Long-form metrics: what to measure and how to measure it

In wonkworld and elsewhere, communicators are experimenting with long-form content. Lengthier essays, feature articles and thinkpieces are becoming another way of getting across the often complex messages we try to communicate. Of course, writing that goes over the 2,500-word limit is nothing new. Online long-form content can feel fresh and exciting though because it allows us to add depth and texture through clever use of transitions, interactions and video, and because for so long the mantra has been ‘longer stuff for print, shorter stuff for web’.

Content has always been king of the internet but the recent explosion of long-form means the king has a new-found swagger and a glint in his eye.

The benefits of long-form content, reasons to use it, how to design it and ways to build it are all covered elsewhere. This article is just about measuring the extent to which your audiences actually pay attention to it.

I’ll be using recent examples from The King’s Fund. We’ve been experimenting with different styles and templates over the last 12 months to see how long-form might work for us. In order of publication they are:

Deciding what you want to measure

Do this as early as you can – before you pick the technology you’re going to use, before you even start creating your actual content. If you know at the outset of the project what you want to measure it’s a lot easier to bake the actual measurement into the code.

Your metrics will likely be based on the objectives of the project (eg reach, taking action) and a desire to understand how your audiences engage/interact with the content. The questions you’ll want to answer might look something like this:

  • How many people read the article?
  • How much of it did they read?
  • Do my audiences prefer long-form HTML to PDF?
  • Did they interact with videos, photo galleries or other elements?
  • Did they share the content, and if so which bits did they share?
  • Did they take any of the actions?
  • Do users behave differently on desktops, tablets and smartphones?

How many people read the article?

This is one area where HTML pages clearly outshine PDFs; it is much easier to have a reasonably accurate idea about how many people have read an HTML page. Yes, you can track PDF download clicks in Google Analytics but that doesn’t capture the many who go direct from search results to a PDF (you need log files for that) and who knows what happens to downloaded copies that have been read and shared offline?

How much of it did they read?

On an HTML page as well you can get an idea of how much of your long-form content your users are reading.

Time on page isn’t a particularly useful metric for this because it ignores pageviews that are bounces or exits – if 80% of people who read an article leave your site after doing so, the time on page metric is only accurate for the 20% who went on to another page afterwards. And it’s not even particularly accurate for them as time on page can include the time users spend looking at other browser tabs or just looking out of their window.

A better method is to track how far people people scroll down your content. There are several different methods you can try, all of which are about a medium in terms of difficulty to implement.

We use Scroll Depth for this. Out of the box, it creates events in Google Analytics whenever a user scrolls past certain points in the page: 25% of the way down, half-way down, all the way to the bottom. If your page has comments or a particularly tall footer you may need to specify where your actual content ends so that a custom event is triggered when the user reaches that point. For example, 50% of users got to the end of this 3,000-word article but a much smaller number got to the 75% scroll point because after a short while that point was halfway down the comments section.

(Hopefully, that ‘50% of users’ figure got your attention. Across all of the long-form content we’ve published, we find that about half of users get to the end. We think this is pretty good.)

Do my audience prefer long-form HTML to PDF?

This isn’t something that we’ve tested a great deal but it is quite easy to answer this question. Near the top of your article, offer a PDF of the same article, then count the number of users who download it, the number who scroll, say, 50% of the way down the page, and the number who do both. Repeat this a few times and you’ll have your answer.

Where we did test it, for example on this report, we found that for every one person who downloaded the PDF, 1.6 people scrolled halfway down the page.

Did they interact with videos, photo galleries or other elements?

As mentioned earlier, one of the benefits of long-form is the ability to embed rich media and interactive elements that add depth and texture to your storytelling. But do your users actually pay attention to these or do they skip past to get to the next paragraph?

On kingsfund.org.uk we use vimeo.ga.js for tracking Vimeo, the CardinalPath method for YouTube and Simo Ahava’s method for Soundcloud. Each of them fires a Google Analytics event whenever a user presses play or pause and tracks how far they get through the piece of audio or video. Similar methods can be used to track clicks in photo galleries or expandable sections of the page.

We’ve not tested this thoroughly, but where we have (on this feature article) we found that around 6% of users pressed play on a video, 5% expanded one of the case studies to read more, 3% clicked on the image gallery and 4% used the sidebar navigation.

Did they share the content, and if so which bits did they share?

It’s often possible to share individual elements of long-form pages. We don’t do a great deal of this at the moment but other sites do. Scroll to the photo of Paul Johnson and Nigel Lawson on this Guardian long read and you’ll see Facebook, Twitter and Pinterest buttons that allow you to share a link direct to that part of the article. This trick is perhaps most effectively used when sharing quotes – move your mouse over the big yellow pullquote near the top of this Buzzfeed long read and you’ll see Facebook and Twitter buttons that let you share the exact text of the quote.

In terms of measuring this, you can either track the click that begins the share as an event in Google Analytics or you can use the new-ish social interactions data model.

On top of this, you should ensure that any URL in the sharing text has appropriate tracking. For example, if you were to share a page from The King’s Fund site using the Twitter button at the top of each page you might notice that the page’s URL in the tweet has this tracking code on the end: ?utm_source=twitter&utm_medium=social&utm_term=socialshare. Anyone who comes to our site by clicking that tracked URL can be identified as someone who has followed a link in a tweet from another site user who shared our webpage using the buttons at the top.

Did they take any of the actions?

Do you want users who read your article to sign a petition, make a donation or take some other form of action? If so, track the click, set the action as a goal and count the conversions.

Do users behave differently on desktops, tablets and smartphones?

Finally, it’s worth noting differences in user behaviour depending on the device they’re using. For example, do users on smartphones read as far as desktop users? We find they tend not to, but tablet and desktop figures are about the same. If bounce rates or scroll depths are wildly different between devices it might be worth exploring the reasons in case there are problems that need fixing.


Have I missed anything obvious from this list? Do you have an example of long-form content that you want to share? Do your audiences behave differently? Add a comment here or twit me at @jiggott.

I do digital comms and projects for The King's Fund. Also: occasional drumming, some culture, a bit of cricket. South London / Crystal Palace / Anerley resident.

Tagged with: , , , , ,
Posted in Guidance
One comment on “Long-form metrics: what to measure and how to measure it
  1. […] Also definitely worth reading is our very own James Higgott’s article on long-form metrics: what to measure and how to measure it. […]

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

%d bloggers like this: