"Arrived late, box crushed, sizing wrong." One star, read on a phone in bed at 10:30pm, and the right move is to close the app, because anything typed at 10:30pm gets regretted by morning. Behind it sit 40 other October reviews nobody has answered, about 20 of them five stars from people who have never been thanked.
And somewhere in those 40 are the same three complaints, over and over, which nobody has seen laid out side by side because nobody has read all 40 in one sitting.
A review team in Murage does the replies and the list. It drafts a reply to every review in your voice, and it turns the complaints into a short ranked list of what to fix first. You still post each reply yourself, and which fix comes first is your call.
A month of reviews, and a few lines on how you talk
You need Murage with an AI engine (Fuigo, your existing Claude or ChatGPT plan, or an API key) and the reviews themselves, pasted in or exported. If your review platform is in the connected-apps catalog, connect it through Flux Router or your own connected-apps key.
Make a project called something like "October reviews" and drop them in, with star rating and date if you have them. Add a few lines on your voice: how you greet customers, how you apologize, what you never promise. Then ask:
Here are this month's reviews. Draft a reply to each in my voice, then tell me the three complaints that come up most and what to fix first. Post nothing.
Four helpers share the job. Support drafts a reply to every review, Copy keeps the tone yours, and an Analyst groups the complaints and ranks them by how often they appear. The fourth is Quality Gate, which checks that no reply promises something you can't deliver. We would not run a review team without that one. "We'll refund you" or "this will never happen again" is a private thought until you post it, and then it is a public promise with your shop's name on it.
Read them as the reviewer would
Good reviews get a short thanks. A bad one gets a reply that admits the problem, says what you'll check, and gives the customer a way to reach you. Cut anything defensive and anything that reads like a form letter. If a reply gets your tone wrong, say so and the team adjusts the rest.
Every reply waits for you, because replies are public and a public post stops and asks even on Full access. Read, approve, post them yourself.
The fix list
A typical one, from a month like the one above:
- Slow delivery, nine mentions.
- Damaged packaging came up six times, mostly on the larger orders.
- Sizing: four.
Each comes with example quotes so you see the customers' own words. In a month like that, we'd sort delivery and packaging before running another ad.
First of the month
Monthly suits most shops. Ask your Chief of Staff to prepare it on the first of each month; it proposes a routine, and nothing is scheduled until you press Confirm. If you collect reviews in a spreadsheet, point the routine at that file.
The same project can also ask for reviews. The Local SEO and Reviews Desk drafts review-request messages and a simple process for asking, and the people to ask are the happy, quiet ones. In the team library, Review Engine does both, and turns the complaints that keep coming back into a ranked list of fixes.
The review replies use case shows the whole job, and if you also run a support inbox, customer support triage is the same idea pointed at email. Download Murage, paste in this month's reviews, and read what your customers have been telling you for a month. The one-star can wait till morning.