Responsible Estimating

My friend Timothy Johnson over at Carpe Factum published some very interesting musings on estimation in project management and why such forecasts (which is what, in reality, any project plan is) go awry. One thing I noticed in his post is the example of the impatient executive demanding a completely accurate forecast well in advance of not only the situation at hand, but also before any further relevant and definitive data appeared to bolster the forecast and/or its assumptions, or change it significantly to adjust for the new findings.

One thing I learned while studying lean software development is the advice to make decisions at the last responsible moment. That is, while we don't delay making the decisions that must be made to keep projects progressing, we don't make them until, given other constraints present, we have to. This allows for further collection of facts and data to fine-tune decisions, or scrap where we are and take a different tack entirely. The word responsible is very important in this concept. If we wait too long, and the situation worsens severely or opportunity passes us by, that is irresponsible, bad practice, and definitely not recommended.

An interesting question emerges from this concept: how do we know what is the last responsible moment, or not? My advise here is to examine your desired outcomes from the project very carefully and map out what dates are critical to the effort. Then, you can work backward and develop a calendar of "last responsible moments" to make significant decisions, and ensure that any data and other information needed to make the calls is collected and available. You make the decisions on or very near that calendar based on information you then have, not what you are waiting for or haven't received. And certainly not, if you can swing it, from an executive that wants "results" at the speed in which he gets his burger and fries from the drive-thru.

Veteran project managers know a couple of things about estimating: a) they are often viewed as reality by management; and b) they are usually demanded well before enough data appears to give the work credibility. If the last responsible moment is too soon in the process, the forecast will have a very low probability of being accurate to any degree. In order to pave the way for this approach, you may want to share the results of the decision-making-date exercise mentioned above with your sponsors and stakeholders before the project starts to execute, to prep them for the technique and why waiting until you must make a decision is the most appropriate tactic, given the benefits and expected outcomes.

Comments

[Timothy Johnson]

Great post! I think it's important to differentiate (as you have) decision procrastination vs. reducing the cone of uncertainty. Getting as much information as you can to make the best decision mitigates the risk of having to revisit the decision later.