Search This Blog

Tuesday, October 12, 2010

Your Estimates Are Wrong!

I have been burned more times than I can count by bad estimates. What can a project manager do to help ensure the accuracy of estimates?  First we should understand the basics behind the estimating process (there are many more than I have listed here). Some are:

• The more unique , complex, or larger the project, the more of a challenge it will be to get good estimates

• Estimates are only as good as the estimator is at predicting the future

• "Padded" estimates are not always bad as long as the padding is communicated (... and as long as the Project Manager is the one doing the "padding")

• An estimate is not a bid

• Estimates using sound estimating practices, performed by experienced estimators from clear specifications should never be negotiated

• Ballpark estimates are guesses and should be treated as much by the project team, management, and the project sponsor

Other items to consider when estimating are:

• Ensure the statement of work or contract is clear and understood by the person(s) doing the estimates

• Ensure that a schedule or mandated date doesn't drive the estimating thought process

• Include Risk Management in the estimating process

• Ensure that estimates take into account the skill level(s) of the person(s) that will do the work

• If your work breakdown structure (WBS) is flawed, your estimates will be inaccurate

Accurate estimating is an art and a science. The estimator (or team) must take into account historical data from past projects, the team's knowledge and experience, the project risks, the statement of work and other project information to make the best estimate possible.

Keep in mind when planning your project that estimates aren't hard and fast numbers. They are guesses, however they should be very good guesses if you have good estimators and are following tried and true estimating practices.


Jean said...

I couldn't agree more on that, especially including Risk Management in the process.

Dave said...

It's discouraging to discover that project managers are still hung up about "accurate estimates." Point #2 in the post, that estimates are only as good as the estimator, reflects a blame-oriented mentality and anachronistic thinking about the purpose of estimates.

No one can predict the future. No one. All estimates are wrong. They are not wrong because of the estimation methodology or the skills of the estimator. They are wrong because they are predictions of the future. In those rare cases when the estimate matches the actual time expended, it is the result of operation of random chance.

High-level estimates whose purpose is to inform go / no-go decisions at a program and project level are necessary. What we must understand is that the best we will ever be able to provide is a "ballpark" estimate that represents our best guess at a point in time. If results or progress are measured against that sort of point-in-time guess, people will be perpetually frustrated and dissatisfied, and will worry about the "accuracy" of the estimates.

Too often I see project managers preoccupied with making the projected timeline match the originally-estimated timeline. They lose focus on the business goals of the project.

It is not important that point-in-time guesses should line up neatly with unpredictable future realities. It is not even a meaningful goal. What is important is that we have enough confidence to make a go / no-go decision at a high level.

At a finer-grained level, such as time-based estimates of individual technical tasks, the very notion of "accurate estimates" is flawed. We should focus instead on making steady progress toward our goal, and making small course corrections throughout the project to ensure we stay on track.

Estimates are not the way to do this. Surely the overall "success" of IT projects throughout the 1970s - 1990s demonstrates this.

We don't need to get better at doing thing. Instead, we need to stop doing things that don't contribute to achieving our goals, even if those things are traditionally assumed to be part and parcel of project management. Forget about tracking estimates versus actuals. Instead, track incremental results against the stated goals of the project.

FWIW, I agree that risk management is important. But "estimation" and "risk management" are orthogonal.