The Case of the Disappearing Users During Onboarding

Few things deflate a growth team more than watching the users they’ve worked so hard to engage slip away during onboarding and activation.

Product Growth Advisor Romain Finot knows this all too well. In today’s How I Amplitude, Romain shares the activation strategies and workflows he developed at Finary and Cyclofix to keep users engaged through onboarding and retain them for the long term.

Here are his strategies.

The disappearing users

In my nine years in growth working with e-commerce, mobile, and web products like Tediber, Finary, and Freetrade, my focus has been simple: leverage product analytics to build for growth better.

Many companies struggle with user drop-off during and right after onboarding because they're looking at the wrong metrics. By focusing on finding your product's “magic number” and building targeted activation workflows around it, you can dramatically improve user retention.

The three key steps to do this are:

1. Set a strong tracking foundation
2. Chart the right metrics
3. Activate your data, activate your users

Set a strong tracking foundation

Everyone—from product to marketing to engineering—needs to understand what you're tracking in your product, how it's tracked, and why it matters.

My approach for strong tracking foundations has four pillars:

1. Event types
2. Event naming
3. Event listing
4. Sharing

Event types: User actions vs. user properties

The two main types of events we’re going to look at are user actions and user properties. I call them .track events and .identify events in my schema.

• .track events capture what users do, like clicks, submissions, and other interactions that happen in your product.
• .identify events capture who users are, like the entries and updates to name, email, user ID, etc.
  • I also always use .identify events to track the user’s state within the product so I can quickly answer questions like “Have they finished onboarding?” and “Have they seen their aha moment yet?”

You can understand a lot about how users engage with your product just by observing these two event types. For example, you can use .track events to create the onboarding journey, then add a layer of .identify events to show which different user populations are taking which actions.

You might be surprised to see who your most successful users really are.

Event naming: Be clear and consistent

The most important thing about naming is once you land on a naming convention, you must stick to it.

I’m a big fan of the object-action naming convention, e.g. “button_clicked,” “signup_completed,” “payment_processed.” It’s straight to the point and helps non-technical people understand what the events are.

Notice the past tense, too—there’s no confusion that the user has done the action.

The same goes for event properties and user properties. Differentiate and be specific about the object associated with the property so you don’t get mixed up if two events have the same type of property.

Event listing: Pick the metrics that matter

Figuring out which metrics matter for activation can generally be thought of with two questions:

• Question #1: What represents the core value you create for users?
• Question #2: How frequently do you need to deliver this core value?

For example, AirBnB’s value is bookings, and they optimize their flows knowing someone will use their platform basically once a year. Uber’s value is rides, and their frequency is weekly. Instagram’s value is measured by daily active users.

Once you have your core metric, you can build out your ideal actions, your flows, and list in important information you want to know about your users. They should all represent value.

Sharing: Communication is key

To help keep track of all these events, I use a dynamic taxonomy Notion template that you can access here.

I start building this at the beginning of all feature development, from the design stage through working with devs. This is my insurance that when we ship features, events are being instrumented at the same time, and we’ll have data flowing in from the moment we launch.

This also leads to better data-driven discussions and data-driven decisions.

Chart the right metrics

With your data tracking foundation set, it's time to build the charts you’ll need in Amplitude Analytics. I’ll use a hypothetical example of a music streaming app looking to boost its user activation.

Personally, I like to keep things simple with four primary charts on my dashboard, plus an optional one:

• Chart 1: Acquisition
• Chart 2: Retention
• Chart 3: Revenue
• Chart 4: Funnel analysis
• Bonus Chart: Engagement matrix

Chart 1: Acquisition

This is where you keep an eye on users entering your onboarding flow. I’ve set this up as a segmentation chart tracking the daily User Sign Up event over the last seven days. If your core value is delivered weekly, you’d want to track this weekly and look at it over a four week span.

Chart 2: Retention

By analyzing how different user behaviors correlate with higher retention rates, you can identify the aha moment or use case that will keep users coming back.

For example, this segmentation chart shows weekly retention for the last twelve weeks. I’m tracking both:

1. all users (blue line)
2. users who have played three or more songs in the last thirty days (green line)

Why three or more songs in the last thirty days? It took some digging to pick that (see chart 5 below). For now, though, we can see that it does indicate an increased likelihood of retention in the longer term, which means it’s a strong aha moment.

One thing I love doing here is to actually pull out a list of the users who fit this criteria and create a cohort. These are your core users! Just click on the segment line and select “View Users” to see them or create a cohort.

Chart 3: Revenue

This shows the monetization of your onboarding—after all, you onboard users to make revenue. I create a segmentation chart and select the revenue event that indicates a successful onboarding, like purchasing a song or video. Then I look at the price property for the event.

