Which type of engagement should I focus on?

  • 25 June 2020
  • 5 replies
Which type of engagement should I focus on?
Userlevel 2
Badge +1

Knowing where to begin with engagement can be kind of daunting. You want to pick out users who are active, but have a bit more commitment or loyalty to your product, and even more than that -- it’s not enough to know who’s engaged, but just how engaged are they? When we’re talking about engagement, those are the two questions we’re asking. How many engaged users do I have? And, of those engaged users, what’s their distribution of engagement? We use engagement metrics to identify users who are getting repeated value out of our product, in an attempt to repeat those behaviors in new, and resurrected users, and drive engagement for everyone who comes across our product. 


In determining the type of engagement that’s right for you, it really comes down to knowing what an engaged action looks like, and how often you’d like to see your users doing that thing. In our Guide to Product Metrics, we say “Engagement could be defined as the number of key actions taken, minutes of video watched or number of transactions completed.” And, as it’s (ideally) the second step in a user’s journey after becoming activated, we usually want to set up our engagement metric as a ratio of [engaged users]/[active usage]. This sets us up to successfully determine: of everyone in my active user base, what percentage is finding the aha! moment in their product experience. 


To build an engagement metric, determine the moment a user finds value -- completing a transaction, viewing a report, making a purchase, finishing a video -- and set it as the numerator. Take your defined active usage metric -- daily active users, weekly active users, or monthly active users -- and use that as the denominator. 

For a subscription based video streaming service, a good engagement metric might look like: 

  • Minutes Watched / WAU 

This will help the product team know how deep into their content their active user base is getting. The moment of value for a service like this comes from watching most or all of the provided content. If they can keep a pulse on how much content their active user base is consuming, they’ll have a good leading indicator on whether they’re surfacing or providing valuable content. For an online clothing retailer, a good engagement metric might look like: 

  • Number of items purchased / Weekly Buyers 

The aha! moment in this instance comes when a user realizes how little lift it is to make multiple, quick and easy purchases from this website. Anyone who’s in the single-item-purchase space hasn’t gotten all the way to the real product value. Discovering the ratio of items bought to users who make purchases will help this team determine just how engaging their product is. If this ratio dips below 2, they made need to iterate -- revisiting the inventory of their site, double checking their workflow is as easy as intended, or even possibly driving up the number of weekly buyers.

5 replies


Not sure I follow this. 

Let’s i’m a nutrition app where customer can log their meals. Why is the ratio between log meals and MAU important? why does the number convey? and should I just calculate using a simple A/B formula of Total Logged Meals (event) / MAU (Cohort)?



Userlevel 2
Badge +1

Hi Sharod -- 

Excellent question. That’s a great use case. The ratio between number of logged meals, and MAU will only be important if logging meals is your moment of value, and your active users are determined monthly. This ratio will map out the number of logged meals per monthly active user. You can set a goal for yourself of 30 meals logged per monthly active user, and using the ratio, see where you end the month or quarter. This kind of ratio approximates how frequently your active users are running into a moment of value in your product.  


You can use the A/B formula of Total Logged Meals / MAU. I would recommend using the formulas feature in Insights, selecting the Logged Meal event, with the “total” operator as (A), then selecting the Logged Meal event, with the “MAU” operator as (B), and building the (A)/(B) formula just below it. 


Hope that helps! 


Yes, thanks @kileysheehy

Everyone forgets about the memory parameter. The customer must want to return to your product after the first purchase or transaction. Customer returns are an important parameter

You're right. Refunds can be received and processed. This is an important statistic that an experienced product manager uses to monitor the quality of their product. Using repeat customer returns is good expertise but not unique. I recently read several texts about human memory. You should focus on the details of the product that will be remembered by the customer. It can be an innovation that sets your product apart from others.