A 0.1★ drop costs more than you think.

The economics literature is unusually consistent here: on Google, a 0.1★ drop on a mid-sized independent restaurant tends to cost between 3% and 7% of monthly bookings in the following ninety days. The mechanism isn't guests boycotting you at 4.2 — it's the map ranking. A tenth of a star can nudge you two positions down on "restaurants near me" and the visits you lose are the ones that would have been walk-ins.

The 48-hour reversibility window

Star averages on Google are recalculated slowly. There is a window — practically, about two days — where a rating dip is still soft enough that a targeted reply pass can pull the average back before the algorithm hardens it. Miss that window and you're waiting for volume to dilute the bad review, which for most independent venues means weeks.

The window matters because it's actionable. You don't need to chase every review. You need to notice the 1★ or 2★ from Saturday night by Monday morning, reply to it in the reviewer's language, and — if there's a real complaint underneath — fix the thing before Friday. That sequence is what the venues that recover from dips have in common, and it's what venues that don't recover almost universally miss.

What's actually happening at 4.2

Below about 4.3 on Google, two things change at once. First, filters. Guests who filter for "4.5 and above" — a growing habit — stop seeing you. Second, the confidence heuristic in the map ranking algorithm becomes less charitable. Both effects compound. The venue that spends two months at 4.2 doesn't just lose the filtered guests; it loses the visibility that would have generated the reviews to climb back.

This is why operators who track ratings weekly, not monthly, catch the problem before it becomes a quarter.

Three signals worth watching

• Weekly rolling average — not lifetime. Lifetime hides everything.
• Reply lag on the last five negative reviews. Anything over 48 hours is a warning.
• Category share of negative reviews. If service complaints go from 20% to 45% in two weeks, that's a shift-level problem, not a rating problem.