On Monday, I spoke at the Open Analytics Summit here in NYC. I talked about our system for measuring Engaged Time: our research into measuring it, the backend that we’ve built, and some places where engagement data sheds an interesting light on users’ experience on the web.

I know you folks are busy, so I summed up my talk in three key points:

1) If the current metrics you have available don’t give you the information you need, develop new metrics instead of trying to back into results with faulty data. When we started thinking about measuring user engagement, we quickly realized that our questions couldn’t be answered with event-based metrics like clicks and social shares. Rather than trying to proxy out engagement, we built a new system for tracking time.

Turning to the conclusions we’ve gotten from our engagement data:

2) Many users don’t engage with a page, but once they do they tend to stay engaged. As we talked about in our recent piece in Slate, a substantial number of users don’t read past the start of an article. But, once a reader starts to scroll, they’re more likely to read at least a few thousand pixels than they are to stop reading. Your key as an editor is to get people to take the first step toward engagement.

3) There’s substantial value in engagement. Through a variety of large studies, we’ve seen that users who engage with content for longer are more likely to return to a site and may be more valuable to advertisers.

Thanks to everyone for coming, and, for those of you who missed it, feel free to ask questions in the comments!