This is your no-fluff, all-action guide to user retention — powered by real data and insights from a live mini app, Drawcast. Written by a mini app builder, for mini app builders. You will walk away with fresh, practical ideas to boost engagement and accelerate your app’s growth.

Would you pour water into a leaking bucket?

Probably not, right?

But what if that leaking bucket is your own mini app?

If you're seeing new users come in - amazing! That means you're getting some distribution and traction. But if those users don’t stick around, your app might be quietly leaking value. 

Like water slipping through a crack, they’re trying your mini app once and never coming back.

To build a sustainable mini app - one that not only attracts users but keeps them coming back - you have to look at user retention.

Because at the end of the day, that’s what we’re all here to build: sustainable, onchain businesses that last.

To build a habit-forming product, you need to follow the Hook Model. The model consists of Trigger, Action, Variable Reward, and Investment.

• Trigger prompts the user to act. There are external and internal triggers.

• Action is the simplest behavior performed in anticipation of a reward.

• Variable Reward satisfies the user’s desire and keeps the experience engaging through unpredictability.

• Investment happens when users put something of value into the product — time, data, effort — making it more valuable and increasing the chance they’ll return.

Drawcast is a social Drawing game, where players draw and others need to guess the drawings. Guessers earn points for correct answers, and artists earn points when their drawings are successfully identified. Click here to see it in action.

Here is how the Hook Model looks for Drawcast:

User retention refers to the ability of your app to keep users coming back over time after they first start using it. It measures how many users continue to use the app after their first visit.

For example, if 100 people open your app today, and 25 of them are still using it a week later, your 7-day retention rate is 25%.

User retention is important because:

• It shows how valuable or engaging your app is.

• High retention often leads to higher revenue, better reviews, and organic growth.

• It’s cheaper to keep existing users than to acquire new ones.

Farcaster and the Base app’s developer platform provide some usage stats for your mini app.

With the latest update, the Base App's analytics dashboard now gives you deeper insights into your mini app’s performance — including median session time, notification CTR, and more.

To access your analytics, head to base.dev, register your app, and start tracking how users engage with it.

While both give you a helpful high-level overview, to get some actionable metrics about user retention, you need to dig deeper.

Take DAU (Daily Active Users), for example—it’s useful, but it lumps together new and returning users, which can be misleading. You don’t just want to know who’s showing up—you want to know who’s coming back.

Let’s assume your DAU is dropping. Is it a user acquisition problem or retention problem? Or maybe both? How do you know what action you need to take to improve it?

That’s why you need to level up your data game. You need actionable analytics.

I started collecting data for Drawcast from day one — and it’s one of the best decisions I made.

Having an anonymous analytics layer in your mini app is crucial for understanding user behavior and improving your product. Without it, you’re basically shooting in the dark.

While I check the Base App and Farcaster dashboards daily (sometimes more than once), I often dive into my own analytics tool for deeper insights.

With the right data, you can:

• Measure real user behavior

• Identify drop-off points

• Optimize the experience

That’s how you turn a leaky app into a sticky one. Win-win for both you and your users.

You can use any analytics tool that tracks retention:

• Google Analytics (easy to set up)

• Amplitude or Mixpanel (for deeper product insights)

• Plausible (privacy-friendly and simple)

Pick what fits your style—but start tracking!

Your analytics tools should be added to the app right when it is available. You need plenty of data, and there is a great chance that you will get some users to your app when you launch.

But it is not enough just to add the tracking code to your app and call it a day.

You need to define events to track in your app, these should cover all the actions users can take in your app.

The type of events you need to track depends on your app, but here is a high-level example from the Drawcast mini app.

In Drawcast:

• Users can draw a prompt: we track if a user starts drawing, and if they submitted the drawing

• Share drawings with friends: we track shares as well

• Users can guess drawings, and we track this event as well

• If they guess correctly, they earn points for themselves and for the drawer: we track correct and incorrect guesses

• Users earn points and climb the weekly and all-time leaderboard: if they unlock an achievement, we track that

• Users can earn badges for winning the weekly leaderboards and completing daily quests: this achievement is tracked as well

• Active users are rewarded with weekly prizes, daily quests and treasure chests.

I highly recommend tracking as many actions as possible in your analytics tool, so you have a large pool of data, and who knows what you can uncover when analysing your user retention data. 

When someone opens your mini app for the first time, your mission is clear:

Deliver the core value as fast as possible.

