Building a ride-hailing app is one of the most technically demanding projects in mobile development. Three separate apps, real-time location sync, dynamic pricing, and payment processing - all running simultaneously. Here's how it actually works.
The System Has 4 Components (Not 1)
Most clients think "taxi app" = one app. In reality:
- Passenger app (iOS + Android) - booking, tracking, payment
- Driver app (iOS + Android) - order acceptance, navigation, earnings
- Admin panel (web) - fleet management, analytics, dispute resolution
- Backend - matching engine, billing, push notifications, maps
You are building four products. Budget and timeline accordingly.
Core Technical Architecture
Real-Time Communication
This is the hardest part. The passenger needs to see the driver moving every 2-3 seconds.
WebSocket (recommended): Persistent connection, low latency. Driver app sends GPS coordinates every 3 seconds → backend broadcasts to passenger app.
Tech stack options:
- Node.js + Socket.io (most common, good ecosystem)
- Go + goroutines (better at scale, harder to find developers)
- Elixir/Phoenix (excellent for real-time, niche)
Driver-Passenger Matching Algorithm
Basic proximity matching:
1. Passenger requests ride at coordinates (lat, lng)
2. Query all available drivers within 3km radius
3. Sort by: distance + acceptance rate + rating
4. Send offer to top driver, timeout 15s
5. If declined/timeout → next driver
For better matching: add surge zones, driver heading prediction, and ETA calculation.
Geolocation Stack
- Google Maps Platform: Routing, geocoding, Places API (~$300-2000/month at scale)
- Mapbox: Cheaper alternative, excellent offline maps
- HERE Maps: Good for Central Asia/CIS markets
- OSRM + OpenStreetMap: Free, self-hosted, requires infrastructure
For CIS markets: HERE Maps has better coverage in secondary cities than Google.
Pricing Engine
fare = base_fare + (distance_rate × km) + (time_rate × minutes)
surge_multiplier = demand_drivers_ratio (1.0-3.5x)
final_fare = fare × surge_multiplier
Variables you need to store per city: base fare, per-km rate, per-minute rate, minimum fare, cancellation fee, surge thresholds.
Feature Breakdown: MVP vs Full Product
Passenger App
| Feature | MVP | Full |
|---|---|---|
| Registration (phone OTP) | ✅ | ✅ |
| Book ride + map | ✅ | ✅ |
| Driver tracking real-time | ✅ | ✅ |
| Cash payment | ✅ | ✅ |
| Card/wallet payment | - | ✅ |
| Ride history | ✅ | ✅ |
| Rating driver | ✅ | ✅ |
| Scheduled rides | - | ✅ |
| Multiple stops | - | ✅ |
| Promo codes | - | ✅ |
| SOS button | - | ✅ |
| Corporate account | - | ✅ |
Driver App
| Feature | MVP | Full |
|---|---|---|
| Registration + document upload | ✅ | ✅ |
| Accept/decline orders | ✅ | ✅ |
| In-app navigation | ✅ | ✅ |
| Earnings dashboard | ✅ | ✅ |
| Driver rating | ✅ | ✅ |
| Online/offline toggle | ✅ | ✅ |
| Weekly payouts | - | ✅ |
| Performance analytics | - | ✅ |
Payment Integration
For international launch:
- Stripe (cards) + Stripe Connect (driver payouts)
- Apple Pay / Google Pay
For CIS markets:
- Yandex Pay, SberPay
- Local wallets: O!Dengi, mBank (Kyrgyzstan), Kaspi Pay (Kazakhstan)
- Cash remains 60-70% of rides in most CIS cities - don't skip it
Backend Tech Stack
API Layer: Node.js (Express/Fastify) or Go
Real-time: Socket.io or native WebSockets
Database: PostgreSQL (trips, users) + Redis (driver locations, sessions)
Queue: Bull/BullMQ or RabbitMQ (notifications, emails)
Maps: Google Maps Platform or HERE
Push: Firebase Cloud Messaging (Android) + APNs (iOS)
Storage: AWS S3 or Cloudflare R2 (driver docs, photos)
PostgreSQL + PostGIS extension for geospatial queries (find nearby drivers) - essential.
Budget Breakdown
| Component | MVP | Full Product |
|---|---|---|
| Passenger app (iOS + Android) | $8,000-15,000 | $20,000-35,000 |
| Driver app (iOS + Android) | $6,000-12,000 | $15,000-25,000 |
| Backend + API | $8,000-15,000 | $20,000-40,000 |
| Admin panel | $4,000-8,000 | $10,000-20,000 |
| Design (UX/UI) | $3,000-6,000 | $8,000-15,000 |
| Total | $29,000-56,000 | $73,000-135,000 |
Timeline: MVP in 4-6 months with a team of 4-5. Full product: 8-14 months.
Monthly Operating Costs (Post-Launch)
- Servers (AWS/DigitalOcean): $300-1,500/month
- Google Maps Platform: $300-2,000/month (biggest variable cost)
- SMS for OTP: $50-300/month
- Push notifications: Free (FCM/APNs)
- App store fees: 15-30% of in-app purchases
- Support team: depends on your scale
Maps API is often the surprise cost. Optimize by caching routes and using static maps where possible.
Legal Considerations
Before launch in any market:
- Driver employment classification (contractor vs employee - varies by country)
- Insurance requirements for drivers
- Data protection / GDPR equivalents
- Payment processing licenses
- Local taxi regulation compliance
Common Mistakes
- Building too much for MVP - launch with cash only, basic matching, no scheduled rides. Validate the market first.
- Underestimating Maps API costs - prototype with free tier, then calculate real costs at scale before committing architecture.
- One app for driver + passenger - never works well. Users need different UX.
- No surge pricing from day 1 - you'll lose money in peak hours.
- Skipping driver verification - background checks, license verification are not optional.