🇺🇸 Conquering the US with AI

Traffic is exploding, and retention is great too?

“If we launch like this, we're doomed”

An Alarmy PO's day begins the moment their 7 AM alarm goes off. It's the only chance to experience the product exactly as a user would. Sure, during the day I thought, "Wow, this is great!" but actually using it in the morning, I realized this feature didn't wake the user up—it just made them angry. "If this goes live, we're in big trouble!" I thought, and we immediately went back in for another round of revisions.

The feature that started with such chaotic energy ended up going viral in the US, driving 4 million DAU. Here's a little sneak peek into the secret of how we pulled it off.

Let's Devour the US Market! 🇺🇸

Our goal for 2025 was "Conquering the US." It's a market with massive potential for incremental DAU growth, and because the LTV per user is high, it's a country that can help us hit our revenue targets. Naturally, the Core squad, which handles Alarmy's core alarm features, jumped right into the mission to conquer the US.

The weapon the Core squad brought to the battlefield was Alarmy's signature missions, specifically the photo mission. You've probably seen a video of Alarmy, the "ruthless alarm app" where you have to go take a picture of your bathroom sink to turn off the alarm. The single weakness of this photo mission, beloved by our hardcore users, was that it was hard to set up. How could we eliminate this friction and make it an even more attractive feature?! To chew on this problem, we put our heads together with the marketing group.

Here at DelightRoom, the product team and marketing team have established a "producting" work style where they build the product together. (If you're wondering what that means, I highly recommend reading our post [Producting: More Than Just a Buzzword].) During our very first brainstorming meeting with the marketing team to kick this off, I realized just how mild my YouTube feed actually was. TikTok and YouTube are absolutely packed with spicy, highly stimulating content.

After an hour of cracking up in our meeting, we narrowed it down to three main hypotheses.

First, make it so users don't have to register an answer photo beforehand.

Second, make random objects appear to maximize the wake-up effect.

Third, add a sense of urgency to make it fun.

Seeing content leveraging these exact traits rack up millions of views gave us the confidence that if we built a feature around these hypotheses, we could take the strong retention of the photo mission and use it to drive massive traffic.

Where Are You Going Without Dismissing the Alarm...

And so, we confidently rolled out the "Find a Household Item" mission.

Since we ambitiously placed it as the very first option during our app installation onboarding, the selection rate hit our target of 10%. But... the rate of people actually dismissing the alarm once it rang was 20% lower compared to other missions. This meant a lot of users were failing to dismiss the alarm and churning by just deleting the app or turning off their phones. Naturally, retention was terrible too.

The photo mission is Alarmy's most attractive feature, but at the same time, it is the highest-difficulty mission. You physically have to drag your body out of bed and take an accurate photo. We kind of expected users to struggle with a new mission, but the data was even colder than we anticipated. So, the product and marketing teams each set out to find angles for improvement.

First, we needed to hear voices from the field. We didn't even have to go far to find users. We started with "dogfooding"—using the feature ourselves. We figured, "Hey, photo mission users use their bathrooms a lot, so maybe it's fine to put bathroom items in the default list?" Well, surprise surprise, when we actually used it, the overwhelming feedback was that it was way too hard. So, based on qualitative feedback and data, we removed the excessively difficult items from the defaults.

There was also a clue hiding in the marketing content. Initially, we thought the fun derived from urgency would be key, but actual ad creative tests showed that using items you'd commonly find around the house grabbed far more user attention than the panic factor. Once we proved that urgency was no longer the core appeal, we stripped away the pressure-inducing mechanics from the mission process and generously loosened the time limits.

Making the first experience easier will lead to higher retention.

Our new hypothesis was spot on. Every time we deployed a revised version, our metrics improved. It was absolutely thrilling.

What was Right Then is Wrong Now

To be honest, it wasn't like we only started dogfooding after the release. We had tested it dozens of times before deployment, but trying out the 100% finished version was a whole different beast, and our users were even more different from us. Because of that, we had to re-validate over 10 hypotheses post-launch. We ended up reversing many policies we had so confidently decided on at the start.

Did my teammates complain or have a hard time with this?

Not at all. In fact, they told me, "I really enjoyed the experience of building a feature and rapidly iterating on it."

There were two secrets to this.

First, the designers and engineers who built it were dogfooding it themselves. Because they had already used the feature, the PO and PM didn't even need to bring in user feedback; the team rapidly understood and fully empathized with the defined problems. They even pitched better suggestions.

For example, in the first version, we tried to simplify the settings screen as much as possible for launch, so the item list was just text. I asked if they could squeeze in tiny images on the right side of the list because it wasn't very intuitive, but they coolly responded, "Wouldn't it be better to just change the whole layout to a grid?" and built exactly that. 😎

Second, we continuously validated our hypotheses by looking at data. What's the selection rate by country? How much does retention differ? Where is the biggest drop-off funnel? Which items make people hit "Choose Again" the most? Which items trigger the most churn? Thanks to diving tenaciously into the numbers, we extracted multiple hypotheses. When the "gut feeling" from dogfooding was backed up by data, everyone felt the same level of conviction, allowing us to push forward with improvements without hesitation.

Ultimately, it comes down to finding problems in the user experience and validating hypotheses with numbers. It's a fundamental principle, but sticking to the basics is what we do best, and the power of the basics was much stronger than we thought.

To 10 Million Viral Views and 4 Million DAU

From Q4 of 2025 until now, "Find a Household Item" has gone completely viral, racking up over 10 million views across social media and becoming a golden goose that improved our install cost twofold. Because the retention is better than typical ad creatives, our DAU shoots up every time we pour traffic in. Hitting 1 million US views was a pipe dream back in Q3 of 2025, but now it's just another day at the office.

Seeing the chart go up is fantastic, but everyone getting to share in a successful experience was an equally massive win. (Nothing quite like the sweet taste of performance therapy...)

I especially loved gaining the conviction that focusing on Alarmy's essence can yield such massive results. Spicy hooks are important, but we learned a valuable lesson: the essence of the service is what actually drives conversion.

4 Million, and Beyond-

The success of "Find a Household Item" is only just beginning. We have around 70 different items we can utilize, plenty of room to advance the feature, and limitless opportunities to target specific countries or seasons. Aren't you curious to see how much more Alarmy will grow in 2026?

If you want to make "truly good" decisions based on data,

If you want to collaborate tightly with teammates obsessed with the user experience,

If you're looking for a company that is conquering the globe,

How about joining us? 😉