← Back to blog
ManagementApril 19, 2022· 5 min read
How we estimate projects so we don't go negative
Three estimation models we tried since 2019. Which works for a site, which for an integration, and why hourly is almost always a bad idea.
Creastra Digest
- T-shirt sizes for discovery, story points for build
- Fixed price only if scope < 4 weeks and 100% clear
- Reserve 25% for unknown unknowns
In 2019 we estimated our first commercial project at "80 hours". It took 240. We've stopped estimating in hours since.
T-shirt sizes for discovery
At discovery you don't know enough to give a number. But S/M/L/XL is easy: S — known task, XL — "team, we need a month just to scope it". Client sees price order of magnitude, we don't lie.
Reserve for the unknown
Every estimate gets +25% on top. That isn't padding — it's insurance against what you didn't account for. Over 5 years the reserve was fully consumed on 70% of projects.