I came across “What is user engagement” article by Jordan Willms of Work at Play. In this post, Jordan defines a good engagement metric as the number of user actions divided by the number of content items. The example provided is of blog posts comments divided by the number of posts per month.
This is certainly a very interesting metric. It made me think, does this also make sense to web application usage?
Following the same logic, in web applications (as opposed to a content site) we would count features and divide those by the number of user actions. I’ll use Google Docs as a simple example:
Say the features are: New, Save, Print and Share document. Total of four features.
For each user the engagement metric would be the number of activities divided by four.
Let’s see the score of two users over a week of usage
| Activity | User A | User B |
|---|---|---|
| New | 10 | 2 |
| Save | 10 | 2 |
| 10 | 2 | |
| Share | 2 | 22 |
| Score | 32/4 = 8 | 28/4=7 |
User A scored 8 while user B scored 7. By this engagement metric user A explores more the feature set of Google Docs and is more engaged.
On the other hand user B uses a more competitive feature of Google Docs, which is Share Document, hence might be a more valuable user of Google Docs.
Although this engagement calculation method isn’t perfect I’m convinced that this it is better measurement of user engagement then by just counting “number of logins” which is what most application owners do.
What do you think?
Adam covati says:
I like the idea of dividing actions by items to normalize the number a bit. Just as with scoring, I think weighting should probably play a role as well.
Some actions require more effort, are more significant, or perhaps are more reflective of target behaviors. As such, those should be given more weight.
Guy Nirpaz says:
Adam, you are absolutely right. We use the term ‘sentiment’ to define not only the weight of the user action, but also whether it’s positive or negative.