How many times have you heard your estimate is ambiguous? How do you validate your estimate? An estimate is bounced off the actual time testing really took by how many testing hours were logged.
- Actual Time logged / Estimate x 100 = XYZ%
- Where if XYZ% is <=69% then the estimate given was to large compared to the actual
- Where if XYZ% is between 70% and 130% the estimate is very close to the actual
- Where if XYZ% is >=131% the estimate given was to small compared to the actual
This is a good baseline/starting point on estimates versus actuals. These would have to be tracked over a number of projects, releases, time-frames before determining any necessary changes. In the following graph we have the exponential lines crossing and close together, meaning our Estimates versus Actuals strategy has been successful and we should continue to use it.
In the following graph we have the exponential lines extremely far apart from each other, meaning our Estimates versus Actuals strategy has not been successful and we may want to revisit where we are going wrong. The estimates could not be being created properly or you may want to look at changing the estimating template itself (the formulas) to better tailor to your environment.
QA Soft-e 