Playbook: the first step in the quest to build world-changing products is measuring retention
Retention refers to continuous usage, which means your solution keeps generating value over time.
The value created can be:
- Active, like when you send a message in Slack to a teammate
- Passive, like using the Twilio API to send a message to verify a new sign up
If you're building a subscription business, you need to win your customers every year and the easiest way to do it is by making sure they keep using your product.
When your business retains its users, your revenue compounds. That's why understanding and prioritizing retention is key for long-term success.
Here’s a famous chart from Slack’s S-1:
In this chart, you can see their customers from each year from 2015 to 2019. Notice how each segment starts small and gets larger over time?
That's retention - having customers stick around, which in turn generates more revenue as they grow and upsell new product lines.
Every dollar of revenue Slack closed in 2015 compounded in the next few years. In January 2019 their Net revenue retention was 143%.
Table of contents
- Why retention is a measure of product-market fit
- How to measure retention
- How to calculate retention
- Step 1
- Step 2
- Step 3
- Retention curve
- Ideal frequency
- Common mistakes measuring retention
- Retention benchmark
- B2B Saas
- Other verticals
- Common retention killers
- Your activation is garbage
- Example: N26
- You don’t integrate into an existing behavior
- Example: Superhuman
- You don’t help users create a new habit
- Example: Calm
Why retention is a measure of product-market fit
Retention is the way to know that you are delivering value to your customers.
If you don’t know whether your company has product-market fit or not, you don’t have it.
At this stage, retention should be the only thing you care about. User retention is often the north star metric for subscription businesses as it makes it easy to predict churn.
In the early stages of your company, retention is even more important than revenue. If people use your product regularly, you have something special since they’re getting repeated value from it. You can’t hack this.
If your company has reached market fit, as in you have a clear persona, use case, retention and go-to-market strategy, then monitoring user retention ensures that the early set of customers don’t leave you as your product evolves. It also helps you ensure that the new signups keep retaining at the same rate.
Something counterintuitive about retention is that if you grow your number of new customers, the retention rate tends to stay the same. So once you figure out how to achieve high retention, you can pump your acquisition through paid channels and generate predictable returns on your ad spend.
How to measure retention
All you need to start measuring and optimizing your retention is a base of users, no matter how small.
What is Retention?
Retention is the measure of how many users return to your product over time.
It’s the percentage of active users at any time after X days of signing up or installing your app.
Why do you need to care about user retention?
Retention impacts every business metric that you care about.
If your retention graph looks like the first chart you’re default dead.
If your retention graph looks like the second chart you’re default alive.
Without retention, your product is a leaky bucket. You can pour as many dollars as you like into marketing and still wind up with no long-term users.
How to calculate retention
There’s only one way of calculating retention. Here’s how to do it in 3 simple steps.
Take the number of users that sign up in the week (or interval) you want to analyze. This is a user cohort.
Then calculate how many users came back in the weeks after. In week 0, you start with all 120 users who signed up in the cohort of 2nd to 8th November. In week 1, 90 of the first 120 users came back, in week 2, 80 of the first 120 users came back, and so on.
Convert the table to be the percentage of users in the cohort coming back.
Repeat this operation for all of your cohorts.
Then take all the columns and average them out and you should get your average retention.
If you take the average retention we calculated above, you can draw a retention line chart.
When the ideal frequency of usage is weekly your should measure the 12 weeks retention
When measuring retention you need to answer:
What's the ideal frequency at which your users should use your product?
You need to figure out what makes the most sense for your business. Usual answers are: daily, weekly, monthly, or yearly.
To help you understand what frequency of usage makes the most sense for you to measure, here are some examples:
Is your product something that should be used as frequently as a smartphone, a washing machine, tampons, or a Christmas tree?
If it’s a one-off like a coffin, then retention is a useless metric to you.
Social networks - you check these apps daily, as there’s always new exciting content. Communication software and productivity tools also fall into this category.
Real life example: Whatsapp. If you’re a chat app like Whatsapp, you should care about daily retention.
Food delivery, ridesharing, grocery shopping, analytics tools, and marketing software fall into this category.
Real life example: Pinterest. It makes sense for users to visit Pinterest weekly to discover new things and look for inspiration. So Pinterest chose weekly active usage (WAU) as a measure of retention.
eCommerce stores, finance, HR, and payroll software all fall into this category.
Real life example: As an employee, you won’t be expensing things every day, but you’ll want to file all your expenses once a month.
Travel, insurance, and tax software are all categories of products that you’ll likely want to be used yearly. When your retention period is this long, you want to rely on more qualitative data like reviews and surveys to measure customer satisfaction, as you can’t wait around for years to fix the problems in your product.
