Have you ever had trouble pinning someone down to an estimate? Maybe a relatively minor risk was blown out of proportion? Or had actual hours be vastly different than the original estimate causing ROI to be, well, messy?

When faced with any level of uncertainty, people will often reach for the answer that will get them in the least trouble. That's human nature. But there's a way to make the vague unease pay off in accuracy and information - it's called Confidence Level.

When you ask for estimates (or a risk assessment), ask for the estimator's level of confidence in their estimate. Use percentages - 100% means they have total confidence in their estimate (or they really don't know what they're doing - you'll rarely get a 100% level). You can take it from there.

If the confidence level is low, ask why. Often there is missing information and the estimator is assuming the worst to be sure everything is covered. Then you have two choices - track down the missing information, or inform your client (or manager, whomever needs to make a decision based on the estimate) that the estimate is necessarily high, but if the missing information identified is provided a more accurate estimate will be possible.

If confidence level is low for something like quality, you know you have a problem. The same method can be applied to risk assessments - how likely does the assessor think a certain risk is to be realized?

Armed with this information you can zero in on good estimates, identify potential problem areas or hazards, and get the folks giving estimates and assessments comfortable with that part of their job.