Read one bad review and you learn about one bad night. Read forty of them with the dates attached and you are holding something closer to an audit — one you did not have to pay for, written by people who were actually in the room.
The difficulty is that reviews arrive one at a time, usually at the worst moment, and each one feels personal. So they get read as insults instead of data, and the pattern sitting inside them never gets seen.
Sort by day and time before you read a word
The single most useful thing you can do with a year of reviews is ignore the text at first and just plot when the complaints happened.
Almost every venue finds the same kind of thing: the negatives cluster. Friday after 20:00. Sunday lunch. The first week of the month. Whatever the cluster is, it is telling you that this is not a quality problem — it is a capacity problem at a specific hour, and the fix is a rota change rather than a new supplier or a new chef.
If your one-stars are evenly spread across every service, that is a different and more serious finding: it points at something structural — a recipe, a policy, a person who works every shift.
The four buckets almost everything falls into
When you group complaints by what actually went wrong, most venues find that nearly all of them land in one of four places:
| Bucket | What it sounds like | Where the fix lives |
|---|---|---|
| Speed | "waited 40 minutes", "forgot our starters" | Rota and pass discipline |
| Temperature | "arrived cold", "lukewarm" | Runner count, pass-to-table distance |
| Attention | "couldn't get anyone's eye", "never checked on us" | Section sizes |
| Money | "charged for water", "service charge wasn't mentioned" | Menu wording and briefing |
That last bucket is the one owners consistently underrate. Complaints about money are rarely about the amount — they are about surprise. A guest who is told about the service charge when they sit down does not write a review about it. A guest who discovers it at the end does.
Watch what disappears, not just what appears
The useful signal is often a complaint that stops. If "cold food" shows up nine times in March and twice in June after you added a second runner, you have a measurement of whether the change worked — and it cost you nothing to collect.
This works in reverse too, and it is the more valuable direction. A complaint type that has been quiet for months and suddenly reappears three times in a fortnight usually means something changed: a new starter on the pass, a supplier substitution, someone senior on holiday. Reviews will tell you about that drift a week or two before your numbers do.
The reviews nobody reads properly
Four-star reviews are the most honest thing you will read all month. A guest who liked you enough to come back but held back a star will usually say exactly why in a single throwaway clause — "lovely, though the room is quite loud" — and because it is not an angry review, it is unguarded. Those clauses are where your next actual improvement is.
Five-star reviews are worth mining for a different reason: they tell you what your regulars think you are for. If eighty per cent of them mention one dish or one member of staff, that is your positioning, written by your customers. Most menus and most websites bury it.
Doing this monthly is the hard part
None of the above is complicated. It is an hour with a spreadsheet, and any owner who did it every month would run a noticeably tighter operation by the end of the year.
Almost nobody does it every month, because the hour lands on a day when three people called in sick. That is the gap OMMU fills: the same pattern read is done continuously in the background and lands as one short email — what shifted this week, what is drifting, and the one thing worth fixing before Friday. No dashboard to log into, no spreadsheet to maintain.
Whether or not you use us for it, do the sort. Even once. The first time an owner plots their complaints by hour, they usually stop mid-way and say some version of "oh — it's Fridays."