Data-backed headlines and how to write them

A data-backed headline puts a real number from your own work or research in the title, such as "How We Increased Landing Page Conversion by 16%". It tells the reader the post contains evidence, not opinion, and it gives them a specific reason to click. It works when you have a genuine result to report and can show the method inside the post. It fails when the number is inflated, vague or unrelated to the article.

How to do it

1. Find the number first. Look in the article for the most concrete result: a percentage change, a count, a time saved or a sample size. If the post has no real number, rewrite the post or use a different headline style.
2. State what the number measures. "Conversions up 16%" says little. "Landing page conversions up 16% after we cut the form to three fields" says what changed and for whom.
3. Write three or four versions. One with the number, one without it, one that leads with the method, one that leads with the outcome. Buffer's blog team wrote five to ten headlines per post, picked their favorites and tested three of them.
4. Test them. Buffer posted headline variations to its Twitter account close together in time, tracked the results in Buffer analytics, and used the best performer as the post headline. You can do the same with a few posts to your social accounts, an email subject line test or a newsletter A/B test.
5. Back the claim in the first screen. Put the number, the date and the sample or method in the opening paragraph, then link to the data.
6. Keep the number fixed. If the result changes, update the headline and the post together.

Examples

• Buffer, "How We Increased Landing Page Conversion by 16%". Buffer's 2014 post puts a measured result in the title and backs it in the body. It reports a 24-hour test with half of visitors sent to a new landing page. The old page converted at 5.23% (762 sign-ups from 14,556 unique visits) and the new one at 6.05% (994 sign-ups), which the author calls about a 16% increase. It is one day of traffic on one site.
• Buffer's headline testing process. In its 2014 article on A/B testing, Buffer described brainstorming five to 10 headlines, testing three on Twitter and keeping the winner. One example it gave was a post that changed from "A Scientific Guide to Hashtags" to "The Research Behind Hashtags", with the article showing a large difference in retweets, favorites, mentions and reach between the two tweets.

When it backfires

• Unsupported claims. In the US, the FTC's substantiation policy says "firms lacking a reasonable basis before an ad is disseminated violate Section 5 of the FTC Act." A number in a headline that promotes a product is a claim, so keep the records behind it.
• Numbers that do not match the post. If the headline says 30% and the post shows 12% in one segment, readers notice, and so will anyone who quotes you.
• A tiny sample. "We tripled signups" from 6 signups to 18 is true but misleading. Say how many.
• Same format every time. A feed of "X% more Y" titles looks like noise. Use a data headline when the post is built around a result.
• Small effects. In a study of 50 headline pairs run by the Engaging News Project with The Huffington Post, the authors wrote that "the difference is modest and many other factors also affect the number of clicks received by each headline." That study compared solutions-oriented headlines, not numbers, but it is a reminder that any headline change moves clicks by a limited amount.

Sources

• Buffer: How to perform an A/B test on headlines, tweets, traffic and more, read Oct 8, 2026
• Buffer: How We Increased Landing Page Conversion by 16%, read Oct 8, 2026
• FTC: Policy statement regarding advertising substantiation, read Oct 8, 2026
• Nieman Lab: Solutions-oriented headlines generate modestly more traffic, read Oct 8, 2026