The hard part of building an app in 2026 isn't building it
The hard part of building an app in 2026 isn’t building it. It’s getting anyone to see it.
Sensor Tower counted roughly 560,000 new App Store apps in the first half of this year, almost matching all of 2025. Downloads grew 2% in the same period. Supply doubled. Demand didn’t move.
Islamic apps are their own flooded corner of that. Most bundle the same four APIs — prayer times, Quran text, Qibla, a dua library — and call it done. The barrier to entry is close to zero, and the shelf is full.
The numbers, four months in
Here is where Rawi stands as of August 2026. The ratio between these numbers is the whole story.
- 835 downloads since late April
- 290 people opened Rawi in the last 30 days, 35% of them still active
- Roughly 1.4 ayahs identified per person per week
- 18% still there after a month
- All 10 App Store reviews are five stars
Getting installed costs a weekend and an API key. Staying on someone’s phone doesn’t. I have the second problem mostly solved. The first one is wide open.
Where the paid spend went wrong
Content is what moved anything at all: 307,000 TikTok views across 11 videos, one of them doing 94,000 on its own. Some paid, some organic, all producing a trickle of people who actually stay.
The mistake was where I spent the paid part. I bought reach in Türkiye, where roughly three in four phones run Android. Rawi is iOS-only today — Android is still in closed testing. I paid to be discovered by an audience that mostly couldn’t install the app.
The more useful lesson arrived second-hand: at this size, ambassadors move the needle further than paid creators. The person doing the most for Rawi right now has never met me, and has an exceptional talent for making content in a region where Rawi lands. No invoice, better conversion.
The question I’m actually asking
It isn’t how to get more downloads. It’s narrower than that: how do you reach someone at the exact moment they’re wondering which ayah they just heard, without buying reach that never converts?
I don’t have that answer yet. But the shape of the problem is clear, and it isn’t a build problem. Building was the part I already knew how to do.