Max KiriienkoShort read · SEO
I made the text five times longer. Rankings fell.
A small test on one of my content sites — and why the honest conclusion is narrower than the headline.
“Longer pages rank better” is the advice everyone has heard. On one of my small content sites I had about eighty reference pages built from the same template, roughly 300 words each, sitting deep in the results. Rewriting all of them would cost money and time. So before paying, I tested the idea on three.
My hypothesis was the opposite of the advice: the pages sit low because the domain is young and has no trust yet, not because the text is short.
The setup
On August 7 I picked three pages that already ranked reasonably well and expanded each from about 300 to about 1,500 words. The other seventy-odd pages of the same template stayed untouched as a control group. Before starting I wrote down the check date — three weeks later — and the decision rule. If the test pages moved up more than the control, we expand the rest. If not, we leave them alone and save the money.
| Page | Words | Avg. position before | Aug 28 – Sep 24 |
|---|---|---|---|
| Page A | 312 → 1,517 | 18.8 | 65 |
| Page B | 321 → 1,589 | 17.2 | 78 |
| Page C | 299 → 1,545 | 14.4 | 80 |
| Control pages | not touched | 59 | 78 |
The result
All three fell hard: from 18.8 to 65, from 17.2 to 78, and from 14.4 to 80. It was already clear at the planned check: between August 8 and 28 the three pages averaged positions 56 to 69.
Average position before and after the test
Page A 18.8 to 65, page B 17.2 to 78, page C 14.4 to 80, untouched control pages 59 to 78.
Now the honest part. Over the same period, the untouched control pages also slid — from around position 59 to 78 — and most of them dropped out of the results: before the test almost all of them had impressions, after it fewer than a third. The whole site was going down. And the test pages had only 4 to 16 impressions before the test: the site had been in Google’s results for less than a week. A tiny sample. So I can’t claim that more text hurt.
One thing the averages hide: the test pages didn’t vanish. Over the next seven weeks each was shown for 59 to 89 different queries. But every one of those queries sat below position 40, and the three pages got one click between them.
What I can claim is narrower and more useful: making the text five times longer didn’t help. That was the question that mattered, because “yes” meant paying to rewrite the rest of the template. We didn’t.
So what’s the best content length for SEO?
I don’t have a number, and this test is why. One more data point from the same niche at the time: the sites that grew were tools and forums. Reference-style sites had lost 30–80% of their Ahrefs-estimated traffic in a year. Word count wasn’t the lever. The type of page was.
How to test content length on your own site
- Write the check date and the decision rule before the test. Otherwise you’ll read the result the way you wanted it.
- Always keep a control group. Without it, this note would have said “long text kills rankings” — and it would have been wrong.
- Check that you have enough impressions to conclude anything. Four impressions before the test is an anecdote, not a baseline.
- The best outcome of a test is often a cancelled expense, not a win.