Ryan Sweeney

Today we have a contribution from NPR digital metrics analyst Dan Frohlich. We’ve often wondered if the length of the lead-in text affects a post’s performance on Facebook. Here’s what he found. – Wright

Here at NPR we’re always looking for different ways to improve our performance on Facebook. It’s one of NPR.org’s biggest drivers of traffic and one of the primary ways we engage with our audience. So recently we decided to take a look at one particular aspect of our Facebook posts – the length of our lead-in text.

NPR’s Facebook posts vary in length, from the very short to the incredibly long.

We exported all of the link posts we published for the last 6 months, over 3,000 in all, and analyzed post performance across groups in 40-character increments. We chose groups of 40 because prior research in this area concluded that Facebook posts that were 40 characters or fewer performed best.

It was illuminating just to see how often we publish posts of different lengths. We also discovered that the typical NPR post was around 200 characters long or around 40 words (using Wolfram Alpha’s approximation that the average English word is around 5.1 characters).  Keep in mind, these results are for our main Facebook page but we see similar results on our Morning Edition and NPR Music pages.

Since driving traffic to NPR.org is one of our main goals on Facebook we looked at click-through rate (link clicks divided by post impressions) to get a sense of how efficient each group was at delivering traffic. What we uncovered is that shorter posts (specifically posts that were 120 characters in length or fewer) tended to have significantly higher click-through rates. Meaning, when all things are equal, these posts are more effective at delivering traffic to our site.

Great! Shorter posts are more effective at getting people to click. Let’s just end it here. Well, hold on. NPR’s mission is to “…create a more informed public — one challenged and invigorated by a deeper understanding and appreciation of events, ideas and cultures.” Given that, delivering traffic back to our site is a great goal (more reading = more learning) but what if someone could learn something from our longer posts without clicking through? We feel that’s also valuable to our audience and there is some data to suggest that could be happening.

One vague metric Facebook provides is something called “other clicks.” Facebook offers this super descriptive definition: Other clicks refers to clicks on the “see more” label (that are on longer posts) or clicks not on the content of the post, such as the page title. Those “see more” clicks are the part we’re most interested in and what we believe causes “other clicks” to increase as posts get longer.

Based on this we came up with a new metric we are calling “Adjusted Click-Through Rate.” This new metric is calculated by taking link clicks, adding in other clicks, and then dividing by post impressions. It produces a new result, which allows us to think twice about how we view our longer Facebook posts.

So how does this impact our approach on Facebook? It means two things:

1.     Our editorial team is going to try writing shorter posts to get people to click-through to our website a bit more often. This doesn’t mean we’re giving up on posts over 120 characters because we still feel like having a broad range of posts is beneficial. Basically, we just want to shift the distribution of our Facebook posts so that we’re writing shorter ones a bit more frequently than we are now.

2.     Since we feel like some longer posts can still educate and inform the public (and perform well according to our measures), we’re going to continue doing them when we feel like a story calls for it. We feel like this allows for some exciting story-telling possibilities on what is quickly becoming America’s go-to source for news.