back
|23 May 2017|Klaus Leopold

Traffic Light, Reloaded!

2

There is an interesting traffic light system at Ultimate Software in Florida. The traffic light logic functions based on a probabilistic forecast. If the probability that the team will work off the backlog by the release date is at least 90 percent, the light is green. If the probability drops to between 70 and 90 percent, the light turns yellow. A probability of 70 percent or less means the light goes to red. This offers the big advantage of always being able to, especially at an early stage, know the actual status of a project with full transparency. Nobody can sugarcoat the state of things—only what is finished can make the traffic light greener. By the way, it’s worth your time to read the Ultimate Case Study by Steve Reid, Prateek Singh and Dan Vacanti.
As you can probably imagine, forecasting is always a topic often discussed in my trainings. You can twist it and turn it as much as you like, but people like project traffic lights, despite overwhelmingly positive experiences occurring only in the rarest of cases.
I can understand this, and exactly because the traffic light is so well-liked, I would like to turn it on its head. The difficulty is (and actually completely skewed) that you start a project at green, although the highest uncertainty exists at this point. To make things worse, the incentive for a project manager is linked to the “green” status. The longer you can keep things on green, the better the project leader. When the difficulties can no longer be covered up, the light jumps to red—two weeks before the project ends. Naturally, now you have no time left to react and the project will have been driven into the wall.
Which is what happens, practically speaking, in most projects: Because the project starts on green, those responsible continuously try to push the risk towards the end of the project. Like I said, the beginning of the project is the phase with the most uncertainty. So, when we absolutely need to have the traffic light, it should start at red. This means, in particular, working and closely inspecting the work frequently so that the traffic light jumps to green and stays green. Progress, as well problems, will be apparent in advance and the risks can be dealt with accordingly.
What do you think? What’s keeping us from turning the traffic light upside down?

Emile
24 May 2017 13:52

Hi Klaus, thanks for the post! I intuitively agree with you that traffic lights is an emotive way to understand risk, and that we should start with a red light.
What if we base the state of the light not on forecasts, but on past data (yesterday’s weather), and then also normalised for time (like the ideal burndown you often see on burndown charts – the line that says “you ought to be here now”). If you’re above the ideal burndown mark in a certain range you get red, vice versa you get green. And yellow for in between. Not sure that this solves the same kind of problem, so curious to hear your thoughts.

Klaus Leopold
24 May 2017 16:20

I think that’s pretty much what I would like to see. Ultimate Software does continuous forecasting based on historical data. When historical data says that we can finish 10 to 15 features per month and a team commits 20 that’s clearly red! For companies that are not yet ready to run continuous forecasting, I would like to find out how to switch the traffic lights in classic projects to achieve a similar effect to continuous forecasting.

This conversation lacks your voice:
Your E-Mail Address will not be published.