Why star ratings aren't enough

A location's average rating moves slowly and says nothing about cause. Two stores with the same 4.1 can have completely different problems: one with slow service, the other with a stock issue. The useful signal is in the review text.

At a few locations, someone can read every review. At hundreds, nobody can, so teams sample, and sampling is where anecdotes start to replace evidence.

Step 1: collect every review in one place

Start with a complete, deduplicated set of reviews for every location, with the fields you'll need later: the location, the date, the rating and the full text.

  • Map each Google Maps listing to the store it belongs to, so results roll up correctly.
  • Keep the review date, not the date you collected it, so trends line up with what happened.
  • Store the full text. Truncated reviews lose the reason, which is usually in the last sentence.

Step 2: classify each review into themes

To compare locations and months, every review needs to land in the same set of categories. A three-level taxonomy works well: a broad theme (Staff), a sub-theme (Helpfulness) and a specific issue (No one available to help).

A single review often covers several things. "Lovely staff, but the queue was long" is one positive and one negative mention, and both should count.

Build the taxonomy from your reviews

Start from what customers actually write about, not an internal org chart. Our guide to building one walks through it.

How to build a review taxonomy

Step 3: decide what to fix first

Volume alone isn't priority. Rank each issue on three things:

  1. Negative share: what fraction of mentions of this topic are negative.
  2. Change: whether that share is rising compared with the previous period.
  3. Sample size: how many reviews sit behind it. Flag small samples rather than acting on them.

Free review snapshot

Want this done for your own reviews?

No commitment. We use your details only to send your snapshot and follow up. Privacy policy

Step 4: keep the evidence attached

A finding like "checkout complaints rose this month" gets acted on when the team can read the reviews behind it. Link every number back to its reviews, so a store manager can see exactly what customers said.

Doing this with Falcon

Falcon does each of these steps for your Google Maps reviews: every review is classified into themes, sub-themes and issues, experience signals show where negative share is rising by topic and month, and every number opens the reviews behind it.