The key to good estimation is not about providing accurate estimates. It’s about providing your environment with as reliable forecasts as you can.
In product development (or other forms of design, development or change) that is best achieved with using a couple of pages from Fredrick Taylors books on Scientific Management. Just with the one very notable exception; we should clock the work, not the worker!
Design and development are non-repetitive as opposed to the manufacturing world Taylor first applied his techniques to. Even though activities seems to be repetitive, it’s the one-off nature of creating and developing the product that your stakeholders care about and want the forecast for.
So, always measure how long time the work takes to complete instead of spending all the time trying to estimate how much effort would be required by the workers. Use lead time data to provide predictability for your customers, users and other stakeholders.
Using the advice from Daniel Kahneman (Professor and Nobel Prize laureate) on how to avoid “The Planning Fallacy”:
- Always start your forecasts with the baseline of actuals from the most similar activity or delivery that you have recently completed. Then very carefully adjust for the particulars of the object in question.
- First state your own prediction, then compare your estimate with others in a small group to calibrate everyone’s understanding without group-think.