The part everyone gets wrong.
WHY ESTIMATES ARE USUALLY WRONG
People estimate the work, not the elapsed time, and they estimate the good case.
WHAT THE DIFFERENCE IS
Two days of work may take two weeks, because of other commitments and waiting.
WHAT TO ESTIMATE SEPARATELY
Effort: how much work
Duration: how long until it is finished
WHAT AFFECTS DURATION
Availability of the person Waiting for others Review and approval cycles Interruptions
WHAT TO ASK THE PERSON DOING THE WORK
Their estimate, rather than imposing one.
WHY
They know, and imposed estimates are not owned.
WHAT TO ASK THEM FOR
A best case, a likely case and a bad case.
WHY THREE
The spread reveals how uncertain they are.
WHAT A WIDE SPREAD MEANS
The work is not understood well enough to estimate.
WHAT TO DO ABOUT IT
Break it down further, or investigate first.
WHAT TO ADD
Contingency, explicitly, at the project level.
WHY AT PROJECT LEVEL
Padding individual tasks hides the buffer and it gets consumed anyway.
HOW MUCH
Proportionate to uncertainty; more for unfamiliar work.
WHAT TO NEVER DO
Cut estimates to fit a deadline and call it a plan.
WHY
It does not change how long the work takes; it only delays the discovery.
WHAT TO TRACK
Estimates against actuals.
WHY
It is the only way estimating improves.
