Retention is King..?

So, if retention goes up by 10%, how much does revenue increase?

Ever since Andrew Chen's post, Retention is King, many people started paying close attention to retention. At the time, many were so obsessed with User Acquisition that the post served as a wake-up call against relatively downplaying the impact of post-install churn. On a smaller scale, it was like saying we should care more about preventing churn within a feature funnel rather than just focusing on the funnel entry rate. It was even more convincing because it proved this point through simple formulas and examples.

Reading his post, it's easy to understand why retention is King in terms of DAU. Honestly, you don't even need a formula to understand it intuitively. If you have a bottomless jug, and you have to choose between plugging the hole or pouring in water, you'd obviously pick the former for having a bigger impact..

Then what about from a revenue perspective? Say our subscription conversion rate went up by 10%, but retention dropped by 10%.

What would be the rational decision to make? Rephrasing this question looks something like this.

If retention goes up or down by 10%, how much does revenue go up or down?

When retention goes up, LTV goes up.

LTV is the user's average Lifetime x ARPDAU. ARPDAU means the average revenue per DAU. In our current assumption, there is no change on the monetization side, so ARPDAU is a fixed value. Only the Lifetime changes according to the change in retention, and this leads to a change in LTV.

Lifetime is made up of the sum of retention.

If 100 people launch the app, 40 launch it the next day, 25 launch it the next... and so on,

if the total sum comes out to 300%, it is numerically equivalent to all 100 initially acquired users launching the app for 3 days without a single person dropping off. Thus, in this case, the Lifetime becomes 3 days.

Here, if D1 retention goes up by 10%, mathematically all the following retentions will also go up by 10%. The retention rate from D1 to D2, or D2 to D3, hasn't changed at all; only the retention rate from D0 to D1 has increased by 10%, so that exact increment is carried over to the rest.

Since the change in D1 retention rate cannot affect the number of D0 users, the total Lifetime increment, including D0, lands at around 7%, which is smaller than 10%. What if the existing Lifetime is 30 days instead of 3 days? A 10% retention increment leads to a 9.7% Lifetime increase. The longer the existing Lifetime, the smaller the proportion D0 takes up in the total, allowing the D1 increment to carry over almost exactly as it is.

To summarize, the user's Lifetime lengthens in proportion to how much retention goes up.

At this point, since ARPDAU is fixed, the longer the Lifetime becomes, the larger the expected revenue (LTV) generated will be.

If retention goes up, how much does revenue go up?

We've observed above that LTV goes up in proportion to the retention increment. Here, the absolute 'margin' of revenue increase varies greatly depending on the figures below. Assuming LTV increased by 7% due to a 10% retention increment, let's see how the margin of revenue increase changes according to the following figures.

a. Number of new user acquisitions
b. Existing Lifetime
c. Revenue per DAU

a. Number of new user acquisitions

For a service with 300,000 new users acquired a month, the expected revenue difference from a 7% LTV increase is 3 million won.

On the other hand, for a service with 1,000,000 new users acquired a month, the expected revenue difference becomes 10 million won.

The concept is that the more active a service's new user acquisition is, the larger the absolute revenue increment caused by the retention increment.

b. Existing Lifetime

As explained earlier, the longer the existing Lifetime, the more you can take the D1 retention increment size straight into a Lifetime increase (because the impact of D0 becomes negligible). Not only that, but even with the exact same 10% increase, the margin of Lifetime increase itself differs between a product with a 5-day Lifetime and one with a 30-day Lifetime. (+0.5 days vs +3 days)

Therefore, you could say that the longer a service's existing Lifetime, the larger the absolute revenue increment caused by the retention increment.

c. Revenue per DAU

The same goes for the revenue per DAU, which we kept treating as a fixed value. The larger that value, the larger the absolute revenue increment caused by the retention increment.

Depending on the product's situation, this margin of increase can vary wildly. A 10% retention increment could boost monthly revenue by up to 50 million won, or it could just stop at the 100,000 won mark. What's certain is that you'll only know if you plug in your own product's current numbers and calculate it yourself.

Summary

1. If retention goes up, Lifetime goes up.

2. If Lifetime goes up, LTV per user goes up, so the total expected revenue goes up.

3. Here, the margin by which revenue goes up depends on the average number of user acquisitions, the length of the existing Lifetime, and the revenue per DAU.

Let's go back to our initial question.

When our subscription conversion rate goes up by 10%, but retention drops by 10%, what would be the rational decision to make?

We can now calculate the margin of expected revenue decrease caused by a 10% retention drop. If that decrease margin is 40 million won a month, we just need to check if the expected revenue increase margin from the 10% subscription conversion rate increment exceeds 40 million won.

Retention is King…

When looking at it from a revenue perspective, retention might be King, or it might not be.

Is retention King in our product?