Product Retention – How to Measure & Improve

Having a well-retaining product is the key to the longevity and success of every company. In the old days, retention was about buying the same physical product over and over; like newspapers, disposable razors, tissues, etc. In the digital landscape, the concept had to adapt and improve – it was no longer about selling something physical, but about grabbing the attention of users long enough to generate revenue. Keeping users engaged for a long time is the important part and how you generate the revenue is the only difference between digital companies. The most dominant strategy in the market is selling subscriptions; but it can also be showing ads, bringing out sequels, and selling in-product benefits.

We investigated in our previous post the basics and different types of calculating retention. In this post we will go over the approaches to managing retention, setting goals, and improving. All the different types of generating revenue are a whole other conversation that will happen later.

On-day retention

This type of retention is excellent at understanding the longevity of your product. Maybe you are selling a 12-week exercise program; then you would be interested in how many people make it to the end of the 12 weeks and where are the biggest drop-off points. Or maybe your product is something that the users will spend many months, if not years on. Then we would want a certain number of days where on-day retention is stabilized and should not decrease.

This rule-of-thumb of measuring longevity is easy to apply as well. The first question that you, as an analyst should ask the product team is: “how long do you expect people to use this product”? The answer to this is going to anchor all our investigations. If our retention drops on Day-30, do we really care if we are selling a product that will fill its purpose by Day-14? Well, maybe we do but it is definitely not as alarming as a serious drop on Day-7.

There will also be many cases where the answer to your question about product life is “forever!”. Especially a prominent trend in gaming, this is when we want people to keep coming back to our product week over week, month over month; and ideally, purchase things here and there. Because you can’t purchase something if you don’t even have the app open, right?

To manipulate and increase on-day retention, the strategy is giving people a reason to come back to your product. A notification here, a reminder there, a super-mega discount promotion thrown in are all great and useful, but the best way to bring people back is content. Adding a limited time content for your website, or dropping a season every month for your game is the most successful way of getting people to visit on certain days. In fact, the most successful method may be daily login rewards, if applied correctly.

Period-over-period retention (PoP)

Our second type of retention is about measuring the length of your product loops. Say, you have a weight-tracking app and want people to check in every Sunday with their measurements. Outside of that, the users can submit their weight, but to make sure our tracking is more accurate we don’t put that much weight into that. As a result, our product loop consists of a user taking measurements, submitting them on Sunday, and repeat in 7 days. Week-over-week retention is very important to us, as that is the frequency we can expect our most dedicated users to use our app.

Similarly, we can add daily quests to our favourite online game and suddenly that is a very fast game loop that we can expect players to do. Especially if it is something rewarding and fun (not an easy balance by any means), we can introduce a core loop of “log in after work, play for an hour, log out”. An hour every day is sometimes much better than 15 hours in a single sitting and experiencing everything.

Maximizing period-over-period retention is much more difficult to do than on-day retention. This is the true measure of the core of your product, and the effort to influence it needs to be much higher to have a lasting effect. Seasonal events are a common strategy, but a lot of companies fail to implement them correctly. A Halloween event is guaranteed to improve month-over-month retention in October & November. And we can make it even better – how about a high-score system for a special game mode? Now we give the users a reason to come back next year in October as well.

The challenge in period-over-period retention is, to impact long-term PoP, the changes you make are mostly permanent. You introduce a new system or a new feature, and you expect the feature itself to create a good and frequent loop. Taking a fitness app, for example, let’s say we added the ability to measure your calorie intake along with your workouts. Suddenly this new feature makes people visit us multiple times every day to track their login, instead of 3 or 4 times a week as our initial loop intended. Once you introduce such a system, you can’t take it back and remove it – it has to stay; needs to be maintained (and needs to be balanced if it is a game system). Even charging for it after a while is usually frowned upon. This is a higher risk in terms of development costs, but a higher reward to guarantee high long-term, period-over-period retention.

Spoilers: The advantage is, the changes made for short-term period-over-period can cover optimising on-day-retention as well!

Connecting the two

The two ways of measuring retention sometimes overlap, and for most of these cases, the common point is high-quality features. Any short-term loop we introduce is going to be beneficial in creating good on-day retention. However, making an impact in the long-term for both types of retention usually needs a much more complex system or feature. Imagine you have a budgeting app and want people to connect their bank accounts and track their spending easily. You gather ideas and need to decide between introducing credit card tracking, or automatic loan tracking. Loans are much easier to track for people as they are the same amount every day of the month, but credit card is difficult to budget around for our algorithm. If we go for the loan tracking, we can expect a small increase in long-term retention maybe, but credit card tracking is going to make people check our app as frequently as they use their credit cards. We not only created daily loops but a monthly one on the payment days of the card. This is not to say it is the perfect choice to build – the cost of building it is also an equally important factor.

Conclusion

There is a very good reason why there are two different ways to measure retention. They both track something vital and need to be regularly checked for the longevity and core loop health of your product. We will see many examples of this in future case study posts, and if there is enough interest a further breakdown of the most common strategies in the industry to increase retention – let me know by liking, commenting or sharing!