Results of second CSS reference title A/B test

So, we finished up our second CSS title experiment. In the first experiment, we found that using long titles along the lines of “CSS foo property: Does foo things with the fooness” not only didn’t help improve SEO for the pages, but actually lowered the click count. This was unexpected. The only question in our minds had been “how much improvement will we get?” rather than “will we get any improvement, or will this make things worse?”

So we ran a second experiment in which we used much shorter titles, but still slightly more specific than the current ones. Instead of, for example, simply “background-color” as a title, we used “CSS background-color property”. That’s it.

The results: the same as the first experiment. The final tally showed that pages got slightly fewer clicks with the new titles. This was even more surprising than the first result. So much so that we were sure it was a bug in the tool we used to run the experiment, so @atopal manually calculated the results, and came up with the same answer.

I still find this very hard to believe. Adding detail to the title should not have this result. Anyway, this suggests that we shouldn’t change the titles of these pages, despite any sense that more information would be useful.

All of the pages have been restored to their original titles until we make a decision on what if any changes to make.