Do you use schedule or tracking tools to predict the future? None of them can predict the future. Worse, too many of them hide the past, especially the recent past.
A schedule or tracking tool can only predict one thing: the first possible date you cannot prove you cannot finish the project/story by. Let me unpack that. First, the prediction part:
- If you were honest in how long a story or task took and you counted overtime and delays, the tool might be able to predict a little about your cycle time.
- Then, assuming the tool knows how, it might be able to predict a range of possible dates for the next bit of work.
Now, the first date you cannot prove you cannot finish by. (Yes, I am sure there is a better way to say this. But the not-proving part is important.) Here's the not-proving part:
- Since the team has not yet done the work, the assumptions about the work might be wrong. (Does the team understand the work?)
- The tool can only guess based on the data it has. How optimistic is the person entering the data?
- Tools do not typically use the three-date range of possible, likely, pessimistic. Instead, the tool chooses one date: the possible date. That date is the first date the tool cannot prove the team will not be done—given the data that went into the tool. (Very few tools use Monte Carlo to get an even better distribution of possible dates.)
Where do teams get that data? Too often, it's top-down estimation or, just as bad, bottom-up estimation. And don't get me started about story points.
I know everyone wants prediction. Here's how you can get closer to something reasonable.
How to Improve Prediction
We can use cycle time and a range of dates to improve prediction. In addition, we know that:
- Teams finish work faster when they collaborate.
- Make stories small so the work is similar to that of the past. (That also helps maintain a similar accidental or essential complication of this next piece of work.
- Map the team's value stream so they can see their bottlenecks and delays.
In addition, rank-order the work. That allows the team to collaborate on just one (or two) things at a time.
Now, use the team's cycle time to predict when the team will finish this piece of work. Look for where the team has feedback loops. Is it outside the “team's” control as in Unearthing Your Project's Delays? Is it because the team cooperates but does not collaborate? (See See and Resolve Team Dependencies, Part 1: Inside the Team to see how separating code review creates a cooperative but not collaborative team.)
Tools will not tell you when a team can finish something in a long roadmap. (See How to Predict A Little About the Future Work Without Long Intricate Plans for what you can do instead.)
Read More
- Predicting the Unpredictable: Pragmatic Approaches to Estimating Cost or Schedule
- Create Your Successful Agile Project
- Manage It! Your Guide to Modern, Pragmatic Project Management
- Project Lifecycles: How to Reduce Risks, Release Successful Products, and Increase Agility
And, good luck!