Dear PM Advisor. May 6, 2013

Dear PM Advisor,

I'm trying to crash my project wherever possible but I'm seeing resistance in adding team members on tasks. I know the old adage of 'You can't ask nine women to give birth to a baby and complete it in one month,' but I'm not doing that. The tasks I'm crashing are things like writing and testing code, and conducting interviews. Where is the problem?

Crash victim in North Carolina.

Dear Crash Victim,

Crashing tasks, adding resources to complete it faster sometimes runs into the law of diminishing returns. This usually comes into play when your company constrains your project on cost. If 'Money is no Object' you can crash all you like and spend thousands of dollars reducing the timeline by a day. You don't have that luxury.

Here is a graph of the law:

The theory works like this. If you have a fixed piece of capital, like a machine, you can add workers to it to achieve higher returns. Run three shifts, run during breaks, run it on weekends and you get a better return on the investment. Of course, if you apply ten people to the machine, they will have to start lining up to take turns on it so the additional investment is providing you with a lower return. Even before that point, you stop having time to maintain the equipment and it breaks down.

How that applies to Project Management is this. If you initially assigned one person to code part-time for four weeks and that task ended up on the critical path, you can speed it up by assigning her full-time to the task. Because you lose the interruptions, the task can theoretically be completed in less than one week. So we are on the left side of the curve moving up the returns axis.

Adding a second person to the coding may move you further up the curve or over the other side, depending on your company culture. I met one company who always uses two people to a computer so that they are overseeing each other's work and learning from each other's style. Other companies believe that two programmers get in each other's way and recode unnecessarily costing more time. In that case they are moving down the right side of the curve.

Your job is to determine where that maximum point is and avoid going past that point. On the curve shown above, they recommend not wasting time trying for that last 10% of improvement. Just keep adding resources until the improvement slows and stay there.

Good luck,

PM Advisor.

Send your questions to bfieggen@gmail.com