What Building a 5 Million Download App Taught Us About Mobile Product Design
We build MedTalk AI for hospitals and health systems, but we also build and maintain a handful of consumer apps on Google Play and the App Store. They're a different kind of product entirely, and the gap between our highest-download app and our highest-rated app taught us something worth sharing about what actually drives each number.
Two very different kinds of success
| App | Category | Downloads | Rating |
|---|---|---|---|
| Fake Call: Prank Call Dialer | Entertainment | 5M+ | 3.4★ (7,214 ratings) |
| Contacts Backup Cloud Transfer | Productivity | 500K+ | 4.2★ (18,721 ratings) |
| Invoicer: Invoicing Simplified | Business | 5K+ | 4.7★ (26 ratings) |
Fake Call is by far our biggest app by volume: 5M+ downloads, built around a simple, shareable idea (a realistic fake incoming call to escape an awkward moment, or play a harmless prank). It's also our lowest-rated app at 3.4 stars. That's not an accident, and it's not something we'd try to spin.
What a viral, low-friction app teaches you
Novelty and entertainment apps win on download volume precisely because the barrier to trying them is almost zero: no account, no ongoing commitment, often installed for a single use case and never opened again. That same low commitment that drives installs also drives a wider spread of ratings, including from people who used it once, got what they needed, and rated based on a single interaction rather than sustained use.
What a smaller, higher-satisfaction app teaches you
Contacts Backup Cloud Transfer solves a narrower, more specific problem — recovering your contacts after switching phones — and gets used with more intent. At 500K+ downloads and an 18,721-rating base averaging 4.2 stars, it shows a different pattern: fewer downloads relative to Fake Call, but a rating base built on people who actually needed the feature to work correctly, which is a much higher bar to clear consistently at volume.
The lesson for anyone building a mobile product
- Download volume and star rating measure different things, and optimising for one can work against the other
- A simple, single-purpose idea with a low barrier to try will always out-download a more complex utility, but won’t necessarily out-rate it
- Utility apps solving a specific, recurring problem build a rating base that reflects sustained reliability, not novelty
- Neither pattern is wrong: the right one depends entirely on what the product is for and what "success" is meant to mean for it
The takeaway we apply to every product we build, including MedTalk AI, is to be explicit up front about which metric actually matters for that product. For a consumer novelty app, that's reach. For a clinical tool a hospital depends on every day, it's reliability and trust, which is a completely different design and engineering brief.