The goal is to help new users reach the AHA moment - that instant when they get what your app does and why it’s worth coming back to. The faster they hit that moment, the better your chances of retaining them.

To improve this, you need to analyze what first-time users do inside your app:

• Where do they enter your app (if your app has different entry points)?

• Where do they click first?

• Where do they drop off?

• Where do they succeed?

But most importantly:

• What actions correlate with long-term retention?

That’s where user cohorts come in.

Because in the end, if you nail the new user experience, you’ll set the foundation for strong retention and long-term growth.

In Google Analytics, you’ll find a built-in retention chart on your dashboard. It’s useful for a high-level overview, but it shows aggregate data—all users, with no filters.

While that gives you a general retention rate, the real insights come from digging into specific user cohorts. These are groups of users who share certain behaviors, especially those tied to your app’s core value.

We create cohorts to:

• Group users who took specific actions

• Compare their retention to other groups

• Spot what behaviors likely drive better retention

Let’s break it down with an example from Drawcast.

My focus was increasing user retention, so I was trying to find data correlating with elevated user retention.

I hypothesized that retention in Drawcast depends on whether a new user:

1. Draws something

2. Successfully guesses a drawing

In the beginning, the entire user flow was designed to get people drawing first.

When someone opened Drawcast, they were taken straight to the drawing screen. The idea was that users would naturally discover the guessing feature afterward — and at the time, I believed this was the optimal flow.

While the week-1 retention of Drawcast wasn’t bad, I felt there was still room for improvement. 

Based on my hypotheses, I created the following three cohorts to identify (and confirm) retention driving activities:

• Cohort 1: All new users (everyone who opened the app)

• Cohort 2: New users who guessed a drawing in their first visit

• Cohort 3: New users who created at least one drawing in their first visit

In Google Analytics, I used these event-based filters to define the cohorts.

For Drawcast, I chose weekly retention intervals, but your timeframe should match your app’s usage pattern (daily, weekly, monthly, etc.).

These are some early analytics from the first few weeks after Drawcast launched. The dataset was still quite small, so while it’s not enough for statistically significant conclusions, it already revealed some useful patterns. Even early insights like these proved incredibly valuable in shaping Drawcast.

Cohort 1 (All new users)

Average week 1 retention for users who joined between Apr 20– May 3, 2025: 18.7%

Cohort 2 (“Players” who guessed a drawing correctly)

Average week 1 retention for players who guessed a drawing correctly (Apr 20 – May 3, 2025): 47.2%

That’s double the average retention rate. Huge!

Cohort 3 (“Drawers”: created at least 1 drawing)

Average week 1 retention (Apr 20 – May 3, 2025): 34.5%

Higher week 1 retention compared to all new users cohort (Cohort 1), but not as high as the player cohort (Cohort 2).

Even with limited data, one key insight stands out:

Users who guess drawings are more likely to come back. It kind of makes sense, since guessing has a much shorter positive feedback loop then drawing. When you draw, your drawing enters the pool but the info about correct and incorrect guesses come later.

That’s a strong insight, but remember: correlation ≠ causation.

Based on the retention data of different user cohorts, the next step was to optimise the new user experience so they perform the core, retention driving actions the easiest and smoothest way possible.

In Drawcast’s case, get more new users to start guessing as soon as possible!

A funnel report can be helpful to see:

• where users end up going once they open your app

• What is the success rate to perform key actions and identify bottlenecks in the process

In the early version of the app:

• New players landed on the “Draw” page, funnelling them to start drawing

• Guess page UI was suboptimal:

  • Back in the early days, on the “Guess” page, we listed all drawings in a long list. Based on user interviews and new user live testing, it turned out the list of drawings caused indecision and was confusing, so I eliminated the long list and created a single button so people can start guessing

Conversion rate of a new user opening the app and guessing a drawing was around 45-55% in May, 2025.

My goal was to increase the share of new players who make guesses, with the expectation that this would also increase retention. So I made some changes to the Drawcast app:

1. I made the “Guess” page default, so new users opening the app can start guessing

2. Removed the long list of drawings and turned it into a single button (one key action, clean interface)

After the changes, around 69% of the new users were guessing in the first session! That's a 25-53% increase compared to the old version.

Did it increase week 1 user retention?

After implementing the changes, average week 1 retention increased from 18.7% to around 20–21%. However, it’s difficult to draw strong conclusions since I’m comparing different user cohorts over time, and the dataset is still quite limited. 

