How to think about retention when the goal is churn?

I haven’t seen articles about this, but it comes down to this: your app needs a way to differentiate good churn (solved the problem) from bad churn (merely stopped using the app). An exit poll would be unlikely to get used by those whose problems weren’t solved, so you’d have to get more creative than that.

That said, in your case, I’m skeptical that you’ll have much hidden good churn in the early days. Depression seems like a long-lasting enough problem - and one that requires ongoing maintenance - that I’d defer this question until you’re receiving anecdotes like “3 months using your app has changed my life! Thank you!” every week (like dating sites do from people in new relationships). Make sure users know you value those stories. When emails like that start showing up regularly, like weekly, then maybe good churn is big enough to be worth trying to calculate. Also, you’ll have some actual examples that show how success manifests in app usage.)

(Obvious exception if your app is totally transactional, like a directory of therapists and all you do is connect the patient. If you’re providing help over time, some users will thank you if it works.)

One of my startup friends brought up that there's a billion dollar company whose work is funerals. I joked that they're doing pretty well for a company with 0% retention rate.

But how does such a company do well? Their focus is customer acquisition and referrals. It's perfect for something like curing depression too.

One of my clients is a mental health app. What we do is track their feedback on discomfort levels (physical, mental, emotional, visual) and how it increases or decreases as the treatment progresses. The payment system has been either pay one shot the first time or via donation. I'm wary of anything that charges for a subscription.

>Dating apps have the same problem to some extent I guess.

No. I think Kent Beck tweeted something along those lines, that the mere fact one was still using the app meant the app failed to do the "job to be done" and my reply was that the tweet assumed that search would stop once a match was found, when clearly it wasn't necessarily the case. Many users do concurrent processing of matches, with some being better than others as their processing algorithm is more efficient, and they can handle more IO and compute workloads. Some do round-robin. Though others pin one thread for processing matches sequentially.

I think this is different. This is similar to hospitals optimizing for patients not coming back after coming in (mitigation), and systems being improved to prevent "failure demand".

What you can draw inspiration from is "remote patient monitoring". I think you have to find differences in behavior from metrics and dimensions. There's that scene from the "Apollo 13" movie where the characters rip off vitals monitoring electrodes and the ground person freaks out looking at the telemetry data thinking they had lost the crew.

I also think the hypothesis that the goal is "churn" is to be revisited. Can we view your device as something that accompanies someone suffering from depression until they recover, and then preventing them from relapse and regression. Your device helps get better, and it is a spectrum, not a binary "depressed"/"not depressed". Similar to physical fitness, you don't stop once you're there, because there is dynamic. So your product adapts and can provide value throughout the journey, as opposed to punctual.

It's not a wedding planner one needs punctually, it's more like a buddy who will be there at the wedding, during the divorce, after a hangover, at the gym, and much more. Support.

So, what are the behaviors exhibited by people who are depressed, by people who are recovering, by people who're giving up (maybe throwing the device). Can you detect those?

You may have started in a niche thing with depression, but could eventually later expand to wellness.

Don't call it churn. Call it "cured". In fact, your success metrics should depend on how many clients you acquired and how many you cured. Imagine saying:

"Over the years, we have helped 1000+ clients get better/cured".