Posts Tagged ‘Mobile’

Understanding Engagement Patterns: Mobile vs. Desktop Users

January 30th, 2013 by Josh

Chartbeat's TotalTotal is something I can watch forever — I never stop being amazed when I see just how much traffic we're measuring at any given moment. Self-congratulating aside, I want us to take a look at one number on this screen that should stand out to you. total_total_mobile As I'm writing this, the mobile traffic we're measuring is sitting at about 3.5% of total traffic, and that's pretty standard for us. By contrast, most estimates out there put mobile traffic as having around a 20% share of total traffic. That's a pretty substantial difference, and it made me want to take a closer look at what's going on. What I found has some interesting implications for how you should look at your numbers and how you might think about your mobile user experience.

How Chartbeat measures

The answer begins with what numbers are being used to generate those percentages. Most estimates of mobile traffic are based on page views — when people say 20% of traffic is mobile, they mean that 20% of page views are happening in mobile browsers.

As you know, at Chartbeat, we measure in concurrents – the number of active browser visits on a site at a given time. Concurrents take into account both how many people visit a site and how long they stick around. What does that mean exactly? Imagine that one new user comes to your site every second for a period of 10 seconds. If each of those ten people bounces immediately, then there’s always just one user on your site (the one who came that specific second), so we’d measure one concurrent. If each person stays for a whole minute, though, there are 10 users on your site after 10 seconds, so we’d measure 10 concurrents. In both of these scenarios, the number of page views would be the same – ten page views– regardless of whether the visitors left immediately or stuck around.

Concurrents aren’t limited to examining the quantity of traffic, and consider the quality of your visitors as well. We’re further able to break down concurrents into more categories based on whether users are engaged or idle, but more on that in a bit. Let’s go back to mobile: if around 4% of concurrents are mobile but 20% of page views are mobile it must mean that mobile users are spending less time on pages.

The mobile vs desktop user dilemma

Naturally, the next question is: what makes mobile users’ time-on-page so much lower than their desktop compatriots?

One major difference between mobile and desktop traffic is that mobile users rarely go idle: a page is either the single open, active page on the device (in which case we’re measuring the user as an engaged concurrent on that page) or closed (in which case we’re not measuring them at all, since they’re no longer there). By contrast, desktop users frequently open tabs, leave them idle, and come back to them over and over, which means desktop users cycle between being measured as engaged and idle concurrents. Almost all mobile concurrents are engaged, while most desktop concurrents at any given point are idle. So, when we look at the pool of concurrent users, there are many more idle concurrents from desktop than from mobile.

What you need to know

I think there are two big takeaways here:

1) Compare apples to apples.

Since almost all mobile users are engaged, if you’re using data from our APIs to determine in real time how much of your traffic is mobile, you should probably be comparing your mobile concurrents to your engaged concurrents (which includes engaged mobile and desktop users) rather than total concurrents.

When we look at the ratio of mobile concurrents to engaged concurrents, we get a number that’s similar to page view figures (around 20%), but with all of the advantages of the concurrent as a real-time measure.
2) Design with the mobile user in mind.
More importantly, the difference in engagement patterns between mobile and desktop is critical when thinking about your site design. On desktop, one very common behavior is for a user to leave your homepage open and idle all day and occasionally come back to read and open stories, and there’s simply no equivalent behaviour on mobile. In terms of metrics, this means for mobile pages (and pages with a high proportion of mobile traffic) you should concentrate even more on pushing your engaged time numbers as high as possible. For a mobile user, when you’ve got their attention you really have it, but when you lose it, you really lose it — plan your site accordingly.

What about all of those idle users on desktops, though?

Of course, there's another side to this data: at any given moment, many of your desktop users are idle. That might seem like cause for concern, but I'd argue that it shouldn't be. Desktop users flip in and out of being idle, returning to your tabbed-out site page and then leaving it in the background again — that's just part of the process of browsing on a desktop. To analyze the specifics of that process, I pulled a sample of about 100,000 page sessions and found some interesting stuff:
  • Of those 100,000 sessions, 54% involved the user going idle for some part of the session
  • But critically, once a user went idle for the first time, they spent an average of 46 more seconds reading that page later in the session
  • So, the bulk of their reading actually happened after going idle for the first time.
So, don't think about idle users as a lost cause; they're just people who aren't currently reading. We've also noticed an interesting property of users' browsing behavior: The amount of time that people spend idle on landing pages actually correlates positively with their likelihood of returning to your site on future days.  If you want to know what sites a person visits over and over every day, looking at their idle tabs will probably give you a pretty good set of guesses. We're doing a huge amount of research focusing much more on this detail exactly — analyzing exactly what types of browsing predict a person's likelihood of returning to your site — so stay tuned. As always, feel free to post your questions or ideas in the comments section below.