Real life example: Airbnb is the place you go when you’re planning a holiday. So it makes sense for them to measure retention year after year.
Common mistakes while measuring retention
In the graph I showed you earlier, all users were averaged together while measuring retention. Doing this is also not a good practice because you lose a lot of granularity, making it difficult to understand how things are trending over time.
The best practice is to create daily, weekly, or monthly acquisition cohorts and measure their retention separately.
This way you’ll be able to understand if the new features you’re building are causing users to retain more or not. If there’s a regression in your retention, you want to learn about it early on to fix it along the way.
A little known fact about Facebook’s daily retention is that it has been over 40% across cohorts over the years. This means that if you looked at the users who signed up to Facebook in June 2010, 40% of them would’ve used the product yesterday.
The good news about measuring retention is that numbers are comparable across companies.
Each vertical and type of business will have different targets when it comes to retention.
B2B SaaS has some of the highest retention rates.
Depending on the company size of your customers, you’ll have different target retention rates.
For B2B products the retention you’ll want to measure will be at an account level.
Employees at companies come and go, you want to measure if companies keep using your product to solve a problem, not who’s using the product within the company. Someone getting promoted or leaving a job shouldn’t impact your metrics.
Each user is worth less than B2B in consumer SaaS, but the addressable market is larger.
Most consumer SaaS tend to be fully self-serve products. Consumer SaaS also has more price sensitivity than B2B SaaS and expanding revenue is more difficult.
If your retention is less than 20% you should talk to your users, and either pivot or build a different product.
If you’re between 20-40% retention you’re in a weird position because even though achieving 20-40% retention is difficult, it’s not high enough for product market fit. While it is possible to move beyond this threshold, you might spend more time than necessary on something that doesn’t work.
For other verticals like travel, e-commerce, or on-demand delivery companies these are the retention benchmarks:
Common retention killers
Building something that retains users is art, not science.
There are some common problems that can kill a great idea.
The activation of your product is garbage
If your retention drops below 20% after the first period then you’ve got an activation problem.
This might be happening for three reasons:
- What you’re selling is not what you built, so users don’t get the value you promised them
- Users can’t figure out how to use your product because it’s too complex
- You built a fidget spinner product, it’s a gadget that is cool only for the first 10 minutes (like a VR headset)
You can solve the first two problems through iteration.
The third one requires some education on why there is pain in the first place. For example, meditation apps weren’t something mainstream a couple of years ago, so education around the benefits of meditation was essential for this category of apps to grow.
If education doesn’t work, consider pivoting.
When the team was building the onboarding at N26 they learned that successful onboarding should consist of:
- The user understands why they’re providing information
- There is a sense of progress as they go through each step of it
- The user gets some value at the end of the onboarding process - In N26 case they’d get an IBAN, and after a few days a debit card
The team decided to build the onboarding around those three rules, at the expense of the time it took for users to complete the onboarding.
The most educational onboarding the team built had the highest completion rate and the users perceived it as the shortest one.
Not integrating into an existing behavior
If there are existing workflows users already have, consider building something that integrates with that workflow.
If you get smart, you can make an existing workflow less repetitive, delightful, and more powerful.
The main advantage of this approach is that you’re not changing any user behavior, you’re just making a workflow more seamless.
As part of the Superhuman onboarding, the onboarding specialist will make you drag the Superhuman app into the same position where your mail app is.
The reason why Superhuman does this is that we all build up muscle memory for where our apps are. So deliberately introducing new apps into an existing workflow can greatly increase retention.
Not helping users create a new habit
Helping people adopt a new habit requires some intention and education and then some proactive re-engagement (emails, push notifications, etc...)
You should build features in your product that keep re-engaging users with no effort while delivering value at the same time.
Examples of such features are email digests, notifications for in-app activity, and Slack bots.
When Calm was starting out they were struggling with their retention. One thing they found though was that 5% of their users, the ones that set up a daily meditation reminder, were retained at more than 50%.
After learning about the correlation of daily reminders with retention the team at Calm tried adding the daily notification feature to the onboarding flow and retention shot up for all users.
The daily reminder feature was something an engineer added for themselves and was deeply nested in the app settings. By understanding user behavior Calm was able to turn a secondary feature into a main part of the onboarding.
Now daily reminders became a core part of the onboarding of every fitness and meditation app.
There’s only one important takeaway from this article:
Measure your cohort retention. It is your most important metric
If you’re a consumer app aim for achieving or maintaining a retention rate of above 40%. If you’re a B2C subscription, aim for over 50%.
If you’re a b2b company depending on the company size of your customers aim for >60% SMB, >70% mid-market, >80% enterprise.
Once you get there you can start stacking those users and in a couple of years, your company’s S-1 will resemble the chart at the beginning of the article.