Qualitative personas for smart segmentation

Welcome to Edition 30 of MAMA Boards, an AppsFlyer video project featuring leading mobile marketing experts on camera.

For today’s mini whiteboard master class, we have Natalie Fitch, CRM Manager for Retention & Engagement at Babbel, one of the world’s top-grossing language learning app.

Natalie shares how developing a personalized and relevant CRM strategy is used to encourage user behavior to drive higher rates of retention. She will show how Babbel has segmented four user personas to build messaging to better segment your CRM and drive a higher LTV.

Real experts, real growth. That’s our motto.



Hi, welcome to another edition of MAMA Boards by AppsFlyer.

My name is Natalie Fitch, and I’m the CRM manager in charge of retention and engagement for Babbel, the language learning app.

Recently I received a presentation from some of our product people, called subscriber segmentation. I was very intrigued. I’m not a big fan of a one size fits all CRM strategy, so anytime that I can kind of segment my users into communications that would be most relevant to them, I’m all there. I opened up the presentation and found heaps of data. I didn’t even know where to start.

So today, I’m going to show you how to comb through all of that qualitative data, to find the most relevant points for CRM segmentation. Then, I’m gonna help you refine those segments using the leftover data. And then look at use cases for each of our user personas at Babbel.


Connecting the dots with people-based attribution


Identifying the four user personas

So I focused on three filters. At Babbel, we have four user personas. We have power learners, striving learners, wishful learners, and dormant learners. So what I needed to do was take all of this data and try to define each of those use cases within the CRM tool.

So first, I split the data into two types of variables. Defining variables, which define each of the user personas, and cohort insights, so data which describes the most recent cohort of that user persona. But first I had to decide what actually makes a good CRM segment.

Two things, first of all, every user should fit into at least one CRM segment, we should not have a single user that fits into none, and second of all, users should be able to move freely between the CRM segments and quickly, so that their communications are most relevant to them at any moment.

So that made it quite clear, the data point which was the defining variable was behavioral. Our power learners learn on a weekly basis. Our striving learners learn but not quite as frequently. Our wishful learners are active in our product, but they’re not completing any learning activity and our dormant learners are not active at all.

categorize data for smarter segmentation mobile marketing

Segmenting personas into your CRM tool

Now let’s talk about how we actually build out those segments within our CRM tool. I needed to make sure that within the CRM tool we had all of the correct filters to make it possible to send each user persona relevant messaging.

So, I focused on three filters: the last date of activity within the app, the last date of learning activity within the app, and the frequency of learning activity within the app. That way all four user personas were represented within these filters, and we were able to target all users from power users to dormant users with relevant messaging.

After building out the behavioral segments within our CRM tool, we wanted to use the rest of the data points in order to further refine the segments that we use to send specific communications to each user, to make sure that they were relevant to them because as we know, not every power user is equal.

build segmented a CRM strategy out of dynamic variables

Demographic and geographic insights

When I think about what I can segment a user on, each of these data points, I think of them in terms of a funnel with behavioral on top, demographic in the middle, and geographic on the bottom. And the closer to the bottom of the funnel that you get, the more personalized that communication is.

As we continued moving down this funnel to further refine our segments, we moved on to demographic insights. For example, we knew that striving learners were less likely to learn a new language for work-related reasons. So we stopped recommending them work-related content, and instead recommended them culture-related content or travel-related content. This was a test that we ran.

Then, we moved on to geographic insights.

We worked on one with our content strategy team to ensure that every geographic region received messaging which was very relevant to their culture and their traditions. For example, we knew that German users prefer to receive messaging that was action-orientated, telling them exactly what the next step is in learning a language. Whereas other geographic regions appreciate getting more content-focused messaging.

First, we combed through all of our data points and found the ones that are most relevant for building out our user personas. Then we built out our segments within the CRM tool, refined our segments using cohort insights and now onto the final user-facing part, tailoring our communications to each of those segments.

refining segments - mobile marketing

What are your KPIs?

When we’re trying to ideate these communications, we focus on hypotheses, what needle can we move using these communications? Is it a product KPI? Do we want users to enter the product more often? Do we want them to move between the different user personas? These are a few things that we can change, or that we can encourage users to do, using communications.

So for example, for our power learners, we want to affirm their positive behavior. They’re already learning weekly, so every week we send them a weekly summary telling them how many lessons and review sessions they’ve completed that week. It might say something like, “Wow Natalie, “You’re making progress!” and hopefully that will encourage our power users to continue doing what they’re doing.

For striving learners, we might want them to move into this power learner segment. So we encourage behaviors that correlate with being a power user, such as using the review feature. So here, we would encourage feature adoption to our striving learners. “Review Spanish now,” is something one of those communications might say.

For our wishful learners, unfortunately, they’ve never completed a learning activity. Sometimes learning a new language can be tough. It’s tough to find the confidence. So we wanna motivate them, show them what their future self might look like. Show them what their motivation was to first learn a new language. So we will send them communications to motivate them to start learning, such as, “Make your past self proud.”

For dormant users, we could also try to get them to start learning and we do, but that’s a really tough needle to move, so instead, we encourage them to engage with our brand in other ways. We might send them things like podcasts or magazine articles. “Another way to learn Spanish.” That’s the copy on one of our communications about podcasts.

tailoring communications for mobile marketing

High rates of retention by identifying user behavior

If you should take one thing away from this, it’s that our entire goal with developing a personalized and relevant CRM strategy is to encourage user behavior which correlates with high rates of retention. The longer a user engages with our product, the higher their LTV. That’s why when we build out our segments, we focus on the past behavior of the user. And when we build out our messaging, we focus on encouraging good future behavior. I hope that you found these insights valuable, and that can take this process back to your own CRM team to build out a highly relevant and personalized CRM strategy.

And that’s it for today, for more MAMA Boards, click this link up here. And as we say in Germany, Dankeschön und auf Wiedersehen aus Berlin.

Ready to start making good choices?