Inputs

The worksheet only requires a few simple data points to create a forecast.

• Backlog items to simulate: This is the number of work items that you want to forecast a completion date for. Typically, this is the number of stories in your product backlog, or the number that need to be complete for a given feature or epic release.
• Simulation start date: This is the date when the team will begin work on the specified backlog. This defaults to today’s date, but users can override it to forecast projects not already underway.
• In the Measured takt times field, enter the number of takt times you have measured for the team, and under Takt Times (weekdays), enter what those takt times are. The worksheet will highlight the cells where it expects to see takt times in blue. For details on how to measure takt times, read the original blog post.

That’s it! As soon as your data is in place, your forecast has already run.

Results

Here, the workbook summarizes the simulations for you.

In the small table up top, the workbook tells you how many Simulations it was able to run – roughly the number of simulated work items divided by the size of the backlog you simulated. You can think of this as the resolution of your forecast: more simulations paint a more precise picture of the future. If you need more precision, just extend down the formulas in the Simulations tab by another thousand rows or so. The summary here should automatically detect the additional data.

The rest of the first table sketches the broad contours of the results: the Earliest, Latest and Median finish dates forecasted. I wouldn’t report any of these as “the forecast” if I were you, but they’re useful for quickly comparing the results of different scenarios.

The larger table is the heart of the forecast. For each week-long Predicted Date Range, the table shows how many simulations fell within that range (N Simulations) as well as the percentage of simulations that were completed by the end of that range or sooner (Confidence). You can read the example pictured, for instance, as “I forecast a 91% chance the team will complete its backlog by October 14, 2019.” See? You sound like an expert already.

The worksheet also plots the same data as a graph. This is useful for identifying more subtle features of the forecast. For instance, are the results tightly clustered around a likely date, or more spread out? It’s also a good way to tell if you have the level of precision you need. You want a clear “bell curve” shape like the one shown here. If there’s more noise in the picture, you may need to run more simulations.