Using Engaged Time to understand your audience

Audience development is a topic on the mind of everyone in publishing, and we here at Chartbeat spend a great deal of time thinking about what actions affect your audience and how to measure their effects. Today, I want to talk about how to use Engaged Time as a metric for audience development. I’ll give you some numbers to watch, walk you through a few use cases, and give you a few hints about some products to come.

The process of audience development can be broken down into four (simplified) steps:

  1. Visitors come to your site
  2. They find content they’re interested in and engage with it
  3. They like what they get, and they choose to come back to your site again
  4. If you’re lucky, they also share what they found with others

Each of these parts is critical, and there’s drop-off at every step along the way — users who come don’t always read and those who read don’t always return — so doing what you can to improve each step can have a huge effect on the growth of your audience over time.

Traditionally, the analytics industry has gotten pretty good at measuring steps 1 & 4 — raw traffic and social media statistics — but quantifying user engagement and propensity to return has always been difficult to do. Engaged Time provides a valuable metric to help fill in those missing data points.

Before we dive in, I want to give a quick definition of how we measure Engaged Time. While a user is reading a page, we count up the amount of time they spend with the page in an active browser tab — a foregrounded tab where the user has recently scrolled, typed, or moved their mouse — and then average that number across users. Note that this number contrasts with traditional time on page, which measures how long users keep pages open rather than how long they actually engage with pages.

Pageviews and engagement — What’s the difference?

If you think that page views are sufficient to measure engagement, think again: all traffic is not created equal. Our goal on the web isn’t just the first step — to bring people in the door — we want folks to actually read the pages we’ve worked so hard to make.

Unfortunately, raw traffic volume numbers don’t speak to how people interact with your pages.

Click data on article pages tells you how enticing links were, not how engaged readers were with the content inside.

To see the difference between clicks and actual engagement, I pulled a sample of 100,000 page visits to Chartbeat sites across a week.1 Of those, about 34,000 of them resulted in users leaving after less than 15 seconds of interacting with the page —

A full ⅓ of the visitors leave without exhibiting any signs of engagement.

But that ratio isn’t the same for all pages: of the 10 most popular articles I looked at, one had 91% of visitors actually engage with the content and another had 93% of visitors leave without ever scrolling down the page.2 From the perspective of page views, these two articles were almost exactly the same, but I think we’d all draw very different conclusions about the success of their content: certainly traffic volume to a story matters, but you also have to ask yourself whether people actually read it.

Readers who don’t engage represent missed opportunities — it’s hard enough to get readers in the door, and we want to make every visit count.

What is Average Engaged Time?

Average Engaged Time is a Chartbeat metric that measures the amount of time that users spend actively interacting with a page – reading, writing, scrolling, watching – and it’s a great place to start when looking at how well your content matches with your audience.

To see what I mean, take a look at the figure below, which shows users’ Average Engaged Time on one specific article, broken down by where they came from.

Screen Shot 2015-12-15 at 5.19.00 PM

Clearly the traffic source dramatically affects not just the volume of traffic, but how long users read for. For instance, the article took off on Facebook — its top referrer — but we can see that readers who came from Facebook links spent dramatically less time reading than readers who came from internal links.

(Related: What happens when Facebook goes down? People read the news)

When looking at an article’s Engaged Time, think about how long you’d expect a reader to take to consume a piece of content. For some rough guidelines: we typically see 60-90 seconds of engagement for news articles, 5-30 seconds for landing pages, and 3+ minutes for long-form content. If an article’s Average Engaged Time is substantially below what you’d expect, it’s not reaching an audience that wants to fully read it:

  • Is the headline out of step with the content, so people are not reaching the story they thought they were getting?
  • Is the format of the article off?
  • Are you reaching out on the best traffic sources for this story?
  • Or is the article simply not as engaging as you’d hoped?

Conversely, if an article’s Average Engaged Time is exceptionally high, you’ll want to get as much traffic as possible on the page – create internal links and take a look at where its current audience is coming from and outreach to similar traffic sources.

Of course, not all content takes the same amount of time to consume, but you should always make sure that you’re getting as much as possible out of each piece of content you produce.

Converting engagement to visitor loyalty over time

We started to measure users’ visits to sites across time and investigate what parts of a visitor’s browsing behavior indicate that they’re likely to spend more time on your site in the future. The call to action here is clear:

Over 60% of visitors to an average site don’t return again in the next 30 days.

But, there’s a lot you can do to have a dramatic effect on that number, and even small changes can have large effects on your site’s traffic over time.

One thing that stands out across the board is that users’ Engaged Time is strongly correlated with their loyalty to your site. Below is a figure showing the relationship between the maximum amount of time visitors spent reading articles one day and whether they returned to the site across the rest of the week.

joshgraph_2c

Visitors who read an article for three minutes returned twice as often as those who read for one minute.

Intuitively, this makes sense: if you can get someone to actually find an article that they like enough to read, they’re much more likely to return.

Of course, we can’t confuse correlation with causation: we can’t say that high Engaged Time causes readers to come back. But, since we know that users with high Engaged Time do come back more often, we can say that when you see an article with high Engaged Time, it’s much more likely that the readers of that article will visit your site again in the near future.

This idea isn’t something we’re seeing in a vacuum on our sites, others have found similar results; ex-New Republic Editor Andrew Sullivan reported similar findings in this great podcast on developing a paid audience.

How does Engaged Time impact my strategy moving forward?

If I can stress one thing, it’s that you want to do whatever you can to make sure that all of your readers find their way to some piece of content that’s highly engaging every time they visit.

If a short piece of content suddenly blows up, don’t miss out on the opportunity the traffic spike presents: try to drive people from that piece to other, longer-form content that’s likely to be engaging to your audience.

Coming from the other side: pages with high Engaged Time are likely being read by your most loyal audience. These pages are your best candidates for things like subscription links and newsletter signups that are designed to take interested readers and make them directly loyal to your site.

The unifying message in everything above is that you want to try to put the right content in front of the right audience.

Doing so will drive up visitors’ time on your site right now and increase their loyalty to your site in the future.


1 Non-paywall sites, to ensure that the numbers weren’t thrown off by people being unable to read articles after they hit a paywall.

2 Both articles were roughly the same height; the lack of scrolling wasn’t because of a shorter page.


More in Research