So I tracked my estimates vs. actual time for 3 months. My accuracy: 47%. I was planning twice as much work as was physically possible, then wondering why I was always behind.
TimeBoxer is an iOS app I built to solve this: - Estimate before starting a task - Timer tracks actual duration - See your accuracy over time - Learn which task types you consistently misjudge
After 100+ tracked tasks, I discovered: - Bug fixes: Off by 3-5x (thought 1hr, actually 4-6hrs) - "Quick" refactors: Off by 4-6x - Code reviews: Off by 4x - Features I've built before: ~75% accurate
Now I estimate based on historical data instead of optimism. Hit 80% of my sprint estimates for the first time in my career.
Built natively in SwiftUI with Live Activities for Lock Screen timer. Free tier, $4.99/mo premium for full analytics.
The ADHD developer community has been particularly interested - time blindness makes estimation even harder, and having external data helps compensate.
Happy to answer questions about the tech stack, estimation patterns I found, or design decisions!