A month of Google reviews for a Bandra café

The scope

For thirty days we logged every Google review of one Bandra café — 61 reviews — coded each by theme and rating, timed every owner response, and compared notes against the footfall counts the owner shared with us.

The pattern

Reviews left in the month's final two weeks, after the owner began replying same-day, averaged 4.6 stars against the café's lifetime 4.2. Reviewers who received a reply naming a specific dish they mentioned were roughly three times as likely to review again after a later visit.

Why it happens

A reply is not really for the reviewer. It is read by the next fifty people deciding whether the owner is awake. A specific reply — one that proves a human read the review — changes what the next reviewer writes before they've tasted anything.

The cost

Estimate: replying took the owner about six minutes a day. His own rough maths: one returning table a week covers the cost of whoever does the replying. Everything beyond that is margin.

The exception

His reply to a two-star review — an apology, what had changed in the kitchen that week, and an invitation to come back — was screenshotted and shared approvingly by the reviewer herself. The complaint became the best marketing of the month.

What we’d do

Reply to every review the same day, and name one specific thing from each review — the dish, the table, the waiter's name. No templates; a template is worse than silence. Six minutes a day, before the dinner rush.