Online reviews and AI recommendations: what helps, and what to avoid
Ask an AI assistant for "a good family dentist near me" and the answer often reads like a summary of reviews: a rating, a number of reviews, and a phrase such as "patients mention gentle care with kids". Assistants that search the web read review sites and the pages that quote them. None of the AI companies publish exactly how much reviews count, so be wary of anyone who claims to know the formula. What is clear is that reviews are one of the main ways an AI learns what customers think of you.
What seems to help
- A steady flow of genuine reviews. Recent reviews show you are still open and still good. A burst of reviews followed by silence looks odd to people and to platforms.
- Reviews where your customers already look. Google is the most common, but many trades and professions have their own sites that customers trust. Focus on the two or three that matter in your field.
- Specific reviews. "They fixed our furnace the same evening" tells a reader, or an AI, far more than "Great service!" You can't write reviews for customers, but you can ask them to mention what you did for them.
- Replies from you. Thoughtful replies show that someone is paying attention, and they add your side when a review is unfair.
- Fixing what reviews keep saying. If several reviews mention long waits or surprise fees, that pattern will show up in AI summaries too. The fix is in the business, not the reviews.
How to ask for reviews
- Ask every customer, not only the happy ones you pick out. Filtering who gets asked (often called review gating) goes against Google's policies.
- Ask at the right moment, just after the job is done or the visit went well, while the experience is fresh.
- Make it one tap. Send a short link that opens your review form directly. Google Business Profile gives you a shareable review link.
- Keep it simple. One request and at most one reminder. Don't pressure anyone.
What to avoid
- Paying or rewarding for reviews. Discounts, gifts or contest entries in exchange for reviews break Google's policies, and those of most review sites.
- Fake reviews, including reviews written by staff, family or a marketing agency. Platforms remove them and can flag your listing. In Canada, the Competition Bureau has treated fake reviews as deceptive marketing, and in the United States the FTC has a rule banning fake reviews.
- Buying "review packages". These are fake reviews by another name.
- Pressuring people to remove negative reviews. Reply and try to resolve the problem instead. Some customers update their review after a good response.
Responding to reviews
- Thank people by name if they used one, and mention something specific from their review.
- For negative reviews, stay calm, acknowledge the problem, and offer a way to continue the conversation offline.
- Protect privacy. Never reveal details about a customer in a public reply. Clinics and other health providers in particular should not confirm that someone is a patient.
- Reply within a few days. Old reviews with no reply look like no one is watching.
What to track
Once a month, note your rating, total number of reviews and how many arrived that month on each platform that matters. Every few months, ask AI tools the same customer questions and see whether the way they describe you has changed.
Related articles
Curious how AI tools describe you right now? The free check records every answer from ChatGPT, Perplexity and Gemini to 10 real customer questions, including what they say about you and who they recommend instead.
Apply for a free check