Why is retention important?
Retention is the most effective and meaningful lever to drive growth. Good retention is a strong signal that there is PMF (product market fit.)
Good retention also signals a higher LTV (lifetime value) of your users. Higher LTV then allows you to power more acquisition channels.
Another way of thinking of retention is that it is the answer to -- "How many users are getting enough value to keep coming back?"
If the answer is none (or a low number), then you are failing at the most essential truth of creating products -- i.e. creating long-term value for users.
What is retention
Retention happens when users repeatedly do a meaningful action over a defined time period.
Let's take an example.
Let's consider Spotify's free users. One of the most meaningful actions on Spotify is to listen to a song (or podcast) for at least x seconds.
Let's assume that 100 users listened for >= x seconds today. So, the W0 (week 0) retention is 100%.
Then, we look at the same 100 users and determine what % of them came back in the next week and listened for >= x seconds. Let's say 80% did. So, the W1 (week 1) retention is = 80%.
Then in W2 (week 2), repeat the process for the same 100 users from in W0. Let's assume that only 75% came back.
Keep repeating for multiple weeks, and you will get a curve like this
In the Spotify example:
- "Listening for >= x seconds" is the meaningful action.
- "One week (7 days)" is the defined time period.
Please note this is just an example. The meaningful action could also be other user actions like opening the app, creating a playlist, doing a search, etc.
Similarly, you could measure retention on a daily or even monthly basis. It is essential to define the right meaningful action and time period.
Whenever I think of retention, I answer the following questions to identify the meaningful action and the time period.
- What action can the user take to get the most value?
- Given the user's needs, how often should they do the action?
To answer these questions well, I highly recommend deeply understanding user's needs and the product's value prop.
My rule of thumb 👍 : If I can't find a good answer to question 2, I start with monthly retention. It is because, I feel that if users do not use the product once a month, there is something wrong.
There are two more types of retention curves that are not typical, but still exist. Both of them look similar, but have a very different story.
Retention curves that go to zero (for some time)
There are some products where retention goes to absolute zero. But only for short periods of time.
These products are very popular to begin with, but then one fine day they are not. And they die.
To put this into perspective, think of games that became extremely popular when they launched (like Farmville on Facebook.) But then disappeared all of a sudden.
Zynga, the creator of Farmville, created many more smash-hit games after Farmville. And, as you can see below, all of those games were bigger hits than Farmville.
So in the long run, the retention curve for Zynga users might look like this
Every spike in the graph would typically correspond to the launch of a new game.
So, when retention curves go to zero in the long-run, it is not always bad. You need to interpret retention curves based on the right context, which includes the type, maturity, and stage of the product.
As we see in Zynga's case, this is just the way things work for them (and other products like them.) They produce one hit after the other, and see retention dropping to zero for older games, but then sharply increasing for new games.
In other cases, like streaming, social media, etc., these types of retention curves are bad. Think of Orkut, Friendster, Blockbuster, Napster, etc. Their retention touched zero, never recovered, and the products permanently died.
It's interesting to note that this type of retention signals a different type of user behaviour -- users get bored after a while, which makes them use the product lesser and lesser, until they stop using it altogether.
Whereas in the case of flatlining curves, the user behaviour is different -- there is a percentage of users who like the product so much that they keep using it even in the long-term.
Retention curves that flatline, but then increase sharply
These curves are also called "smiling" curves, because they look like a smile.
First, these curves are downward-sloping. But, over time, they move upwards again, and a smile occurs
Theoretically, this happens because the product offers more value than before or more users are get value than before. This could happen because of three reasons:
- Network effects: when there are more users using the product, the product becomes more valuable for the existing users. And that is why the flattened curve starts moving upwards. Airbnb is a great example. With an increasing number of hosts signing up, you see more users booking on Airbnb. Whatsapp and Instagram are two other examples -- when users see more of their friends using the products, they also start using it more.
