The latest research suggests that 29 is the optimal number to use in the title of a listicle. That specific number—29—brings in more traffic, far and away, than any other number. For example, the 29 Life-Changing Quesadillas You Need To Know About are more valuable than 25 or 30 life-changing quesadillas.
Gilad Lotan, chief data scientist at Betaworks, published his findings on Tuesday. Here you can see a screenshot of his data, showing the traffic generated by BuzzFeed listicles based on the number in the title:
Data: Gilad Lotan
After 29, Lotan says, the best results came from nearby odd numbers, such as 27, 31, and 35. He reports finding “a statistically significant difference in the performance of odd-length listicles compared to even ones.” Notice in the graph how the orange bars (representing listicles with an odd number of items) are generally higher than the blue bars (representing the evens). Having 29 life-changing quesadillas is definitely better than 28 or 30.
Prime numbers don’t seem to have a particular advantage over other odd numbers. Such larger numbers as 71, 81, and 91 tended to do well, usually because of outlier individual posts with extremely high traffic. Notice how that “1″ at the end is a common theme.
In summary, three distinct patterns stand out in drawing the biggest audience:
• keeping the overall range of items in the 20s or 30s;
• using odd-numbered totals;
• using “1″ as the final digit in especially long lists
Lotan’s code is available here and shows a more complete description of the data. Cells in red represent more traffic.
Data: Gilad Lotan
As we just saw recently with Facebook’s attempt to manipulate user content (not that that makes Facebook (FB) unusual in the social-media world), Lotan asks the right question about the ethical boundaries of data scientists working to improve business results:
“The tougher question is where do we draw the line? As a data scientist I am tasked with finding techniques to optimize performance, not only for algorithms, but for businesses. Part of the commonly used tactics involves this type of behavioral analysis, comparing datasets based on parameters that may be descriptors of the data itself (such as listicle length) or based on user metadata (your typical user segmentation). By building a recommendation system that gets users to interact with more content than they typically would have and spend more time on my site, am I crossing an ethical boundary? What if I tweak the recommendation system to affect user purchase behavior? Or emotional state?”