Good question. I started by reading basically everything I could on the subject after watching a prez Josh Elman gave on it (https://www.youtube.com/watch?v=uCIQPV6lOdY). Also, looking at how other companies have defined it (Facebook, Twitter, Medium, etc) has been helpful. I still think the metric that answers the question how many people are actually using your product is the best way to approach it.
If you define active users right then the # of active users is actually a pretty good answer to that question. Note that almost all of the analytics platforms (Google, Flurry, Mixpanel, Amplitude, etc) define “Active User” as one who opens the app but just going by that will not get the job done and will not answer that important question.
For example, Rapchat is an app that lets users record freestyles over beats and share them with friends. For us, an active user is somebody who completes at least one core action in a given time frame (recording a rap, sharing a rap, listening to a rap or message). Since we have both content creators and consumers on the platform we needed to include core actions from both sides.
Time is another important factor. For us, we found weekly to be most fitting because we don’t expect users to record or play raps everyday but we DO expect users to engage with the app at least once a week. For a messaging app like Facebook Messenger, they’d be better off focusing on DAU rather than WAU or MAU because they fully expect users to communicate with other users every single day.
Hopefully that helps! Also, it’s worth noting your OMTM should evolve over time so I fully expect ours to change as we scale.