I’ve frequently said that project management is about bringing predictability to uncertainty and while a lot of my writing focuses on managing the impacts of uncertainty through effective risk management, estimation is another area where uncertainty needs to be addressed.
If you are fortunate enough to be managing projects where sufficient historical data exists to account for nearly all sources of variation then I envy you.
For the rest of us, here are some estimation principles which should be applicable regardless of the project’s domain or specific context.
- Use more than one estimation method: Consider this a form of hedging your bets. If you derive close to the same estimated value using different techniques, the likelihood that your estimates are significantly incorrect drops. Consider combining bottom-up and top-down methods.
- Document and share assumptions related to the estimates: It’s always a good idea to have a second pair of eyes review estimates, but without understanding the underlying assumptions which those estimates were based on, it can be very difficult to provide quality feedback. If the assumptions are not documented, it eliminates your ability to identify proactively when internal or external changes might invalidate the assumptions & their associated estimates.
- An estimate should never be provided as a single value: Without some understanding of the variation in the stated estimate, it’s impossible to manage the uncertainty resulting from it. This variation could be stated as a range or a confidence level.
- Discourage multitasking during estimation: Estimation takes significant mental effort and if the subject matter experts involved are distracted, the quality of your estimates will suffer. Consider breaking up the estimation process into more short meetings as one way to overcome this. Also, avoid estimation right after lunch time or at the end of the working day as your team might be less focused.
- Evaluate estimates across multiple dimensions: Just as it is helpful to use more than one estimation technique, it’s also valuable to analyze estimates in different ways to try to identify flaws. For example bottom-up activity estimation will give us detailed effort data at the work package level. But aggregating these estimates by resource role or by project phase will enable us to evaluate whether the ratio between roles or phases is logical.
- Remove the person from the evaluation: It helps to have peer-level reviews of estimates without knowing who produced them. This sounds wrong, but no subject matter expert is omniscient and our positive biases about them might cause us to unconsciously miss estimation flaws.
Improving estimation is not a silver bullet to slay all sources of project schedule or cost variation but elevating project management capability is about evolution, not revolution!
(Note: this article was originally accurately estimated and risk reduced in July 2014 on kbondale.wordpress.com)