads as indifference to humans and to models alike.
Reviews are no longer a score. They are the primary text an AI reads to decide what your business is good at. Write the ask so the answer is worth reading.
Questions
- Do AI assistants read the text of reviews?
- Yes. They analyse the language in reviews to build a reputation profile — which services are mentioned, how outcomes are described, whether the tone is specific or generic. Review substance is about six times more important for AI visibility than the raw count.
- How do I get customers to write detailed reviews?
- Ask a specific question instead of asking for a review. "What was the job, and how did it turn out?" produces detail; "please leave us a review" produces "great service, 5 stars". Automate the ask by text or email right after the job, with a one-tap link.
Sources
Related service
Review and Reputation Management