App estimates vary more widely than website estimates — sometimes by a factor of ten. The reason is that the word "app" covers both three screens and a banking system.
Prices by complexity
- MVP — three to five screens, one core scenario, simple authentication. 2,000,000–5,000,000 AMD.
- Mid-complexity app — user account, push notifications, payments, admin panel. 5,000,000–15,000,000 AMD.
- Complex product — real-time maps, chat, heavy integrations, offline mode. From 15,000,000 AMD upwards.
These figures cover both iOS and Android built cross-platform. Native development for two platforms usually adds 40–60%.
Native or cross-platform
Cross-platform (Flutter, React Native) — one codebase for two platforms. Cheaper and faster. It suits most jobs: a shop, a service app, a customer account, a booking flow.
Native (Swift, Kotlin) — separate code for each platform. Worth it when you need heavy graphics, deep camera or Bluetooth work, or maximum performance.
If you do not know for certain that you need native, you need cross-platform.
What pushes the price up most
- Authentication and user profiles. Sign-up, password recovery, social login — more work than it looks.
- Payments. In-app purchases run under App Store and Google Play rules, each with its own requirements and commission.
- The backend. An app almost always needs a server side and an admin panel. That is a separate project.
- Offline mode. "It should work without internet" doubles the complexity of everything touching data.
Store costs
These tend to be remembered at the last minute:
- Apple Developer Program — 99 USD per year.
- Google Play Console — 25 USD one-off.
- Commission on in-app purchases — 15–30% depending on the programme and your turnover.
Note that Apple can reject an app. The first review usually takes one to three days, but a rejection restarts the cycle. Plan your launch date with that risk in mind.
After launch
Unlike a website, an app cannot simply sit there. iOS and Android ship new versions every year, and an app has to be updated or it stops working — or gets delisted.
A realistic support budget is 15–20% of the project cost per year.
How to reduce the risk
Do not start with the full product. Write out the feature list and split it in two: "without this the app is pointless" and "nice to have". The first half is your MVP.
Launch that, watch how real users behave, and only then pay for the second half. It almost always turns out that half the list was not needed.
