Dear PM Advisor,
How can I predict the duration of a truly untried activity? In R&D we are asked to develop a product that has never been developed before. What can I base my estimate on?
Flummoxed in Phoenix
I once worked for a Director of R&D who dealt with this problem by getting on his high horse and telling those who asked for schedules: 'You can't predict discovery!' While I genuinely liked and respected this man I didn't buy into this philosophy. While that is fine in the pure Research part of R&D, it doesn't fly in Product Development where we need to determine when activities will complete so that the other parts of the project can be coordinated with the development.
My version of this is an extension of his. 'You can't predict discovery but you can manage it.' Whenever confronted with a truly unknown activity, I list a series of assumptions and questions to be answered with it and ask management for a certain duration of time that seems logical to me and my team. I don't promise that I will find the answer in that time, just that they will give me that amount of time to answer the questions and test the assumptions and, either complete the activity or have enough information to predict when the activity will complete in the next iteration.
The most extreme case of this was at a company where the yarn used to knit or weave the medical device grafts was being discontinued by our current supplier and needed to be replaced with a product that performed identically to the existing product that had been in production for 40 years.
There was an 11 week process between the raw yarn arriving the involved spinning, processing, knitting, weaving, coating, sterilizing and many other steps before the product could be tested to see if it performed the same. We could test multiple versions of the yarn to get results and then could iterate until we found a version of the yarn that worked exactly the same as the current yarn. But how many versions and iterations would it take before we found our solution?
After lengthy discussion with my team we settled on six versions at a time with six iterations before we found our answer. This resulted in a contract with management for two years and two million dollars of effort to develop the solution. In each iteration we would find the answers to multiple questions and test assumptions and narrow down the choices until we reached the version of yarn that worked in all cases.
Why six? Mostly because I am hesitant to use five, simply because he humans have five fingers on each hand. And it seemed like we needed about that many to be successful.
Send your questions to Bruce@RoundTablePM.com