We built 71 city pages. A quarter of each one was ours.
26 July 2026 · 5 min read
We have a page for LGBTQ+ dating in each of seventy-one places. Every one was hand-written — a real paragraph about that city, a line about its scene, a question specific to it. We were quite pleased with them.
Then we measured them, which we should have done at page three rather than page seventy-one.
The numbers
We pulled four city pages, stripped the markup, and compared the actual words a reader sees.
- Any two pages shared roughly half their vocabulary.
- Only 21–29% of each page was unique to that city.
- 135 words appeared, identically, on all of them.
- Each page carried about 380 visible words, of which perhaps 82 were genuinely about the place.
That is not seventy-one pages. It is one page with seventy-one variations, and search engines have been able to spot that arrangement for well over a decade.
Where the duplication came from
Not from laziness — from features. Two product questions we thought every visitor deserved answered were appended to every city page. Three benefit rows explaining how the daily set works. A call to action. All reasonable individually; all identical seventy-one times.
The hand-written part, the bit we were proud of, was the smallest thing on the page.
The fix we didn't make
The obvious move is to pad. Write four hundred words about the queer history of every city on the list.
We can't do that honestly. We have not been to most of these places. Anything we produced would be plausible, unverifiable and quietly false, which is exactly the kind of content that reads fine and deserves to rank nowhere.
The fix we made
We asked what is genuinely different between these pages and already true.
Distance. We already store city-centre coordinates, because the app's radius filter runs on them. So each page now says what a radius actually reaches from there — computed, not written.
Brighton: nothing else listed within 50 km; widen to 250 km and you reach five more cities, starting with London. Reykjavik: no other listed city within 250 km at all, which is precisely the situation online-first dating exists for.
No two cities produce the same paragraph unless they genuinely have the same neighbours. It describes how the product behaves rather than how a place feels. And we didn't have to invent a single thing — the facts were already in the database, doing a different job.
We also moved the two duplicated product answers to a single page and linked to it, which removed seventy of the seventy-one copies.
What we'd tell anyone doing programmatic pages
Measure the unique fraction before you scale, not after. It takes ten minutes: fetch two pages, strip the tags, compare the words.
And when a page is thin, the question is not "what could we add" — that road ends in invention. It's "what do we already know that is different here, and true?" Usually the answer is sitting in a database column that was put there for something else entirely.
Know someone this would help?
Dating built on alignment, not volume
The whole LGBTQ+ spectrum, a few genuinely compatible people a day, and matching and messaging free forever.
