Time Spent: A Vanity Metric
Worthless when Independent, insignificant when combined with other metrics
Prelude
In the last 6 years, Google Analytics has been the Grand daddy of Analytics, it supposedly runs on 67% of the top 1M sites as reported by BuiltWith. Due to this Google Analytics defines the mental models for Analytics for most folks in the field. The model that most people are tuned to is that of page views, bounce rates, time spent , etc. For today’s discussion I will be trying to breakdown time spent and it’s perceived value.
Problems
It can’t be tracked effectively
In an age of tabs & side by side Apps, just because an app came to the foreground does not mean your product is being used.
It does not imply that the user was engaged
The human attention span can be as low as 8 seconds, combine that with a Smartphone where you can get a notification every few mins, it is pretty much certain your users will try to multitask
It does not provide any insight
Let’s say you are an E-Commerce app. If the time spent on App goes down, is it because you optimised your flow with a faster search & checkout experience or is because you folks are not finding good deals? Both scenarios are plausible. It can’t singly handedly provide any meaningful insight.
Diving deeper
Time spent independently provides no value
Let’s take 2 users for e.g. Lets assume both spent 90 minutes on your platform in the last 4 weeks. How do you value this user over the other? You can’t, you need more data.
Time Spent needs a secondary dimension
Let’s split the minutes by the number of videos viewed by both. With this we get a sense that User B might be more loyal to us, but we can’t define why or how yet just by looking at time spent.
To do that, let’s split the video views by weeks of the month, this should give us a better idea of when the users engaged with us.
Now we can clearly see that User B is more engaged with us, also as a bonus we can state that she likes short form content.
Replacing time spent with events
Let us take it a step further and replace time spent with videos start & completed events (depicted here with smaller blocks).
It is now easier to make a comparison as we are comparing oranges to oranges.
A counter argument to this exercise might be that time spent is important as that defines the number of Ads someone might watch. But we would be better off tracking Ad impression as events in such a scenario. As long as you know what is the ratio of your content to Ads you can make simple calculations when needed. Another aspect the events show us is that user B might have started consuming more content from us.
Conclusion
Since we now know that time spent in itself does not really indicate anything, we don’t need to take the effort of tracking it. The only place it adds value is on video consumption. But as we showed above, it can’t stand on its own. If you were to read the spec of Digital Video Ad Serving Template (VAST), they suggest tracking of the video in quartiles not in minutes. Another point to note, Facebook released Video Ads minutes watched metric in Feb, but also provide 10 Sec video viewed metric & completion metric which are more valuable than the time spent.
What do you think?
Do you track time spent? If so why?
Struggling to make sense of your data? We are building a product to solve your woes.
Did you find this post useful? Please do hit the 💟 button.
Interested in keeping up with our posts? Want more? Follow us :)
Photo credit: Cyberslayer via Foter.com / CC BY-NC-SA
