Buzzfeed is full of numbered lists with sensational headlines. Have you ever wondered what number results in the most viewed articles? We certainly have. We asked our coworkers what number they would be most likely to click on and the answer was pretty much unanimous. It amounted to “Any number as long as there are cute baby animals.”

This answer didn’t satisfy us very much, so we decided to look at the data. Fortunately, there’s a view count on the bottom of every Buzzfeed article, so anyone with a little time and curiosity could do it.

First test – Frequency distribution

We compiled a list of posts, their view count, and the date it was posted. The search we used to compile the list was:

site:buzzfeed.com “top 3″ OR “3 of the” OR “3 best”

We grabbed the first page of results on Google for each number from 3 to 35, then sorted them by most views per post. We then selected the top 100 posts and did a quick frequency distribution analysis. 25 appeared to be the best number, but only by one or two posts.

This didn’t give us the confidence we had hoped for, but a picture was starting to emerge. We noticed that once you got up into the large number of views, there were no posts with a number under 10 in the title. With confidence, we could say that the posts with the most views were clustered around 25, ranging from 21 to 27. We could also say that numbered lists ending in 5 or 0 garnered more views than other numbers.

Love exploring the world with data?

So do we. Our data journalism has been cited by The New York Times, The Wall Street Journal, and Fast Company.


Second Test – Changing the search parameters

We went back to Buzzfeed and noticed that the posts that ranked on the first page of Google search weren’t the posts with the most views of all time. The most viewed posts had titles outside of our search parameters, for example, “21 Pictures That Will Restore Your Faith In Humanity.” Our second test involved searching “most viewed on Buzzfeed” and compiling a frequency distribution chart based on the results.

The search resulted in this list:

Looking at the most viewed posts, three had the number 25 in the title! While this gave us more confidence in the number 25, we noticed that each one of the three posts had views under 7 million, for a total under 21 million. The number 21 in the title had two posts over 10 million views, (including the most viewed of all time) for a total number of views of close to 25 million. This gave us the idea for our third test. What number in a blog post title has the highest average number of views?

Third Test – Comparing the average views for each number

Looking at the total views by number in the post title gave us an idea for a third test. We wondered if there was a reliable number for more views, regardless of whether it went viral or not. We took the ten posts we collected for each number in the post title and then found the average number of views.

Based on this data, the average post with 25 in the title far outperformed every other number.

Conclusion

At this point, we feel confident about the number 25 being the top performer in “best of” lists. While we can’t guarantee a million views each time you include 25 in the title, we can say that it consistently performs better than any other number.

Bonus: Top 25 things we learned from looking at all those Buzzfeed posts

Just kidding, there are only four takeaways. We noticed a few common themes among viral posts on Buzzfeed, regardless of the number in the title:

  • Three Buzzfeed contributors had the highest number of viral posts and those three dominated the top ten most viewed of all time list. These three contributors have 2000, 2500 and 6500 posts attributed to them. It’s fair to say they aren’t hitting it out of the park with every post they publish.
  • Time doesn’t matter when it comes to looking at total views count. Older posts that don’t “go viral” never reach the same level of views as new posts that do.
  • The most viewed posts didn’t involve cute baby animal. The list titles incite emotion with titles that include words like, “most powerful” and “restore your faith in humanity.”
  • The Buzzfeed stats page includes views from social and it is clear that Facebook is the content amplifier of choice. Circling back to the top three contributors, their viral posts all have over 50% of their views from Facebook. Each of the contributors have a relatively modest amount of followers, but those followers provide the starting block for content getting up to viral speed.
  • http://www.mikemerrill.com Mike D. Merrill

    My webinar on 25 Ways to Get more customers wasn’t scientific so glad to see the number is ideal. Great data here.

  • Douglas Morris

    I think this is pretty poor analysis. It relies upon the absolutely massive assumption that there’s an even distribution of good content amongst posts of varying length. But even if there was, perhaps there are just more posts with top 20-25 and therefore more viral ones?

    • http://rjmetrics.com/ Tristan Handy

      Hey Douglas! Thanks for the feedback. You raise two great points.

      1. Assumption of randomness in quality of content.

      This is accurate, we did assume randomness here. We didn’t feel that there was any reason to believe that the number in the title of a post would impact the quality of content. As long as these two factors aren’t related, then an assumption of randomness is reasonable given a large enough sample size (a bar which has clearly been met).

      2. Only measuring # of top posts with a given number, not controlling for frequency of occurrence

      This is a bigger issue. While we did consider this factor in pulling together the post, there wasn’t a good way to get total # of posts with a given number in the title without writing a full-blown web scraper. It does seem that some numbers would appear more frequently than others in the title of posts, so we can’t just assume randomness (as above). In the end, we consoled ourselves that, while this might not have been a perfect approach, ultimately it was just for a spot of fun ;)

      Thanks for the comment; much appreciated!

  • http://www.referralcandy.com/ Visakan @ ReferralCandy

    I’m not a fan of buzzfeed/upworthy type articles, but I’ve always been interested in what we can learn from them, and what they tell us about ourselves. This was interesting to read. I do agree with Douglas that we can’t really tell much from the frequency distribution.

    What I did find interesting- the subject of the top grossing posts (change your view on humanity, see before you die, most powerful images, most powerful photos ever taken)- a lot of them promise “drastic” perspective.

    Couple that with the fact that they’re shared primarily on Facebook, where people post and share things to co-create their identities and to earn goodwill from one another… and I think we’d get to the conclusion that the “magic formula” for list posts is to pick the right subject matter, to give a certain sort of promised experience that people would be interested in.

    I also suspect that lists with <10 points might seem "trivial", not worth clicking through. 50 pictures, now that's something you can consume fairly rapidly, scrolling past stuff that doesn't bother you.

    Interesting to think about. Thanks for putting this together!

  • Eric Van Buskirk

    You
    guys are doing original data research for your blog posts. Love it.
    Opinion pieces are becoming far too pervasive for topics like this.
    And actually 25 makes sense to me. 1-9 don’t have any meat on them.
    However, for those of us over 40 (including me) I think the answer is NO
    NUMBER is the best. Does the NY Times, Wash Post, or Miami Trib use these
    in there headlines? No. In fact way, way back in 1992 the only
    people using the numbers in headlines constantly were the know “fluffy”
    or “low-brow” publications. So, Inc magazine never used to do
    that, while entrepreneur mag I think did
    (’cause the kinda suck and have a lot of sensationalistic articles).

    I know titles with numbers get far more clicks, but they also miss the savvy
    people, not just over 40, who are turned off by click bait and headlines that
    insult their intelligence be by have useless info in them (1-25)

    Keep up the original research! +1 to you all.

    • JanessaLantz

      Eric,

      You make some good points about the overall value of “listicles.” Thanks for taking the time to weigh in on this.