We’re Not Trying to Keep You in the App

In a randomized controlled trial, how much people used Sonia didn’t predict how much better they felt. What did: whether the first week earned their trust.

At Sonia, we think about you a lot. Not in a metrics-on-a-dashboard way, but in a *what does this person actually need* way. That leads us to questions most tech companies would rather not ask — like whether spending less time in our app might be a sign it is working.

We recently finished a randomized controlled trial, independently analyzed, looking at how members actually use Sonia and what that use means for their wellbeing. The results gave us a new north star.

Does using a wellbeing app more often help more?

People use Sonia in all kinds of ways. Some go deep — journaling every morning, reflecting every evening, working through everything the app offers. Others check in now and then, a few sessions here and there. Some find what helps them quickly and step back until the next thing comes up.

Here is what surprised us: across all of those patterns, members reported meaningful improvements in how they were feeling. Even brief contact seemed to be enough for many people. Among members who checked in sporadically, 56% reported a meaningful improvement in mood and stress, compared with 67% of the heaviest users — a difference that was not statistically significant (p>0.5).

But the finding that really stuck with us was *why* that happens.

What actually predicted feeling better?

It was not how many times someone opened the app. It was not session length, or streaks, or how many features they touched. The strongest predictor was something harder to measure: whether, in that first week, people felt like Sonia *got* them. Whether the approach made sense to them. Whether they felt genuinely connected to it.

Researchers call this working alliance — the sense that you and your support are on the same page, working toward the same goals, in a way that feels real. In our study, early alliance was the strongest predictor of improvement, and it accounted for 51% of the relationship between how much someone used Sonia and how they ended up feeling. Not the volume of engagement. The quality of the connection.

What this means for how we build

Plenty of apps are built to maximize the time you spend in them: more opens, longer sessions, more notifications. Wellbeing apps are not immune to it. We could build Sonia that way, and it would probably lift our engagement numbers. Our investors would be delighted.

But that is not what our data say helps you feel better. What helps is whether Sonia earns your trust in that first week — whether the techniques feel worth trying, whether the goals feel like *your* goals, whether the support creates a real sense of connection. So that is what we have decided to optimize for.

In practice, that means more of our work is going into your first week: making onboarding feel personal, giving you an honest sense of what Sonia does and why, and building trust before anything else.

What we won’t do

We won’t send endless notifications to pull you back. We won’t gamify your wellbeing. We won’t design Sonia to be something you feel compelled to open, because compulsion and care are opposites, and making you dependent on an app is not what is best for you.

If you find what you need in three sessions, good. If you want to check in every day, we are here. What matters to us is that when you are here, the time means something.

Thank you for trusting us

This research exists because real Sonia members showed up, shared their experiences, and let us learn from them. We are grateful for that, and we take it seriously. Every finding we publish and every decision we make is in service of the same thing: being genuinely useful to you, on your terms, when you want it.

Source: Forman-Hoffman, V., et al., 2026. Submitted to a peer-reviewed journal on July 31, 2026. Not yet peer-reviewed and not yet publicly available. The study was conducted by Sonia Health; author affiliations and competing interests are disclosed in the manuscript.