Spiko English

Problem

English learning apps in Central Asia largely localize content from Western markets, missing the grammar pain points specific to Uzbek/Russian speakers. The top-ranked apps also require subscriptions that aren’t accessible at local pricing.

Approach

• Android-first with Jetpack Compose + Clean Architecture
• Offline-first content delivery; syllabus works without internet
• Spaced repetition engine calibrated for native-language drift patterns (Russian/Uzbek → English)
• Localized pricing via Google Play country-specific SKUs
• Analytics-driven paywall experiments (A/B tested 10+ variants)

Outcome

• 100,000+ installs and 4.8 stars on Google Play
• 99.9% crash-free sessions
• Cash-positive after infrastructure, AI inference, platform fees and taxes
• 44-minute average session length
• ~1,320 commits over the lifetime

What I learned

• Solo-shipping a revenue-generating app forces you to care about every layer equally — one slow DB query costs you real money
• Cohort-driven feature decisions beat gut feel; A/B’ing the paywall added 20%+ conversion uplift
• 44-minute average sessions tell you the content + spaced repetition loop is working; optimize for that metric first

Available on Google Play and the App Store. Product updates and support run through the Telegram channel, with 1,598 subscribers.