Our most engaged users weren't our best users. They were our most anxious ones.
For months, we celebrated our power users at Murror — the ones journaling every day, sometimes multiple times a day. They had the highest session times, the most entries, the best retention curves. On paper, they were our success story.
Then we started reading what they were actually writing.
Many of them weren't journaling out of growth or self-reflection. They were stuck in loops — writing the same anxious thoughts over and over, looking for reassurance the app couldn't give them. The product was becoming a crutch, not a tool.
This forced us to rethink what "engagement" even means for a wellness product.
We stopped optimizing for daily active usage and started tracking what we call "resolution rate" — the percentage of users who journal about an issue and then naturally reduce their journaling frequency about that same issue over time. In other words: did they work through it?
The counterintuitive result: our best users are the ones who use the app less over time. They come in with something weighing on them, they process it, and they move on with their lives. That's the outcome we should be designing for.
We also added gentle nudges when our system detects repetitive anxiety patterns — suggesting the user might benefit from talking to someone (a friend, a therapist, a trusted person) rather than continuing to journal alone.
Our DAU dropped. Our investor-friendly charts got a little less impressive.
But our NPS went up 20 points. And our word-of-mouth referrals doubled.
I think a lot of products, especially in the AI space, are accidentally optimizing for dependency when they should be optimizing for resolution. The best product experience might be the one that teaches you to need it less.
Curious if other founders have faced this tension between engagement metrics and actual user wellbeing — and how you navigated it.
Replies
The hard part about that kind of loop is that from the inside it doesn't feel like avoidance. Writing the same thought for the fifteenth time feels like you're taking it seriously, like you're doing the work. The app probably felt responsible to use and that's what makes it a crutch rather than a tool... not that it feels bad, but that it feels productive while not actually moving anything🤷♀️
@edikan_peters I’ve started questioning engagement metrics because of insights like this. If users can’t comfortably step away from a product, is that really success?
Murror
@edikan_peters @new_user___090202674ab6e030a7a9c52 That's the right question to ask. I'd say real success is when users can step away feeling lighter, not when they can't stop coming back. We've started thinking of it like a good therapist — the goal is to eventually not need one. A product that keeps you dependent isn't serving you, it's serving its own metrics.
Murror
@edikan_peters You nailed it. That's exactly the trap — it feels like effort, so it feels like progress. One of our users told us "I journal every day, I'm doing the work" but when we looked at her entries, it was the same three worries on repeat for weeks. She wasn't processing, she was rehearsing. That's when we realized the app had a responsibility to gently interrupt the pattern, not just facilitate it.
@monatruong_murror Yes, she just needed something to notice the loop before she could. Most products would've just celebrated the daily streak and moved on. That's so thoughtful of you...
Murror
@edikan_peters Thank you for seeing it that way. That's exactly the mindset shift we had to make — from celebrating streaks to celebrating breakthroughs. The hardest part was accepting that a "successful" user might look like someone who stops using the app. But when you reframe it as the app doing its job, it actually feels more meaningful than any retention chart ever did.
@monatruong_murror What convinced the team to change direction?
Murror
@zara_noelle Honestly, it was one specific user interview. She broke down crying and said "I feel worse after using your app but I can't stop opening it." That sentence haunted us. We ran a deeper analysis and found she wasn't alone — a meaningful segment of our most active users showed the same pattern. Once we saw it, we couldn't unsee it. The team agreed we'd rather build something that actually helps people, even if it meant our DAU charts looked less impressive to investors.
I completely relate to this. Some platforms kept me highly active, but mentally exhausted at the same time. The products I respected most were the ones that didn’t demand constant attention from me.
Murror
@mathew_chang That distinction you're making — active vs. exhausting — is so important. We've started calling it "calm engagement." The products that earn long-term loyalty are the ones that respect your attention, not hijack it. It's a harder business to build, but the retention is real. People don't churn from something that genuinely makes them feel better.
This one landed for me. We had the exact same thing happen. High session times looked like love on the dashboard, but it was anxiety. Users kept coming back to double-check, re-read, re-confirm everything.
The terrifying part is how easy it is to optimize for that without noticing. You see the retention numbers spike and think "ship more of this." But you're just feeding the loop. When did you first realize the gap between the metrics and what people actually felt? Was it something that jumped out or more of a slow realization?
