A forecast needs feedback from what really happened and an honest representation of uncertainty.
What to look for
- 01
Break work into pieces whose completion can be observed.
- 02
Learn from the difference between expectations and outcomes.
- 03
Treat the prediction as a range that changes with evidence.
Keep in mind
The article’s individual velocity model and tooling are dated. Team composition, AI tools, and changing work can make old histories poor predictors.
Try this next
Inspect a recently completed project. Separate implementation effort, review, dependencies, and waiting; identify which uncertainty most affected the outcome.
Get the planning & delivery templateNotes and exercises by Awesome CTO. The original work belongs to Joel Spolsky.