---
name: review-health-assessment
description: Assess review count, recency, rating distribution, and fake-review risk from visible page data.
---

# Review Health Assessment

## Use this when

review health, review velocity, fake reviews, review response rate.

## Process

From visible page data and schema, assess: total review count against a rough sufficiency threshold of 10, star rating, `aggregateRating` fields (ratingValue, reviewCount, bestRating), any visible recency signal, and owner-response patterns where shown. Review recency matters more than total count: a healthy cadence has no 3+ week gap in new reviews; a long gap is a risk signal worth flagging even with a high total count.

Flag possible review manipulation when two or more of these co-occur: uniform timing (multiple reviews the same day), an unnatural spike in 5-star reviews with no corresponding marketing event, near-identical review text across entries, or reviewer profiles with minimal history. Never assert manipulation from a single signal alone, and always frame it as "worth investigating," not a confirmed finding, since this team can't independently verify reviewer identity.

Flag any pattern that looks like pre-screening customers before directing them to a public review platform ("gating") as a policy compliance issue, not just a quality one — it's prohibited by major platforms' terms and, in some jurisdictions, by consumer-protection law.

For industries with disclosure constraints (healthcare cannot confirm or deny a reviewer is a patient in a public response; legal has attorney-client privilege considerations), note the constraint rather than recommending a generic response template.

Report: review count and rating summary, recency assessment, response-rate estimate, any manipulation risk flags with the specific co-occurring signals cited, and a review-generation strategy recommendation sized to the gap found.