A while back, I wrote about how Session Duration (sometimes called Time on Site) is not a good metric. My primary argument: it’s not accurate. But let’s imagine that new technology made it accurate. It’s still not a good measure of engagement.
A dozen or so years ago, when I was a wee production support specialist at an agency, we launched a new website. It had an old-school log-based analytics system. My boss asked me to run before/after numbers from the launch (one of my first forays into analytics). I gloomily sent him the results. Time on site had gone down by a significant measure.
The following week, I sat in the presentation where he gave the results. He moved to the slide that showed the decrease of time on site. I held my breath. “Clearly”, he said with a wide grin, “the changes we made have made it easier for users to find what they are looking for.”
“Man, that guy can spin ANYTHING,”* I said to myself. I was sure, after thinking about it, the client would call his bluff. I took a look at the older website. I tried completing a few high level functions. Then, I went to the new site, and did the same. It took less time on the new site, and was a better experience. Time on Site going down was a good thing. No spin required.
Think about it. Would you rather spend 20 minutes on the website of your Health Insurance provider? Or would you rather spend 3 minutes?
So whenever anyone asks me to report session duration (or time on site), before I say yes, we conduct an exercise. I ask them to imagine a user. See the device they are on. Imagine their location. Then ask yourself, what does this user need to do before I consider them “engaged”. Do they need to read an article? Or 5 articles? Do we want to make sure they read it all the way through? Do we want them to participate in an activity? Any activity? A specific activity? With this information, you can find a series of metrics that are within the context of your needs, your website, and define engagement.