Yes. With a condition attached.
Mental health apps produce real, statistically significant effects on anxiety and depression. The effects are small. They erode when you use the app alone. They strengthen considerably when any human contact is layered in.
Bemellou was built on exactly this logic: a physical, zero-friction first step for people who wouldn't open an app, let alone book a session. The question isn't whether apps work. It's work as what, exactly.
Small effect, enormous variation
A small effect is still a real one. Across 176 randomized controlled trials, researchers found mental health apps produced statistically significant improvements in depressive and generalized anxiety symptoms compared to control groups, with effect sizes of roughly g = 0.28 and g = 0.26. Those numbers sound modest. In a population where a large share of people receive no support at all, modest beats nothing.
The more useful frame: a small average effect contains enormous variation. Some apps, for some people, under some conditions, work very well. Others do almost nothing. The meta-analysis was averaging across all of them, the rigorous and the shoddy, the supported and the fully self-guided.
That variation is where the real story lives.
When another person changes everything
Apps with human contact consistently outperform fully self-guided ones, and the gap is not subtle. A meta-analysis of 92 RCTs found that apps incorporating human support showed stronger effects, and the support didn't have to be intensive. A check-in. A prompt from a real person. Something that signaled: someone knows you're here.
The attrition data reveals the mechanism. Dropout was lower in trials that incorporated reminders or human support. People stop using apps when the apps feel indifferent to whether they show up. The presence of another person, even minimally, changes the felt stakes.
This reframes the whole debate. The binary question "do apps work?" obscures what matters: for whom, and under what conditions? A person who already has a therapist and downloads an app as a supplement is in a completely different position from someone managing everything alone, quietly, and downloading an app at 1 a.m. because they couldn't sleep. The evidence suggests those two people will have very different experiences.
The thing nobody talks about: getting to week two
The opening rate is not the problem. Most people activate the app.
What happens after week one is.
Posttest attrition averaged 18.6% and climbed to 28.4% at follow-up. Nearly one in three people who enrolled in an app-based study had disengaged by the time researchers checked back in. In real-world use, without the mild social pressure of being in a trial, that number is almost certainly higher.
This isn't a failure of the apps exactly. It's a failure of the conditions. An app sitting on your phone can't meet you where you are at 11 p.m. when you don't have the activation energy to open it. A tool is only as good as your ability to reach for it.
That's the gap most "do apps work?" articles skip. They evaluate the app in use. They don't ask what it takes to get a person to the point of opening it in the first place, and staying.
If you're carrying something quietly, the kind that makes you feel like you should be fine but aren't, you might find more on that in When You're Fine and Also Not Fine or High-Functioning Doesn't Mean Fine.
What each type of support is actually good at
| Tool | Best for | Human contact | The honest edge |
|---|---|---|---|
| Bemellou plush + app | People not yet ready to open a mental health app | Via community + optional coaching | Physical, always-available, zero-friction first layer |
| Calm | Sleep, everyday stress, breathing practices | None | Beautifully produced; low barrier for daily relaxation habits |
| Headspace | Structured mindfulness courses, focus | None | Excellent for people who want a curriculum-style approach |
| Finch | Low-energy days, self-compassion, gentle check-ins | None | Warm gamification; well-suited for emotionally depleted users |
| Wysa | Mood tracking, CBT-based chat, crisis access | Optional paid coaching tier | Strong evidence base; anonymous feel reduces shame barrier |
| Therapy / counseling | Moderate to severe symptoms, complex needs | High | Irreplaceable for real clinical work, apps are not substitutes |
No tool in this table treats or cures anything. All are framed as support, access, and companionship, not clinical intervention.
Before the app comes the threshold
Here's the mechanism that rarely gets named. Before someone opens an app, before they book a session, there is a prior question: can I acknowledge that I need something? For a lot of people, particularly high-achievers, people who look fine, people for whom asking for help feels like failure, that step is the hard one.
Within-person barriers, the lack of trust in professionals, fear of stigmatization, or the desire to cope independently, contribute to the treatment gap. Self-guided mental health apps may support treatment-seeking by reducing some of those within-person barriers. But apps have their own barrier: you have to decide to download one. You have to recognize yourself as someone who needs it.
A physical comfort object has no such threshold. You don't have to call it anything. You don't have to admit anything. You just hold it. That is the design logic behind the Bemellou plushies, not a replacement for an app or therapy, but the layer before the layer. The thing that asks nothing of you until you're ready to ask something of yourself.
For more on why this isn't as strange as it sounds, The Comfort Object, Revisited: What Adults Actually Need covers the attachment science directly.
FAQ
Do mental health apps really work, or is the evidence just hype?
The evidence is real, not hype, but it requires context. A 2024 meta-analysis of 176 RCTs found statistically significant effects on both depression and anxiety, modest in size but consistent across a large body of trials. The more honest caveat is that effects vary widely depending on the app, the person, and whether any human support is present. "Apps work" is too broad a claim; "some apps, for some people, under certain conditions, provide meaningful support" is closer to what the data shows.
Are apps with human support actually better than fully self-guided ones?
Yes, consistently. Trials that include any form of human contact, even light-touch check-ins or reminders, show lower dropout and stronger outcomes than fully self-guided versions. The gap matters because most commercially available apps are fully self-guided. If you're evaluating an app, it's worth checking whether it includes any coaching, community, or human-supported features.
What if I'm not ready to use a mental health app at all, where do I start?
That's a more common position than most people admit. Bemellou is built specifically for that in-between space, a comfort plush paired with optional psychoeducation and community, designed as a zero-pressure first step for people who aren't yet ready to open a wellness app or book a session. You don't have to know what you need yet. You just have to reach for something.
Written by Rodrigo Arismendi, Co-founder & CEO of Bemellou
I'm a 19-year-old student at Northwestern, and I started Bemellou because I've felt that quiet weight of stress and pressure myself. Along the way I learned something simple: the first step toward feeling better shouldn't have to feel like a step at all, so I set out to build the softest one I could.
Last updated: 2026-08-07
Keep reading
Comments
Bemellou isn't a crisis service. If you're struggling right now, please reach a local helpline.