Murror
AI that helps you understand your emotions & connect deeper
Start new thread
Mona Truong

We stopped adding AI features to Murror. Our retention went up.

by
Murror
•

Three months ago, we had a backlog full of AI features. Auto-generated mood summaries. Predictive journaling prompts. Smart scheduling for reflection sessions. Voice tone analysis. The works.

Every feature demo'd beautifully. Our team was excited. Investors loved the roadmap.

Then we looked at our data and noticed something uncomfortable: the users who engaged with the most AI features were actually churning faster than users who used fewer features but spent more time with each one.

The power users weren't going deeper. They were skimming the surface across a dozen AI-powered capabilities.

So we did something that felt terrifying as a founder: we froze the AI feature backlog and spent three months making our three core features significantly better instead.

Here's what we changed:

  1. Instead of adding AI mood summaries, we made the manual journaling experience richer — better prompts, more space for nuance, slower pacing.

  2. Instead of predictive prompts, we built a reflection review that shows you patterns only after you've done the work of noticing them yourself first.

  3. Instead of voice tone analysis, we added a simple "how did writing that feel?" check-in after each entry.

The results surprised us. 30-day retention jumped 18%. Average session time increased by 40%. And the qualitative feedback shifted from "cool features" to "this actually helps me."

The lesson: in AI products, more intelligence doesn't always mean more value. Sometimes the AI's job isn't to do more — it's to create the conditions for the human to do more.

We're back to building AI features now, but with a completely different filter. Every new feature has to answer: does this help the user understand themselves better, or does it just make the product feel smarter?

Anyone else wrestling with where to draw this line?

53 views

Add a comment

Replies

Best
Gwendolyn Kira

I actually think this shows maturity. For me, removing unnecessary complexity usually improves user trust. Sometimes the best product decision is knowing what not to build.

Mona Truong

@gwendolyn_kira Totally agree. We spent so long thinking about what to add that we forgot to ask whether we should. Once we shifted to "does this help the user understand themselves better?" as our filter, it became way easier to say no to shiny features. Trust really does come from simplicity.

Anna Jarrett
This is exactly the thinking that takes a product from great to extraordinary. I think of the “AirPods” strategy. They created so much buzz (aside from the logo) because of how simple they were to use. Other models required you to connect it and tune your sound profile and download an app where you are to make a profile. The key I walked away with was, how simplistically intuitive can we make our product?
Mona Truong

@anna_jarrett Love the AirPods analogy. That's exactly the question we keep coming back to. For Murror, intuitive means the app should feel like talking to a thoughtful friend, not configuring a dashboard. If users have to think about how to use it, we've already lost. The magic should be invisible.

Anna Jarrett
@monatruong_murror yes! That’s the magic. It should feel like second nature.
Mona Truong

@anna_jarrett Exactly. That's the bar we hold ourselves to — if it doesn't feel like second nature within the first session, we go back to the drawing board. The best features are the ones users never have to "learn."