← Back to all notesJUL 11 · 2026

The app that needs you to need it

Start with a question that sounds cynical and isn't.

Why would an app want you to get better?

If it makes money while you struggle, then your recovery is the moment its revenue ends. Not a milestone. A loss. And a business does not, as a rule, build machinery to produce its own losses faster.

This is not an accusation. It's arithmetic. And once you see it, you can't unsee it in any product that promises to fix you.

"Free" was never the price

There are two ways a recovery app earns from a person in pain. Neither of them is the subscription fee. The fee is the part you can see.

The first way is your data.

In 2023, the FTC ordered the online counselling service BetterHelp to pay $7.8 million, settling charges that it had shared consumers' sensitive information with advertisers after promising to keep it private. The information included email addresses, IP addresses, and answers to a mental health questionnaire — including whether a person had experienced depression or suicidal thoughts. That data went to Facebook, Snapchat, Criteo, and Pinterest, to make advertising work better. BetterHelp denied wrongdoing and settled.

A year later, the FTC reached a settlement of more than $7 million with Cerebral, a mental health and prescription service. The agency charged that Cerebral had passed the sensitive data of nearly 3.2 million people to third parties including LinkedIn, Snapchat, and TikTok, through tracking tools built into its own website and app.

And Monument — a service for people trying to stop drinking — was barred by the FTC from disclosing users' health data to Google and Meta for advertising, after promising that same data would be "100% confidential."

Read the pattern, not the names. A person reaches out at their lowest, hands over the truest thing about themselves, and is told it is safe. Then the truest thing is quietly turned into an advertising signal — because the truest thing is exactly what advertising pays most for.

The vulnerability isn't a side effect the company tolerates.
The vulnerability is the inventory.

The second way is quieter

The data story has been told. The other one hasn't, and it's the one that should worry you more, because it needs no data broker and breaks no promise on paper.

The second way an app earns from you is simply by keeping you.

Return to Cerebral, because the FTC's own account contains a detail that says everything. At one point the company added an easier way to cancel. Two weeks later, it removed that button — after cancellations went up. And over roughly two and a half years, it collected more than eight million dollars from people who had already asked to leave.

Nobody had to sell a single record for that to happen. No promise was broken in the fine print. This is just what a business does when its revenue depends on your staying: it makes leaving a little harder, a little slower, a little more likely to be abandoned halfway.

Now widen the lens past any one company. Think about the streak that resets to zero. The daily check-in that guilts you back. The progress that lives on their server, so leaving means losing it. The badge, the counter, the gentle notification that arrives exactly when you were about to forget the app existed.

None of these are malfunctions. They are the model, working.
An app that earns when you stay has a reason to keep you staying.

This is not a story about bad people

It would be easier if it were. You could name a villain, feel clean, and move on.

But most of the people building these products are not cruel. They are answering to a structure. When your revenue is a subscription, retention is survival, and every incentive — funding, growth, the next round — points toward one instruction: make them come back. You do not have to be a bad person to follow that instruction. You only have to not notice you're following it.

Which is the same thing we say about the person using the app.

The man refreshing at two in the morning isn't weak. He's inside a loop that was built to be hard to leave. The company optimising for his retention isn't evil. It's inside a loop that was built to be hard to leave. The machinery is the problem, on both sides of the screen. Blame is just a way of not looking at the machinery.

So we don't ask whether an app is run by good people. We ask a colder question:

Does it need you to need it?

What we did about our own incentives

We are not going to claim we're better people. We built a different machine, and the machine is what you should trust — not us.

On data: your information does not leave your phone. Not because we promise restraint, but because the app is built so the promise isn't ours to break. Your entries are encrypted on your own device. Our server does not hold them, cannot read them, and therefore has nothing to sell — to an advertiser or to anyone who breaks in. There is no vault of your worst nights, because there is no vault. We wrote about the architecture of that in Not another place to hide.

On staying: Meridin is built to graduate you. The whole thing is designed to become unnecessary — to work itself out of your life in a few months, not to install itself for years. There is no streak to protect. There is no progress trapped on our side that you'd forfeit by leaving. If it's working, the correct ending is that one day you stop opening it and don't notice for a while.

That is a strange thing for a business to want. It is, honestly, a worse business than the alternative. A product built to let you leave will always grow slower than a product built to keep you.

We think that's the point. The slower thing is the honest thing.

What this means for you, tonight

You don't need to become an expert in advertising technology or subscription economics. You need one question, and you can ask it of anything that promises to fix you:

If I got completely better tomorrow, would this company be glad — or would it lose me?

If the honest answer is "lose me," then some part of that product, however well-intentioned, is quietly working to make sure tomorrow doesn't come too soon.

You were looking for a way out.
Make sure you didn't walk into a nicer room.