Review Response Quality Gate (QA)

Review Response Quality Gate (QA)

1. 01

   Sample 20 published replies per client location before the first month goes live

   Pull a mix of 5-star, 3-star, and 1-star responses. Score each on specificity (names the service or staff member), accuracy (no invented facts about the visit), and tone match to the client's brand voice.

2. 02

   Flag every AI-drafted reply that could apply verbatim to a different reviewer

   Generic phrasing like 'thanks for your feedback' is the tell. Platforms such as GatherUp and Reviewshake generate drafts in seconds, so the review step is where agency judgment gets added, not the drafting step.

3. 03

   Confirm each negative review reply commits to a named next action and an owner

   A reply that apologizes without stating who will follow up and when reads as deflection. Route the commitment to the client contact who can actually keep it before publishing.

4. 04

   Check that no reply discloses customer details the reviewer did not share publicly

   Mentioning an order number, appointment date, or account status in a public thread creates a privacy problem the client will inherit. Strip it and move the specifics to a private channel.

5. 05

   Verify response time against the client's stated SLA, not the platform default

   Log median hours-to-first-reply for the month. A 48-hour target on 1-star reviews is defensible for a dental group and too slow for a restaurant.

6. 06

   Run a second-reader pass on any reply touching legal, safety, or refund claims

   One reviewer alleging injury or a billing dispute can escalate. Have a second person on the account read it before it posts, and note the escalation in the client file.

7. 07

   Record the QA score and the count of rewritten drafts in the monthly client report

   Two numbers per location: percentage of replies passing first read, and number of AI drafts edited. The second number is the honest evidence of agency labor in the retainer.