Story points feel like astrology at this point. Anyone actually solved estimation?
I’ve been on five different teams in six years. Every team has the same conversation every two weeks: how many points is this? Nobody knows. We guess. We’re wrong. Repeat.
The thing that frustrates me most is that we have all the data to do this better. We have years of git history showing which parts of the codebase are complex. We have Jira history showing how long similar tasks actually took. We know who on the team has deep knowledge of which modules.
But none of this gets used during planning. We still sit in a room and play cards.
I’ve been thinking about what it would look like if planning used actual data instead of intuition - not just velocity metrics, but real signals: codebase complexity at the file level, individual developer patterns, historical risk of specific modules.
Honest questions:
1. Has your team ever tried data-driven estimation? What happened?
2. What would make you actually trust an automated estimate?
3. Is the real problem the estimates themselves, or is it that nobody respects them anyway?
4. Would you pay for a tool that improved planning accuracy if it could prove it with your own historical data?
Not a product pitch. I’m at the research stage and want to hear from people actually doing this work before I build anything.