The Optimal Blog Post Length to Maximize Viewership
Anyone who has ever penned a blog post has asked, how long should this post be to maximize viewership? I’ve often wondered the same thing, particularly in the moment before I click the publish button and broadcast a perhaps-too-short-perhaps-too-long post into the Interwebs.
I’ve written 256 posts in the past 18 months and I sought to understand the impact of word count on every metric I could measure: page views, time on site, time on page, bounce rate, exit rate, retweets and favorite activity.
Here’s the data. In short, post length has no meaningful impact viewership, engagement or sharing.
Below is a chart depicting the correlation of word count to six key metrics. In each of these cases, the correlations are small (<25%) and relatively speaking insignificant predictors of traffic.
Metrics
My posts vary from 100 to 1100 words, with the majority between 300 and 750 and a median of 461 words. These posts aren’t exceptionally short or exceptionally long, but I imagine they are comparable with most blogs that post near-daily. Below is the distribution of the 256 posts by word count.
Plunging into the finer points of the analysis, below is a box plot demonstrating the relationship between page views and word count. I’ve bucketed posts by keyword in 200 word increments. Note the Y-Axis is Log10. The boxes show the distributions of page views in each bucket. The middle line in each box is the median, the borders of the rectangle show the 25th and 75th quartiles and the circles are outliers. More on reading box plots.
There is no difference in the traffic generated by longer or shorter posts. Though the ranges of the box plots vary, the medians are all very close to each other and while the 200 keyword bucket does seem to indicate a narrow distribution of page views, the three outliers at the top indicate the conclusion is likely a product of the smaller sample size.
In the end, the data reaffirms what I probably already knew to start. The best length for a blog post is the length required to capture and convey the message. No more, no less.
Talent Acquisition Expert- Talent Transition Consulting - Customized Training - Learning and Development - Coaching
6yI needed that, thank you.
Helping Technology Marketing Teams Streamline Technical B2B Content Creation and Get Their Lives Back.
9yI don't think the question is really what gets someone to read your blog as much as what gets them to share it. Data since late 2013 and Mid 2014 show that blogs above 1500 words get more shares. Yes they need to be compelling and have lots of subheads because the people sharing the blog are often skimming, but that said more substantial blogs with highlights called out seem to be more effective. https://blog.bufferapp.com/the-ideal-length-of-everything-online-according-to-science http://www.business2community.com/blogging/blogging-best-practices-the-ideal-length-for-the-perfect-blog-post-0577303
publisher
10y[2nd post] I agree the length needs to match your purpose. And where there is much information, breaking it into parts seems to me essential for a blog, just as collecting it together is essential for a book. Holding surfers' interest seems to me another matter. How well have you targetted to get the interested rather than the casual? All analysis and information is of great value, but I do very often feel the real question is "How long is a piece of string?" :-)
publisher
10yNot only do the article and comments already made add up to a deal of information that is useful as part of understanding, but the key is probably something else. Your readership. That reflects back into the market you address, and the purpose of your posts. The biggest problem is misleading advice; Dawn, the figure you were given relates to the capacity of computer hardware. Early on, when RAM, in particular, was in low Mbs, It might possibly have been a wise limit. But I'm fairly sure that was in the days of DOS - and maybe early Windows95, and I don't recall starting on the net with so little working capacity. You might still see advice to keep to a limit of 256 colours; I don't think this limit has any meaning any more, either. [more in second post]
Freelance Writer at Gittle Publishing (Pen Name Aviva Gittle)
10yTomasz, you've got me thinking. :-) Technically, your study is inconclusive; the data set isn't large enough as Tamera Flora pointed out. But individual bloggers are concerned about *their* readership. Conducting experiments using your own site makes sense. You will learn which variables have impact on time per page, page views, bounce rate, etc. on your own site. Seems like heat mapping software would be useful this. Would love to hear others' ideas on the best way to conduct a single-site study.