It Predicted when Our Projects Would End: How We Started Using Evidence-based Scheduling to Predict the Future

Posted on 2024-10-17

A few years ago I was working in a software engineering team. I came across Adrian Fittolani’s article, Agile Project Forecasting - The Monte Carlo Method. I ran it past the engineering team. It sounded too good to be true, but we decided to give it a try.

Our first project

We were near the end of a project. The engineering team was adamately unanimous. We’d be finished in 2 weeks, 3 weeks tops. We gathered historical data from previous work and ran monte carlo simulations to forecast when the project would end. 95% of the simluations said the project would end on May 3rd, 8 weeks. The team was incredulous. No way would it take that long.

The actual result. We finished the project on May 1, slightly under 8 weeks.

Our second project

We then applied it to a new project. After mapping out the project, we had the team estimate how long it would take and then we ran the simulations. The team was confident they could complete it in 6 weeks. The monte carlo simulation forecasted that it would take 12 weeks. They said impossible. There’s not that much work. The determined that they needed to beat the simulation’s forecast.

The actual result. The project took 12 weeks.

It worked!

We were surprised and intrigued. What if we could accurately predict our project end dates without having to do those long estimation exercises that everyone hated?

Monte was born

Monte started as the tool I created to run those early simulations. Monte makes it easy to input the historical data and run the simulations. It simulates the project 1,000,000 times in a couple of seconds and then reports the results in an easy to read chart. And bottom-line: it works.