Don't Gamble, Experiment Instead

My hackathon team recently prototyped the feasibility of my idea to use machine learning to customize an anonymous web site browsing experience. Flush from the excitement of proving that it *can* be done, the team was eager to tackle the question of *how* it must be done properly.

Then the team started to hate me. Once a visionary, I had suddenly turned into a wet blanket. "Look", I told my team, "when we prove that something can be done, the next question is whether it *should* be done". By that point they already had their minds set on figuring out *how* it must be done. I'm sure they still hate me.

Akin to when we dip our toes in the bath water to test the temperature, we might use experimentation and the scientific method to weed out the bad ideas from the good ones before going all in.

Use the following project inquiry and cost pyramid to uncover bad ideas early and to fail cheaply.

• Project #1 - Can we do it? ($)
• Project #2 - Should we do it? ($$)
• Project #3 - How must we do it? ($$$)
• Project #4 - Did we really do it? ($$$$)
• Project #5 - How do we scale it? ($$$$$)

With any new idea start at project #1 and only proceed to the next step if the previous one was positive. Skip a step at your own peril. As we all know, time is money. So if we are to be thrifty in the early projects, then they must be fast, focused and honest.

Money saved by abandoning bad ideas early can then be allocated to the good ideas that have made it through the pyramid or be reinvested in funding experiments for other new ideas. Doing otherwise is no different than putting all the chips on red.

About the Author

Byron Jung is a proven and certified specialist in high-performance project management that is disciplined and agile.