To get monthly revenue numbers, I change the “Measured As” selector to “Sum of Property Value,” and voilà!

Chart 4: Funnel Analysis

This one is super important for telling you how onboarding and activation flow is going. If you don’t get this part right, the rest of the product won’t matter.

To set this up, I built out a Pathfinder chart by selecting the end event of activation (purchasing song or video), segmenting for new users, and seeing what events took place beforehand.

I then create a funnel chart to see dropoff more clearly.

Bonus Chart: Engagement matrix

You don’t always need an engagement matrix on your dashboard, but it’s super helpful for digging into your numbers. This chart type allows you to compare the frequency of engagement activities by different users, like the number of songs played versus the daily active user percentage over a certain timespan.

Here we can see that the average active user plays about three songs in their first 30 days. We can reason then that if a new user also plays three or more songs in their first 30 days, they’ll become a daily active user—i.e., be retained! This is how I picked the aha moment in the Retention chart above.

Activate your data, activate your users

So now at this point, you have your metrics and you have your charts—what do you actually do with those to improve your user activation? The steps:

1. Lock in on your magic number
2. Create targeted user segments
3. Deploy action-based messaging
4. Build a data-driven comms flow

Lock in on your magic number

Your company’s magic number is the specific threshold of engagement that signals when a user is likely to become a long-term customer. Some people also call this a North Star.

For example, Facebook’s North Star was for users to add 7 friends in 10 days. For Slack, it was sending 2,000 or more messages—and to activate their magic number, Slack made their first 2,000 messages free.

Your magic number can almost always be found in one or more of these key phases of activation:

1. Setup: How long until a user experiences your product? Some products like Shazam are instant, while banking apps might need 15 steps.
2. Aha Moment: When do users first get value from the product? Think Shazam showing you the song name or getting your first real-time payment notification in Revolut.
3. Habit Formation: How often does someone need to have that aha moment to stick around?

Once you define the magic number, tell the rest of your organization right away and make it a core goal of our onboarding flow.

Create targeted user segments

Use .identify event properties to segment users based on their stage in the product journey for more targeted, personalized messaging. Segments can include:

• Newly signed up users
• Users who completed setup
• Users who experienced the aha moment
• Users who formed a habit
• Premium vs free users
• B2B vs B2C users
• Account size segments

To implement this segmentation, I work with my tech team to set up boolean properties in the .identify events. For example, I’d create an “is_activated” property that’s either True or False.

Deploy action-based messaging

Instead of sending generic outreach that may not match users’ needs or usage patterns, leverage your behavioral user segments to send highly targeted messaging that guides them toward value-driving behaviors.

For example, if a user signed up for the music streaming app but only played two songs, send them a message about a third song to play to get them to their aha moment.

But for a user who’s already played three songs, you’d target them with a different message, relevant to their behavior and the next action in their lifecycle.

You’ll also want to customize messaging based on identifiers (premium, B2B, etc.), not just actions.

Build a data-driven comms flow

As with all things, consistency is key, so plan out an activation communication strategy. I created a FigJam flowchart of how our music streaming app could message by behavioral segments. It can be implemented using customer engagement platforms like Customer.io or Braze.

While it may seem complex, it's a systematic approach to guiding users toward activation based on their actual behavior rather than arbitrary timing.

Initial welcome

1. User signs up → Trigger automated welcome email
2. Welcome email focuses on driving toward the magic number: "What are your three favorite songs? Find them on our platform!"
3. Wait X days (use your frequency analysis as a guiding point here)
4. Check activation status using .identify boolean

If activated?

1. Send a "power moves" email highlighting advanced features like playlist creation
2. Wait a few days
3. Send a product-market fit survey (based on Superhuman's model)
4. Check premium qualification
   • If qualified: Send premium upgrade email highlighting benefits (ad-free, sharing features, etc.)
   • If not qualified: Send referral email (active free users can be valuable advocates)

If not activated?

1. Check setup completion status
2. If setup incomplete:
   • Send a "need help" email with specific guidance
3. If setup complete but aha moment not reached:
   • Send targeted content that helps reach the aha moment
4. If after that they don’t activate:
   • Send habit-forming prompts (similar to Duolingo's approach)
5. If the user still hasn’t activated, send an additional email depending on their status:
   • For users who never complete setup, send a churn survey. For users who complete setup but don't activate, send a feedback survey to understand obstacles.
6. If a user becomes dormant:
   • Send a re-engagement campaign focused on habit formation

Ready to activate better activation?

There’s a lot here! And in putting this into practice, it may take some digging and experimentation. Just remember to start with clean tracking, find your key metrics, share your taxonomy and charts with your team, and gradually build out your communication flows based on real user behavior. Your data will tell you where to focus next.