Predicting effort.
WHY ESTIMATES ARE WRONG
Unknowns are discovered during the work.
WHAT TO DO ABOUT IT
Break work into pieces small enough to understand.
WHAT PEOPLE OMIT
Migrations and data handling Error paths and edge cases Tests Integration with anything external Deployment and configuration Review and revision
WHAT TAKES LONGEST, CONSISTENTLY
Anything depending on a system you do not control.
WHY
Their behaviour, availability and documentation are all outside your influence.
WHAT TO DO ABOUT IT
Investigate the integration before estimating anything that depends on it.
WHAT TO STATE
Assumptions, explicitly.
WHY
Unstated conditions become disputes.
WHAT TO PROVIDE
A range, not a single figure.
WHAT TO FLAG
Anything you do not understand well enough to estimate.
WHAT TO PROPOSE FOR THOSE
A brief investigation first, estimated separately.
WHAT TO TRACK
Actual against estimated.
WHAT YOU WILL FIND
A consistent bias, correctable once known.
