Auto-reply to good Google reviews, escalate the bad ones

Every weekday at 8am, triage the new Google Business Profile reviews collected in Metricool and handle each one based on its star rating. Metricool does not push review events to us, so run this on a schedule with a lookback window and dedupe carefully rather than expecting one run per review.

Start by listing the brands on my Metricool account with Get Admin Simple Profiles, so the run covers every location I manage rather than just one. For each brand, pull the recent reviews with Metricool's Get Stats Gmb Review, which returns the GMB review list with metrics for a period. Use a lookback window of the last 3 days by default, which gives enough overlap that a missed or failed run still gets caught on the next one. Use Get Stats Gmb Reviewbyid if you need the full detail of a single review that came back thin in the list.

Before doing anything with a review, dedupe. Skip any review that already carries a reply, since that means either I answered it in the Google Business Profile app or an earlier run of this workflow already handled it. Also skip any review whose id is already present in the Notion log described below. Deduping is the most important part of this workflow: the lookback window deliberately overlaps, so without these checks the same review would get answered again on every run.

For each genuinely new review, read what the customer actually wrote and draft a short reply in my brand voice. The reply should be under about 60 words, thank the person by name where a name is available, respond to the specific thing they mentioned rather than using a template, and avoid marketing language, discount offers and anything that sounds automated. If the review is a star rating with no written text, keep the reply to a single warm sentence.

Reviews rated four or five stars get their reply posted straight back to Google through Metricool with Stats Gmb Review Reply. Reviews rated three stars or below are never answered automatically under any circumstances. Instead, send them to my Slack channel using Slack Send a Message, and include the full review text, the reviewer name, the star rating, the location or brand the review belongs to, and the suggested reply, formatted so a person can read it quickly and either approve it as written or rewrite it before posting the response themselves.

The star cutoff is the key business rule here and should be an obvious, clearly labelled setting near the top of the workflow, defaulting to auto-replying at four stars and above. Someone should be able to raise it to five, or set it above five so that everything goes to Slack for approval, without touching any other logic.

Log every review the run touched to my Notion database with Create a Page, whether it was auto-replied or escalated. Each entry should capture the review id, the reviewer name, the star rating, the location or brand, the date of the review, the review text, the action taken (replied automatically or sent to Slack for approval), and the reply text that was posted or suggested. This log doubles as the dedupe record and as a running history of reputation activity across all locations.

Be resilient about failures. If posting a reply to Metricool fails for one review, still log it to Notion with the failure noted, escalate it to Slack, and carry on with the remaining reviews rather than aborting the run. If there are no new reviews, finish quietly without posting anything to Slack.