- New use cases / categories: users start using the product more often when they see new categories or use cases that become available to them. A few examples: Uber experienced increased usage after they launched food delivery, grocery delivery, etc. Google Maps: would have seen a usage increase when they broadened their value prop from being an "only navigation" app to also being a "business discovery" app.
- New platforms: this happens when products launch on newer platforms like Android, iOS, Web, etc. Every launch leads to an increase in retention, pushing the curve in an upward direction. Instagram and Clubhouse were iOS only when they launched. And saw a great surge of users when they launched on Android.
Liking this post? Get the next one in your inbox!
What is good user retention?
A typical retention curve (like the above) will trend downward in the initial periods. But over a more extended period, it will flat line above zero and will not decrease any more.
That is precisely what you want -- a curve that flatlines above zero.
(There are two more types of retention curves, which I will talk more about in the next edition)
The next question is: when should it start flatlining, and should it be at 25%, 30%, 50% or some other number?
Short answer: The sooner it starts flatlining—the better. The higher the level—the better.
Long answer: it depends.
When it starts flatlining depends on the kind of product, user needs, competition, and many other factors.
Similarly, the level at which it starts flatlining depends on multiple factors. Luckily, there is enough industry data to create good benchmarks for common industries.
- Consumer Social: ~25% is GOOD, ~45% is GREAT
- Consumer Transactional: ~30% is GOOD, ~50% is GREAT
- Consumer SaaS: ~40% is GOOD, ~70% is GREAT
- SMB / Mid-Market SaaS: ~60% is GOOD, ~80% is GREAT
- Enterprise SaaS: ~70% is GOOD, ~90% is GREAT
(Sourced from Lenny's Good and Great Retention research.)
How can PMs increase retention?
A few caveats before answering:
- Retention is very context-specific, i.e. what works for one might not work for others
- Increasing retention is always challenging. It takes a lot of trial and error.
- The below pointers are generic. They will help you think of the right things. You can create concrete action items once you start thinking of these things.
Maximise user value
Build your product to deliver maximum user value. Make it easy and quick for users to get value. If you're in a competitive market, deliver it better and quicker than others.
The best way to do this: build user empathy - learn about your users, their needs, and their problems. Then, solve them in the best possible manner.
Example: Spotify creates personalized playlists through its Discover Weekly and Daily Mix playlists. They help users find relevant songs easily and quickly. This makes the user feel Spotify understands them, and users perceive more value.
Make value explicit
Explicitly show users the value they will get from your product. Remind them if they forget.
The best way to do this: try to deliver value as often as possible, so users remember and appreciate it.
Example: As soon as you sign up, Dropbox shows users value like easy file sharing, secure storage, and cross-device access. It also updates users about new features and how users can benefit from them.
Give them so much that they fear giving it up.
If you deliver a lot of value very often, users will not want to give it up. And that is what you want.
Example: In one membership, Amazon Prime offers so much value -- free shipping, music, video, and exclusive deals -- that users find it hard to cancel their subscription.
Don't let them go away.
Identify users who are not coming back often or are about to leave the product. Find ways to engage them -- give them a reason to stay.
The best way to do this: look at past users who did not retain, identify their behaviour right before leaving, identify others who are displaying that behaviour today, and encourage them to change their behaviour so they don't leave.
Example: Netflix predicts when users might be losing interest. Then, they show personalized recommendations and send emails about new releases similar to the user's viewing history.
Win them back after they've left.
Remind users who have left your product about the value you provide. Help them understand how your product solves their essential problems. And how you do it better and quicker than other alternatives.
Example: when users stop using Grammarly, they send emails showing recent improvements that can help users write more effectively. These messages remind users of Grammarly's value, specifically focusing on new features that might address past reasons for leaving.
Implementing some or all of these strategies will help you increase customer retention.
The main learning that you take away from this article should be -- focus on delivering value to the customer throughout the journey. And ensure that they keep getting that value in the long term.