A recent Bo Sacks post included this interesting paragraph.
“Norwegian publisher Amedia reaches 2.4M readers daily across all platforms. It uses a machine learning algorithm to identify characteristics of user behavior that indicate happy and satisfied customers. The algorithm analyzes up to 70 reader behavior statistics and boils it down to a single number called Engagement Index which indicates the likelihood of a reader to stay loyal. What adds to its value is that the index captures factors that are under the publisher’s control.”
“70 reader behavior statistics”!! That’s a lot.
Depending on how you count them, I can come up with about 25.
- Visits / month
- Recency of visits
- Regularity of visits
- Gaps between visits
- Page views / visit
- Page views / month
- Likes on articles
- Comments on articles
- Social shares on articles
- Time on site
- Logins
- # of e-newsletters they subscribe to
- # of paid subscriptions
- # of special report downloads
- other purchases
- Clicks on emails
- Likes on social media
- Comments on social media
- Video views
- – Likes, comments, shares on videos
- Podcast downloads / listens, likes, etc.
How do you get to 70?
But more importantly, are you thinking about this? Are you measuring engagement, and intervening when people are not engaged?
Subscription publications often have “zombie subscribers” — people who sign up, and then never use the service. They don’t stay long. You need to have a strategy to get them on-board, show them how to use your service, and keep selling its benefits.
