Reviews and AI recommendations: why ratings now decide who gets named

Reviews now affect AI recommendations in two ways. AI assistants read reviews and review summaries when deciding which businesses to name, and customers check reviews before trusting what an AI told them. So the number of reviews, their rating, how recent they are and what they actually say all decide whether you get named and chosen.

Key takeaways

• 45% of consumers already use ChatGPT or other AI tools for local recommendations (BrightLocal, 2026).
• Customers expect recent reviews: 74% look for reviews from the last three months.
• Reviews that describe the job ("replaced the e-bike battery in one day") tell an AI what you're good at. "Great!" doesn't.
• Ask every happy customer, at the right moment, with a direct link. Don't pick only the happy ones, and never pay or reward for reviews.
• Reply to every review, calmly and specifically, including the bad ones.

How do AI assistants use reviews?

AI assistants with web search, like ChatGPT, Perplexity, Gemini and Claude, read pages about businesses before they answer, and review platforms are among those pages. Google's own AI goes further: in Google Maps and Search it answers questions about a business using its profile, its reviews and its website, and Google says on its help page that "more reviews and positive ratings can help your business's local ranking". In practice reviews do three jobs for an AI:

• Proof that you exist and are active: recent reviews show the business is open and serving customers.
• Reputation: the rating and the balance of positive and negative comments.
• Facts and words: what services people mention, which area, how fast, how friendly. This is how an AI learns that you're "the place for e-bike repairs in Žirmūnai".

What customers expect (the numbers)

BrightLocal's Local Consumer Review Survey 2026 (1,002 US adults, published February 2026) shows how much review standards have risen:

• 45% use ChatGPT or other AI tools for local recommendations, now the third most common source after Google and Facebook.
• 82% read AI-generated review summaries, and 59% use them as a starting point before checking ratings or full reviews.
• 47% won't use a business with fewer than 20 reviews.
• 74% look for reviews written in the last three months.
• 68% want at least 4.0 stars, and 31% only consider businesses with 4.5 or more.

These are US figures, and Lithuanian habits may differ. The direction is the same, though: customers ask AI, then check the reviews.

Step 1: Ask every customer, at the right moment

Most happy customers never leave a review because nobody asked. Ask right after the job is done, when they're happy: in person, in the receipt email or by SMS. Make it one tap: use the direct review link from your Google Business Profile ("Ask for reviews"), or a QR code at the counter. Ask everyone, not only those you expect to be happy. Filtering out unhappy customers before they reach the review page ("review gating") breaks Google's rules.

Step 2: Help them write a useful review

You can't tell customers what to write, but you can remind them what helps other people: "If you have a minute, a few words about what we did would help other cyclists find us." A review that names the service, the bike and the result is worth more than five stars with no text, to customers and to AI. Never write reviews for customers, and don't post reviews yourself under other names.

Step 3: Reply to every review

• Positive: thank them and mention the job in natural words ("Glad the new battery is working well"). That adds more useful text to your profile.
• Negative: reply calmly, admit what went wrong if it did, and say what you did about it. AI summaries often mention complaints. A calm reply shows you take them seriously.
• Fake or abusive: report it through your profile instead of arguing in public.

Step 4: Don't stop at Google

Different assistants read different sources. Google reviews matter most for Google's AI, but ChatGPT, Perplexity and Copilot also read other sites. Collect reviews where your customers already are: Google, your Facebook page and any well-known review sites for your industry. Keep your name, address and phone the same on all of them, so an AI can tell the reviews are about the same business.

What you must not do

• Don't buy reviews or offer discounts, gifts or prizes for them. Google removes them and can restrict your profile.
• Don't ask only the happy customers (see review gating above).
• Don't post fake reviews of your own business or fake negative reviews of competitors. Since 2022, EU consumer law, applied in Lithuania too, explicitly bans fake reviews and misrepresenting reviews.
• If you show reviews on your own website, don't pick only the best ones and present them as all reviews. EU rules require businesses that show reviews to say whether and how they check that they come from real customers.

How long does it take?

Reviews build up slowly, and that's fine: a steady flow of recent reviews is exactly what customers and AI look for. A small business that asks every customer typically sees a difference within a few months. Keep going; the 3-month window means old reviews count for less over time.

How to tell if it's working

Ask the questions your customers ask ("Who repairs e-bikes in Vilnius?") in ChatGPT, Gemini and Perplexity every few weeks, and note whether you're named and what they say about you. A RecoPulse report does this across four AI assistants, shows what they say about you and which competitors get named instead. Start with the free check.