I just read an excellent deep dive from the iOS team at inDrive on app performance optimization.
One insight stood out:
Improving a critical user flow by just **750ms_ increased orders by _1.5%**. When you’re operating at scale, milliseconds literally translate into revenue.
Some of the most impactful lessons:
⚡ Measure TTI (Time To Interactive), not just launch time
⚡ Track performance with percentiles, not averages
⚡ Eliminate unnecessary startup work
⚡ Run network requests in parallel whenever possible
⚡ Move heavy operations off the main thread
⚡ Build performance monitoring into CI/CD pipelines
⚡ Treat performance as a product feature, not a technical afterthought
What I liked most is that this wasn’t theoretical optimization. The article shows real-world examples where reordering requests, caching data, and removing startup bottlenecks reduced TTI by seconds and significantly improved user experience.
As mobile developers, we often focus on adding features.
But users don’t notice your best feature if they’re still waiting for the app to become usable.
Performance is a feature.
🔗 Worth reading for every iOS, Android, and mobile engineer working on apps at scale.
Source: <a href="https://medium.com/indriver-tech/milliseconds-make-millions-how-and-why-to-speed-up-your-app-5c814a8e29ce" target="_blank" rel="noopener nofollow">https://medium.com/indriver-tech/milliseconds-make-millions-how-and-why-to-speed-up-your-app-5c814a8e29ce</a>
#iOS #Swift #MobileDevelopment #Performance #AppDevelopment #SoftwareEngineering #iOSDev #Engineering #Programming #TechLeadership #UserExperience #PerformanceOptimization #MobileApps

Heart

A community for iOS developers to learn, build, and grow together. Share Swift, SwiftUI, UIKit, Xcode tips, architecture, AI, app development insights, career advice, and the latest Apple ecosystem updates.

iOS Developers

A summary of key lessons from inDrive's iOS team on app performance optimization. The central finding: reducing a critical user flow by 750ms led to a 1.5% increase in orders, demonstrating that performance directly impacts revenue at scale. Key recommendations include measuring TTI (Time To Interactive) rather than just launch time, tracking performance with percentiles instead of averages, eliminating unnecessary startup work, parallelizing network requests, offloading heavy operations from the main thread, and integrating performance monitoring into CI/CD pipelines. The core message is that performance should be treated as a product feature, not a technical afterthought.

Milliseconds matter more than most engineers realize.

The 750ms → 1.5% more orders result is impressive, but it made me think about priorities.
If you could only invest engineering time in one area, what would you choose?
• Faster startup/TTI
• Eliminating UI hangs and jank
• Faster API responses
• Shipping new features
At what point does performance optimization deliver more business value than building the next feature?

A very important post! The idea can be extended to any sort of friction, we want our user to have the smoothest flow possible from app start to action finish. This means speed of course but also clear UX with one clear call to action in every step for instance.

Made my career reducing time. I am a performance fanatic. But mostly it’s just the question of why is it taking this long? Pursuit of this question ends up fixing mostly any bottleneck in any system. I remember bring down a regression suit in a financial company from 7 days to 5 hours just by automation and parallelism. Then reducing the time on another transaction in another company from 16 secs to 35 ms. Then recently worked on another bottleneck where an ML pipeline used to take 56 hours to 3 hours.
You don’t really know once you start that what will be the uplift you will end up with, it’s mostly just reducing by any amount to be honest. But digging deeper, burning the midnight oil and you end up with results that shock you. You realise that people don’t really know what they are doing, which I think is sad but also a good thing as that leaves people like me with puzzles that are really satisfying and also makes normies go ape shit on how the F did you do that.