SaaS:Engagement, Retention all the sticky stuff

This is my first post on engagement and stickiness. My goal is to take a lot from the book Lean Analytics along with other sources but most from this book, and create posts that I can then pass on to anyone to read or for myself to reference when I want to address different topics in application development and management. This book has become somewhat of a bible to me. Whenever I am at a stage in business, this book address each stage to such detail. This book has become my most referenced book over the past two years.

I'm going to jump past the first steps to any application in this post acquisition (pirate metrics - AARRR/attention & enrollment (lean analytics style) to engagement/retention.

Step 1: Segmentation of users

The first step to tracking your engagement is segmenting out your high interaction users from causal users. This is especially true if you have a freemium product. Find what is common among your high level users. Such as location, all referral from a social network, all under an age, over an age, all in the same industry or work position, anything you can workout that is common among your high usage users.

Workout who is using your product at a the highest level and see what is common about them. Maybe you get more than one segment, that is great!

Step 2: Define your stickiness actions

Now that we have these segments lets define your stickiness actions. From these segments workout what is the common actions these users all do when they login. Try to see what they do 85% of the time or above. Example might be, users all login, check notifications, click on at least one notification link, if they have mail they check that, they check user posts of at least three of their friends. If you established more than one segment do this for each segment. Once you have done this for all what is common? Once you have this you have your stickiness actions.

Step 3: Establish usage level

Next we are going to establish what the usage level of your user segments are. How often do these users login and use the application? Every 3 hours, every day, every week? Track all of the established segments and average their access to the application. This will be your current usage level.

Step 4: Set usage goal

Now lets set your usage level goal. Do you feel the current segment level is on target? Do you feel this should be higher? For some products like Facebook, they want this to be supper high. They are looking to get users to have usage level on the scale usage of minutes. Other products like Google Maps, might be daily now but as they grow I am sure they are moving this to be more and more. There are products that are weekly and monthly. Set the bar a bit high, but be realistic. If you set this wrong that is fine, your data (datum for all you grammar heads) will lead you to the right goal later and you can adjust with time.


So now we have:

  • User segments
  • Stickiness actions
  • Usage level
  • Usage level goal

Example 1 of what we can do with this data

Now you can act on this data. The first actions on this data could be when you create features. Think of your application like a backpack. This backpack you have to live out of for the rest of the life of the application. If you have ever over packed a backpack on a trip you will really understand this. At first it does not feel so heavy, but then you start to climb, move...and you are like why did I pack all this crap. Don't let your application become the same way. You want every feature to be something that is used. You will have to maintain this application so make sure you can justify each feature with data.

Using this data with feature releases. Now of course you should track other things with each feature. I am a believer that each feature should have key performance indicators that are attached to each feature. That when you rollout a feature it should be advancing some metric.

What you can do with this data is A/b rollout features to your segments first.

  • Does your stickiness action for the segment change?
    • Does it become part of the stickiness action?
    • Does it lower the stickiness actions?
    • Is the change good or bad you believe?
  • Does your usage level go up/down?

Once a feature passes your split test on the segments then roll it out to all of the segments. From their look at the data again and the questions above again.

Once it passes the full segment rollout, roll it out to all users. From here you can now ask did I gain a new segment? Is the over all usage level going up?

Example 2 of what you can do with this data

A real simple use, you now have clear segments of your users. You can now use tools for advertisement like Facebook and target similar users to try and acquire the type of user that you know will love your product.

I write this post from myself really, but I hope that it helps others. I find that it helps me gain more understanding of the topic if I try and teach the topic and I also will use this to share with my team as well. If you have feedback please let me know, this is for sure an every learning process. I would love feedback!