We gated our best writing to grow. Google saw 144 words.
26 July 2026 · 5 min read
A while ago we made a decision that sounded obviously correct: put the newer guides behind a short teaser, so people who want to read them create an account. The older ones stay open for search engines. Best of both.
This week we measured it, and it was one of the worst decisions on the site.
What a search engine actually sees
A crawler is a logged-out visitor. It has no account and never will. So when we asked what Googlebot received from a gated guide, the answer was 144 words — a title, an intro, and a wall.
An open guide, same page, same code: 875 words, with headings, a full argument and a FAQ.
Six of our fifteen guides were in the first category. They sat in the sitemap. They got crawled. They could never rank, because there was nothing on them to rank — and worse, they contributed thin-content signal to every other page on the domain.
The arithmetic we skipped
The gate was supposed to trade a little traffic for a lot of sign-ups. Here's the part we never wrote down:
- You cannot convert traffic you do not receive. A gate converts a percentage of arrivals. If the gate is the reason nobody arrives, the percentage is applied to zero.
- The people it did stop were people who had already found us, already clicked, and were already reading. That is the most qualified audience a piece of writing ever has, and we showed them a wall.
- Meanwhile the end-of-article call to action — the thing you reach after the writing has actually earned something — was only ever shown to people who were already members.
What we changed
Every guide is open. All nineteen of them, no account, no teaser, no wall. The invitation sits at the end where it belongs, after we've been useful rather than before.
We also added three real questions and answers to each one, visible on the page — not markup describing content a reader can't see, which is both against the rules and the kind of trick that eventually gets you removed from results entirely.
The general shape of the mistake
The gate wasn't stupid. It was a reasonable idea evaluated against the wrong number.
We were measuring conversion rate, which the gate improves. We were not measuring total conversions, which the gate destroyed by shrinking the pool it operated on. Both are real metrics. Only one of them is the business.
That failure mode is everywhere once you look for it: a filter that improves the quality of what passes while quietly collapsing the amount that arrives. Spam rules that block real mail. Interview loops with a great signal-to-noise ratio and no hires. Requirements that make a form more useful and stop it being filled in.
The test we now try to apply: what does this number look like if the thing it measures is a fraction of something I just made smaller?
We were pleased with the gate for months. Nobody complained, because the people it cost us never knew we existed.
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.
