Oki — Android client

Problem

The product was iOS-first with an existing user base, and Android was still on the roadmap. The Android client had to match the iOS feature set, run its own subscription flow, and pass Google Play review without breaking the monetization that the marketing spend depended on.

Approach

• Built the Android client in Kotlin with Jetpack Compose, matching the iOS information architecture instead of inventing a second one
• Wired subscriptions through RevenueCat, including re-attribution on login so purchases made during signup attach to the right account
• Fixed purchase reporting for signup-funnel purchases
• Kept the store listing honest about which languages the app actually supported

Outcome

• Production release live on Google Play, version 1.0.2
• 10,000+ installs on Google Play
• 400 to 800 daily active users
• Crash rate 0.27 percent and ANR rate 0.11 percent, both well inside Google Play’s bad-behavior thresholds
• Subscription attribution issues closed in the 1.0.2 release

What I learned

• Billing is where Android releases actually break. The screens are the easy part; attribution is the part that costs money.
• A store listing is a promise. If it claims languages the product does not have, the reviews will say so publicly, and they will be right.
• Release work is dependency work. Store access, billing accounts, and review timelines gate the schedule more than the code does, so getting those unblocked early is part of the engineering job, not admin around it.