Mobile Performance Optimization

Mobile performance is measured differently than web performance. Users notice jank at 16ms, battery drain over hours, and data usage over cellular connections. The constraints are harder — limited CPU, limited memory, limited bandwidth, limited battery — and the expectations are higher because native apps are supposed to feel instant.

---

Performance Metrics

Metric	Target	Measurement Tool
Cold start time	< 1 second	iOS: Instruments, Android: adb
Warm start time	< 500ms	Time to interactive
Frame rate	60 fps (16.6ms per frame)	Core Animation, GPU rendering
Memory usage	< 200MB typical	Instruments, Android Profiler
Battery impact	< 5% per hour active use	Energy Diagnostics, Battery Historian
App size	< 50MB download (ideally < 20MB)	App Store Connect, Play Console
Network data	Minimize redundant transfers	Charles Proxy, Network Monitor

---

Startup Optimization

Phase	What Happens	Optimization
Process creation	OS launches process (10-50ms)	Cannot optimize
Runtime initialization	Load frameworks, initialize runtime (50-200ms)	Reduce linked frameworks, lazy load
App initialization	AppDelegate/Application setup (50-500ms)	Defer non-critical work
First frame	Layout + render first screen (100-300ms)	Pre-computed layouts, skeleton screens
Data loading	Network/cache → populate UI (200ms-2s)	Cache first, background refresh

Techniques

Technique	iOS	Android
Lazy initialization	Defer service setup to first use	Lazy properties, WorkManager for deferred
Module loading	Reduce linked dylibs	Reduce dex file count
Skeleton screens	Show placeholder UI immediately	ShimmerLayout, Compose placeholders
Pre-warm data	Cache critical data on disk	Room/DataStore pre-population
Reduce main thread work	Move initialization off main	Move work to background dispatchers

---

Memory Management

Issue	Symptom	Diagnostic	Fix
Memory leak	Growing memory over time	Instruments Leaks, LeakCanary	Fix retain cycles, clear references
Image memory	Large memory spike on image load	Memory graph debugger	Downscale images, use disk cache
Cache unbounded	Memory grows with data	Memory monitor	LRU cache with size limit
View controller leak	Memory grows on navigation	Instruments Allocations	Weak references in closures/callbacks
Retained fragments	Android memory growth	Android Profiler	Proper fragment lifecycle

Image Memory Rules

Image Size	Raw Memory	Optimization
4000×3000 (12MP photo)	~48 MB	Downscale to display size
1080×1920 (phone screen)	~8 MB	Acceptable for full-screen
100×100 (avatar)	~40 KB	Cache in memory

---

Battery Optimization

Battery Drain Source	Impact	Mitigation
GPS (continuous)	Very high	Request only when needed, significant location changes
Network (polling)	High	Push notifications instead of polling
CPU (background)	Medium-high	Background task time limits
Screen rendering	Medium	Avoid unnecessary re-renders
Bluetooth scanning	Medium	Scan only when user initiates
Animations	Low-medium	Use hardware-accelerated animations

---

Network Optimization

Technique	What It Does	Savings
Response caching	Cache API responses locally	50-80% fewer requests
Image CDN	Serve right-sized images	60-80% image data savings
Request batching	Combine multiple API calls	Fewer connections
Delta sync	Send only changes since last sync	90%+ data savings for sync
Compression	gzip/brotli responses	60-80% response size reduction
Offline-first	Work without network, sync later	100% savings when offline

---

Profiling Tools

Tool	Platform	Best For
Xcode Instruments	iOS	CPU, memory, energy, network profiling
Android Studio Profiler	Android	CPU, memory, network, energy
Charles Proxy	Both	Network traffic inspection
Firebase Performance	Both	Real-user performance monitoring
Flipper	Both	Debug tool with plugins
LeakCanary	Android	Automatic memory leak detection

---

Anti-Patterns

Anti-Pattern	Problem	Fix
Loading full-res images	Memory spikes, OOM crashes	Resize to display dimensions
Polling for updates	Battery drain, bandwidth waste	Push notifications, WebSocket
Synchronous network on main thread	UI freezes	Async networking, loading states
No image caching	Re-downloads same images	Disk + memory cache (Kingfisher, Coil)
Unbounded lists	Memory grows with data	Recycler views, lazy lists
No startup profiling	Slow cold start unnoticed	Measure and track app startup time

---

Checklist

• Cold start time < 1 second
• Maintain 60 fps in all scrolling and animations
• Memory leaks monitored and fixed (LeakCanary, Instruments)
• Images resized to display dimensions before loading
• Network requests cached with appropriate TTL
• Battery impact profiled and optimized
• App size monitored with alerts on regression
• Background work minimized and scheduled appropriately
• Real-user performance monitoring in production (Firebase Performance)
• Performance regression testing in CI

---

:::note[Source] This guide is derived from operational intelligence at Garnet Grid Consulting. For mobile engineering consulting, visit garnetgrid.com. :::