Beyond the numbers, I ran live beta tests. The feedback was overwhelmingly positive — users found the app much easier to navigate, and confusion about where to click dropped significantly. As a result, a much higher share of users started guessing, meaning more people reached and experienced the core value of the app.

Fast-forward 5 months later (November, 2025)

Drawcast’s daily active users increased a lot, from 50-60 DAU (daily active users) in May 2025 to 300-450 DAU in November 2025. Week 1 retention has fluctuated quite a bit, 5-6 months later it is between 15–19%.

I also compared the week 1 retention rates of the same user cohorts from May to November:

• Cohort 1 – All new users (everyone who opened the app):
  May: 18.7–21% → November: 15–19% → Slightly lower week 1 retention.
  Cohort 2 – Users who guessed a drawing correctly in their first visit:
  May: 47.2% → November: 19.7% → A significant drop, though the November dataset includes roughly 10× more users, which makes the result more representative.

• Cohort 3 – Users who created at least one drawing in their first visit:
  May: 34.5% → November: 27–33% → Retention decreased here as well, but remained relatively stronger than the overall user base.

Interestingly, those users who draw and guess in their first session the week 1 retention rate is 32.67%.

At first glance, the drop in retention might seem like a step back — but context matters. Over time, Drawcast reached a much larger and more diverse audience. In May, retention data came from a few hundred early adopters who were naturally more curious and motivated to explore a new app. By November, the user base was 10x larger, with many casual or one-time visitors included.

This kind of decline is common when a product scales — as you grow beyond your most engaged early users, the average retention rate typically drops. 

The retention-focused improvements weren’t wasted effort — they made the product easier to use, improved the user journey, and created a more scalable foundation for growth. Without these changes, retention would likely have dropped much faster as new users joined.

Running retention-focused experiments is something you need to do continuously. So I didn’t stop there. 

What if, besides the joy of guessing and drawing, we added a little win factor to the mix?

And what’s better than winning some tokens?

The assumption is that this combination of instant gratification and core gameplay engagement will further boost retention.

This is a new retention experiment I’m running: after guessing or drawing a random number of times, a treasure chest appears, letting users claim a random amount of tokens.

But here’s the twist for new users:

First-time players see the treasure chest much sooner — just once, during their first session. That means they quickly experience both the fun of the game and the excitement of earning their first reward.

After a few weeks of running the treasure chest feature — and ensuring it appears early for new users — I’m now seeing a 38-41% week 1 retention rate among users who claim a treasure chest. That’s more than double the average retention rate for all new users, and significantly higher than the other recorded cohorts.

The takeaway is clear: the ideal onboarding flow in Drawcast for a new user looks like this 👇

1. Guess at least three times (ideally correctly)

2. Create a few drawings

3. See and claim the treasure chest reward

This sequence seems to create the strongest early engagement and retention loop.

It is still very early to drive conclusions, but the results so far are very promising!

So am I winning? What’s the take away?

• Extract actionable insights from your data — look for patterns like “users who do X are more likely to come back.” Then, optimize for that action.

• Identify your key retention drivers. Figure out which actions best predict long-term engagement in your app.

• Optimize the UX to make it effortless for users to perform these actions. Brainstorm and test ideas to improve the early user experience.

• Experiment relentlessly. Keep testing new features, tweaks, and flows.

• Treat data with caution. Early data is limited and often compares different user cohorts over time — don’t over-interpret trends.

• Run live beta sessions with new users discovering your app on their own — it’s the fastest way to see where they get stuck or drop off.

There’s no silver bullet for improving retention — but the goal of this guide was to share the mindset you need to approach it effectively.

In the early days, it’s easy to obsess over user acquisition, which is important to get things off the ground. But once users start coming in, your focus needs to shift to retention — otherwise, you’re just pouring water into a leaking bucket and wasting precious resources.

I hope this gives you a useful starting point to better understand your users, increase retention, and build mini apps that people genuinely want to come back to.

So dig into your analytics. Talk to your users. Learn from what you see — and keep iterating.

Let’s bring people onchain — and give them a reason to stay onchain.

Huge thanks to my friends, Gabor Zold and Jonathan Colton for brainstorming and reviewing this post!

Last updated: 11 November 2025

Sources

• Hooked: How to Build Habit-Forming Products by Nir Eyal is an incredibly valuable read for anyone building products. I highly recommend picking it up and applying its principles to your own product — it’s truly a must-read!

• Hook Model in short: https://www.productplan.com/glossary/hook-model/