Just How Badly Do You Want A Number?
Thomas M Cagley Jr.
An audio version of this essay can be found in the Software Process and Measurement Cast 38 (www.spamcast.net)
Every project begins with a prediction of how much it will cost and when it will be delivered. Project managers, as a rule, admit this behavior delivers results that are a mistake, can lead to perception problems and might actually warp space and time. Most projects I see begin with this sort of incantation. Why? The rationale for this seemingly irrational behavior is generated by many competing forces. It might be a reaction to a market deadline, a requirement to secure funding or based on the mistaken impression that the project team actually knows what they need to know to complete the project. Every IT manager I know understands the fallacy of the initial estimate, however I know very few if any that will actually stand up and just say no. This behavior causes cognitive dissonance (stress caused by holding two contradictory ideas simultaneously), but it is assuaged by continuing to look for a solution to stop the insanity (or by giving up and embracing the dark side).
Why, if we all know this type of behavior is wrong and that the outcome of the behavior is rarely effective, do we continue to do it? Why do we feel compelled to act in a manner that is non-nominal? Do Information Technology (IT) managers and project managers have a touch of a victim complex? The driver for this behavior begins at the interface between IT and finance known as the budgeting or funding process. While not the root of all evil, financial procedures and thinking are at times at odds with many standard ideas for planning and estimating a project. Concepts like ROI and tax accruals have precise predictable financial definitions and calculations. Financial control and analysis requires a level of precision in reporting project data. The problem is that the level of precision is generally at odds with initial project estimates. There will always be a mismatch between finance’s needs and IT’s data needs, unless projects are being actively measured and are using mechanisms to continually access they need to know to complete the project.
Agile methods such as xP and Scrum attempt to make peace with the initial estimation conundrum by breaking projects into small bites and only making promises for each small bite prior to beginning the work for that bite. In Scrum terms, the sprint planning exercise is an operational illustration of how agile methods have used a short term planning horizon to address the ‘how much’ and ‘when’ questions. This does not address the need to know when the overall project will be done and how much it will cost. You will encounter the same problem we began this essay with if you extend the short term planning windows to the entire known project backlog by counting the number of sprints or iterations runs. .
Non-agile shops have been adopted other tools to deal with the vagaries of early estimation and the perceived need for precision. The estimation funnel is an example of other strategies. An estimation funnel helps enforce the understanding that the variance of any prediction is larger earlier in the project and reduces as project team learns what they need to deliver and how to deliver it.
In the end however, nature, our society, finance and our fellow project managers betray the quest for abandoning the initial fixed estimate. Nature’s betrayal comes in the form of the belief that Sir Isaac Newton had something to do with project management. We believe that each action has an equal and opposite reaction, or that if we do step A then outcome B will magically appear. Remember that the combination of humans and physics does not a straight line make. Projects are human ventures, and therefore are more akin to herding cats than the laws of physics. Society’s betrayal is driven by ingrained consumerism. As a consumer, you and I have an expectation that if we’re going to buy something, regardless of complexity, someone will tell us for how much to write the check. An example of this type of behavior can be seen in the general hullabaloo that occurs when NASA overruns a cost estimate for the space station, despite the incredible level of complexity. How can anyone know exactly how much a project of this type will cost when it begins? The finance department’s betrayal is through their need to plan, secure funding and to understand cash flow. Making a significant mistake in the financial arena will have disastrous consequences for any company’s value and potentially its ability to make payroll. Most importantly, project managers let themselves down by saying “yes” and giving the number to whomever asks. Why? They feel they must, if for no other reason that there is always someone who wants the job badly enough to say anything and managers want a number badly enough to believe the lie and drink the Kool-Aid. Failure to play the game is seem to be career limiting.
I believe we need to rethink our concept of initial estimates. To drive this change we will have to change both our vision of the world and that of other constituencies within our companies. We need to de-link estimation from other non-estimation behaviors (we are not shopping for an HDTV), we need to change how estimation the process works, we need to measure projects which means embracing concepts such as function points for size and a proxy for functional knowledge. But most of all we will need to set a standard of behavior for ourselves and our fellow project managers and follow it.
Recognize these issues? Leave a comment or drop me an email at spamcastinfo@gmail.com
