Same test, six cities, 14,130 replies. For each one we asked whether it concedes fault, promises compensation, guarantees the future, pushes the customer off the page, or turns on the reviewer. Then we counted by city.
The table
- Liverpool — 14% break a check · 31 cafés · 478 replies read · median 25 words
- Oxford — 13% break a check · 40 cafés · 529 replies read · median 25 words
- Birmingham — 12% break a check · 55 cafés · 1,099 replies read · median 24 words
- London — 10% break a check · 160 cafés · 2,108 replies read · median 28 words
- Nottingham — 10% break a check · 48 cafés · 790 replies read · median 31 words
- Cambridge — 9% break a check · 40 cafés · 535 replies read · median 22 words
- Leeds — 8% break a check · 58 cafés · 1,026 replies read · median 18 words
- Manchester — 8% break a check · 52 cafés · 918 replies read · median 26 words
- Edinburgh — 7% break a check · 45 cafés · 687 replies read · median 18 words
- Brighton — 6% break a check · 46 cafés · 832 replies read · median 26 words
- Bristol — 6% break a check · 57 cafés · 756 replies read · median 18 words
- Newcastle — 6% break a check · 46 cafés · 545 replies read · median 27 words
- Bath — 5% break a check · 34 cafés · 585 replies read · median 17 words
- Belfast — 5% break a check · 32 cafés · 408 replies read · median 20 words
- Sheffield — 5% break a check · 49 cafés · 791 replies read · median 13 words
- York — 5% break a check · 31 cafés · 599 replies read · median 18 words
- Cardiff — 4% break a check · 48 cafés · 682 replies read · median 14 words
- Glasgow — 4% break a check · 58 cafés · 762 replies read · median 17 words
The spread runs from 4 per cent to 14 per cent. That is a real difference and it is not explained by sample size — every city here contributes hundreds of replies.
What we are not claiming
This is not a ranking of cafés, or of coffee, or of service. It measures one narrow thing: what happens in the box underneath a Google review. A city can have superb cafés and careless replies, and several of these do.
Nor is it a ranking of how often cafés reply at all. We only read reviews that already carry a reply, so a city that answers rarely but carefully will look good here. We measured the answer rate properly in one city only — London — and there, 57 per cent of the cafés we checked had replied to none of their recent reviews.
The one pattern that does travel
Reply length barely moves between cities. The medians sit within a few words of each other, and the overall median across the whole corpus is 22 words, with ninety per cent at 73 or fewer. Whatever separates these cities, it is not effort measured in words.
What separates them is what the words do. The cities at the bottom of this table are not writing less; they are writing more apology per sentence.
What each city gets complained about
Across all six, the subject mix is remarkably stable. Coffee dominates everywhere, price is second almost everywhere, and the complaints that provoke the worst replies — staff and waiting — sit in the middle of the pile by volume:
- coffee quality — 4,324 reviews across all six cities
- price and value complaints — 680 reviews across all six cities
- allergy and dietary requirements — 427 reviews across all six cities
- slow service and long waits — 344 reviews across all six cities
- rude staff — 307 reviews across all six cities
- cold food and lukewarm drinks — 212 reviews across all six cities
- dogs, prams and access — 211 reviews across all six cities
This is the useful part of a city comparison. Not that one place is better than another, but that the job is the same job everywhere. A café in Leith and a café in Digbeth are answering the same complaints in the same proportions.
The number that does not vary at all
Repetition. 664 sentences in the corpus appear word for word at two or more separate businesses, and 122 appear at five or more. The most repeated closing line turns up at 68 different cafés — and those cafés are not clustered in one city. They are spread across all six.
Which is the quiet finding underneath this whole table. Whatever separates the best city from the worst, none of them has solved the thing that actually shows on a listing: that the reply could have been written by anyone, anywhere.
The sentences that cross every border
If the cities differ in how carefully they write, they are identical in what they copy. These are the most repeated sentences across the whole corpus, and none of them belongs to one city:
- "hope to see you again soon!" — 68 cafés
- "we hope to see you again soon!" — 54 cafés
- "thank you for your feedback." — 52 cafés
- "thank you so much for your lovely review!" — 47 cafés
- "thank you for your review." — 46 cafés
- "we hope to see you again soon." — 36 cafés
Length, city by city
A reasonable guess would be that the careful cities write more. They do not:
- Liverpool — median 25 words per reply
- Oxford — median 25 words per reply
- Birmingham — median 24 words per reply
- London — median 28 words per reply
- Nottingham — median 31 words per reply
- Cambridge — median 22 words per reply
- Leeds — median 18 words per reply
- Manchester — median 26 words per reply
- Edinburgh — median 18 words per reply
- Brighton — median 26 words per reply
- Bristol — median 18 words per reply
- Newcastle — median 27 words per reply
- Bath — median 17 words per reply
- Belfast — median 20 words per reply
- Sheffield — median 13 words per reply
- York — median 18 words per reply
- Cardiff — median 14 words per reply
- Glasgow — median 17 words per reply
The medians sit within a handful of words of each other. Whatever separates these cities, it is not effort measured in keystrokes. It is what the keystrokes are spent on.
Why we are publishing a table nobody asked for
Because it is the kind of thing that is easy to assert and almost never counted. Everyone has an instinct about which city is friendlier or more direct. This is one narrow, checkable version of that instinct, and it is checkable because we read every reply rather than sampling opinions about them.
If your city is at the wrong end of it, the useful response is not embarrassment. It is that the gap between the top and the bottom here is a few hundred replies, written by a few hundred people, each of whom could write a different sentence tomorrow.
How this was measured
14,130 review-and-reply pairs from 914 cafés across 90 neighbourhoods in 18 cities, read from live Google listings in August 2026. Cafés were found by neighbourhood, not by rating, so this is not a list of the best or the busiest. No café is named.