Murror
@nolan_vu It was more of a slow burn that suddenly clicked. We'd been seeing high session times and celebrating them in standups for months. But the real wake-up was a user interview where someone said she opened the app every night but dreaded it — she felt like she "had to" journal or her anxiety would spiral. That hit hard. We went back and cross-referenced our highest-engagement users with our NPS detractors and found a surprising overlap. The numbers had been telling us a story we wanted to hear. Once we started reading what people actually wrote instead of just counting how often they wrote, the gap became impossible to ignore. Sounds like you saw the same thing — curious what you changed on your end once you noticed it.
@monatruong_murror I began to explore more about what make clients anxious and try to come up with some solutions to them. Not effective all the time but at least it can improve the client experience towards your products or services
Murror
@nolan_vu That's a really healthy approach. The fact that it's not effective every time is actually a good sign — it means you're being honest about the complexity instead of trying to force a one-size-fits-all fix. We've found the same thing: some patterns respond well to a gentle nudge, others need a completely different kind of support. The key shift for us was accepting that the app doesn't have to solve everything — sometimes the best thing it can do is help someone recognize they need a conversation with a real person, not another journal entry.
@monatruong_murror I agree with you on that. Hope that we can all improve our product/ service better off in such way that users find it useful for their personal usage. Thanks again for your kindly response
Have other founders related to this story?
Murror
@sarah_butler1 A few have reached out privately since I posted this, actually. It seems to resonate especially with founders in health, wellness, and education spaces where the line between "helpful" and "habit-forming" gets blurry fast. Would love to hear more stories if anyone else has navigated this tension — it's a conversation that doesn't get enough attention in product circles.
This is a much healthier metric than raw engagement. I especially like that you’re measuring whether the same issue naturally shows up less over time, not just whether someone opens the app again tomorrow.
One product question I’d be curious about: do you distinguish “repetition with new detail” from “repetition as rehearsal”? In writing/reflection tools those can look similar in a simple activity chart, but the user experience is totally different. A good nudge probably needs to know whether the person is adding new evidence/clarity, or just restating the same fear with different wording.
Murror
@jim_jeffers That's a really sharp observation. We actually struggled with exactly this. Our early detection was too blunt — it flagged anyone who wrote about the same topic repeatedly, which caught people who were genuinely working through something layer by layer. What we landed on was looking at emotional texture, not just topic overlap. If someone journals about work stress but each entry explores a different angle — a conflict with a manager, then a values misalignment, then a boundary issue — that's processing. But if the entries echo the same phrasing, the same circular worry, with no new framing or insight, that pattern looks more like rehearsal. It's not perfect yet, but combining semantic similarity with sentiment trajectory gave us a much better signal than topic matching alone.
DAU and session time were built for social and media apps where more consumption is the product. Wellness, education, and productivity tools inherited these metrics without questioning whether they applied. Your DAU drop — NPS jump result isn’t a tradeoff. It’s proof you were measuring the wrong thing all along.
Murror
@habibferdous You've put into words something we felt but couldn't quite articulate at first. We kept trying to "fix" our DAU numbers before realizing the frame itself was wrong. When we finally stopped treating lower session time as a problem to solve and started treating it as a signal of progress, everything clicked. The hardest part was explaining that to stakeholders who were trained to read those charts one way. We've started sharing NPS alongside resolution rate in every report now — it tells a much more honest story about whether we're actually helping people.
This is a really useful distinction. “Resolution rate” feels much closer to the job the user actually hired the product for than DAU, especially in a wellness context.
One metric I’d be tempted to pair with it: whether the user can name the thing they’re no longer looping on. A drop in usage could mean resolution, avoidance, or churn; but “I came in anxious about X, I now understand Y, and I don’t need to keep reopening it” is a much stronger product signal.
The brave part is accepting that the best retention story may be episodic trust rather than daily dependency.
Murror
@jim_jeffers Love the framing of "episodic trust over daily dependency" — that's a much better way to describe what we're building toward. And yes, your proposed metric is close to what we've been exploring. We've started pairing resolution rate with a simple self-report: "Do you feel like you understand this better now?" The combination of behavioral signal (they stopped looping) and self-reported clarity has been the closest we've gotten to measuring real progress. Still iterating, but it feels like the right direction.