How do you calculate your Customer Retention?

Flurry Analytics Retention

Today I want to talk about Customer Retention, because most of the time it is calculated wrong way (or at least nudges you to think in a wrong way).

Many analytics platforms show very common picture (like Flurry or Google Analytics)

Google Analytics Retention

On a left side is a User Acquisition day/week/month etc. and on the right side — when did they show up in the app / website.

So when you see graph about your app like here in Google Analytics (GA), do you have to worry? None of 426 users (Jan 12) came on 6th and 7th day from day of join. Is it bad?

It depends.

The main idea I want to cover here, that you should calculate User Retention considering your service (app) specificity.

Let me show you an example.

In our app — App in the Air — people track their flight, explore airports, talk with fellow passengers. Some people fly a lot, and some not. One person can fly every week while other can fly twice a year.

My next flight will be in 55 days

So what kind of retention do you expect from them (assume that they love your app hence want to use it each time they fly)? For the first person — weekly retention and for the second — annual or quarterly.

My flight will be in 55 days. Do I have to visit the app in the next (at least) 54 days?

Here is a problem. You need to make segments, based on each user. Regular analytics tools can’t satisfy such needs if you observe this complex behaviour.

What we’ve done? We started to calculate retention with one simple metric, but in terms of each person perspective. Again, Einstein’s relativity idea help us to solve this issue. We call this metric same day: does the user open the app (or interact somehow) in the same day as departure day of each flight?

Of course we take into account that user can add past flight or couldn’t appear in the app because his flight is in the future.

Now the starting point is the timeline of each user, but calculates very easy. Simple solution.

Hence, we re-ask the questions from what’s weekly / monthly retention to flight retention (but we still check regular retention)

Believe me or not, from that point of view, our retention grows up to 4 times, but it unveils more questions:

  1. What’s the fraction of users add their first flight?
  2. When is their first flight?
  3. Do their fly rarely or just forgot about your app?
  4. Will they remember you on the next flight?
  5. Is there any patterns within people who show up in the app?

Such questions are a second level of analyis, that brings much more insights and calls to actions.

Do you calculate your retention correctly?