Skip to main content

Turn each week's reviews into themes, phrases, and flags.

Overview

Reviews arrive one at a time, so nobody reads them in aggregate, and the aggregate is where the signal lives: a complaint theme building for weeks, a rating quietly slipping, a phrase that sells better than anything your team wrote. Every week the agent clusters the new reviews into themes, compares ratings to their trailing average, and pulls the best verbatim lines out ready for pages and ads. Let a few weeks accumulate; trends need history.

Before you start

  • A supported review platform connected: Okendo, Yotpo, Judge.me, Junip, or Stamped
  • Enough review volume that a week has something to say. If not, set it monthly.

The prompt

Paste this into the agent. The bold values are yours to change:

Every [Friday], read the week's new reviews from [Okendo].

Report: how many came in and the average rating vs the last four weeks, the themes in positives and negatives with counts, and [5] verbatim customer phrases that would make good page or ad copy.

Flag any product whose rating dipped more than [0.3 stars], and tell me which page its reviews suggest updating.

What the agent does

  1. Pulls the window's new reviews from your review platform.
  2. Clusters them into positive and negative themes, with counts.
  3. Extracts the verbatim phrases worth reusing, exactly as customers wrote them.
  4. Compares each product's rating to its trailing average and flags real dips.
  5. Delivers the digest on schedule, with page suggestions where reviews point at one.

Try it on your store

The prompt opens in the agent ready to edit, with your store and tools connected.

More use cases

Browse all use cases