Why Your Reviews Are Not Helping Your AI Visibility (Five Fixable Reasons)
A firm with 150 Google reviews and no AI citations usually has one of five problems: the reviews are all on one platform, they are old, they say nothing about the practice or the place, they are unanswered, or the profile they sit on is inconsistent with everything else. Each is visible in ten minutes and fixable in a quarter. What the engines actually do with review data, and the fixes in order.
Short answer: AI engines use reviews as evidence about a firm, and they read more than the star count. They read recency, spread across platforms, the text (what matters and where), the firm's responses, and whether the profile the reviews sit on matches the firm's other listings. A firm with a large but old, single-platform, generic, unanswered review base on a profile that disagrees with its Avvo page gets little from it. Five causes, each fixable: concentrate on recency, add Avvo and Yelp, let clients describe the matter and the place, answer everything, and clean up the entity.
What do the engines actually do with reviews?
From the way their answers are built, described in words:
- Gemini quotes review themes directly ("clients describe clear communication during custody matters") and shows Maps ratings and counts. Recent review text feeds the sentence.
- Perplexity cites Yelp and Avvo pages, including their review sections, as sources for the firm list.
- ChatGPT leans on directory profiles where the review count and rating are part of the page it reads.
- Google's AI surfaces inherit the map pack's judgment, where review recency and velocity are strong signals.
None of them counts to 150 and stops.
The five reasons
1. Everything is on Google
A firm with 150 Google reviews, 2 on Avvo and an unclaimed Yelp page looks, to an engine that reads all three, like a firm with a review program rather than a firm with clients. Perplexity in particular cites Yelp and Avvo; it cannot cite what is not there.
Fix: Rotate the ask. Three clients to Google, one to Avvo, one to Yelp, repeat. Within two quarters the spread looks natural because it is.
2. The reviews are old
Ninety of the 150 arrived in 2021 and 2022, when the firm ran a campaign. Six this year. The engines read the dates; the profile reads as past tense.
Fix: Velocity. Milestone asks, four to eight a month, sustained. The ninety-day count is the number to watch, not the total.
3. The reviews say nothing
"Great lawyer, highly recommend." Five stars, no matter, no place, no practice area. An engine building "who handles custody in Burbank" cannot use it.
Fix: Ask at the milestone, when the matter is on the client's mind, and ask in a way that invites description without suggesting words: "If you'd like to leave a review, it helps other families to hear what the process was like." Never draft, never suggest content (Rule 7.1). Clients asked after a hearing describe the hearing.
4. Nobody answers
Unanswered reviews, especially negative ones, read as a firm that does not engage. Responses are also text the engines read: a response that says "thank you for trusting us with your estate plan; our office is here for any follow-up" is a sentence about what the firm does, in the firm's voice, without confirming anything about the reviewer.
Fix: Every review, two business days, within confidentiality rules. A quarter's backlog can be cleared in an afternoon.
5. The profile disagrees with everything else
The reviews sit on a Google profile named "Smith & Lee Law Group" at the old address; the site says "Smith & Lee LLP" at the new one; Avvo lists a partner who left. The engine cannot be sure the reviews belong to the firm it is being asked about, so it hedges or names someone else.
Fix: Entity consistency. One canonical set of facts across the profile, the site, the bar record and the directories. A week's work, then quarterly checks.
How do you diagnose it in ten minutes?
| Check | Where | Red flag |
|---|---|---|
| Reviews per platform | Google, Avvo, Yelp | One platform has over 90% |
| Reviews in the last 90 days | Google profile | Fewer than six |
| Share of reviews that name a matter or a place | Read the last twenty | Under a third |
| Unanswered reviews | All three platforms | Any negative unanswered; more than a handful positive unanswered |
| Name, address, phone across platforms | Profile, site footer, Avvo, bar record | Any difference |
What does not work?
- A review campaign to a past-client list. A burst, filtered, and a Rule 7.1 exposure if the list includes non-clients.
- Asking clients to mention the practice area. Suggesting content is the line; do not cross it.
- Deleting old reviews (not possible) or starting a new profile (a suspension and a lost history).
- Buying anything.
What does it look like fixed?
Reviews arriving steadily on three platforms, most of them describing a matter and a courthouse, all of them answered, on a profile that matches the firm's every other listing. Within a quarter or two, Gemini's sentence about the firm changes, Perplexity's citations include the Yelp and Avvo pages, and the ninety-day count sits above the local pack's third profile. The review signals service runs it, the review velocity planner sets the pace, and the free visibility check runs the five checks above for a firm and shows what the engines currently say about it.
Check your firm
See whether ChatGPT, Perplexity, Gemini and Google AI Overviews name your firm for your practice area and city.
Free AI visibility check