Agile Planning

Release forecasting: When might we finish?

Gather your historical sprint data and enter it in the box below. You can use either story points or simply use a work item count. 

Historical Data:

How much work is remaining? Again, use either story points remaining or simply use a work item count. You should also include any work in progress, i.e. partially done work that doesn’t meet the definition of done. 

Remaining work:

How many trials would you like to run? The more trials you run the longer the simulation will take to run.

Select number of trials to run:

Back Story

You’re eleven sprints in when your boss asks “based on the remaining known work, when will the MVP be ready for release?”.

You’ve still got 61 stories to get through in your backlog. That is, 61 stories that you’re aware of at this moment in time. More stories could emerge as we iterate our way through our sprint plan but let’s put that aside for now.

You set to work on answering your boss’s question.

Your team have delivered a total of 847 points over eleven sprints. We have 581 points remaining to hit MVP. This means we’re 59% finished. If we’re averaging 77 points per sprint then you’d be tempted to estimate you’ll need another 8 sprints to complete the work. 

Looking at this probabilistically, using the average will only give you a 15%-30% chance of delivering the remaining scope with an additional 8 sprints. Are you sure you’re comfortable giving your boss an estimate based on such a low level of confidence?

Use you own data in the monte-carlo simulator above or use the data below to get familiar with confidence levels.

Story points delivered in each sprint – from Sprint 1 to Sprint 11: 56, 21, 33, 87, 31, 84, 85, 150, 58, 91, 103, 41, 62, 55
Story points remaining: 581

How does this monte-carlo simulation work?

The simulator runs thousands of sprint scenarios based on your historical data to compute the probability of an outcome as a confidence level. The algorithm works as follows:

  1. pick a random number from the historical data
  2. Reduce the amount of remaining work by the number picked in step 1 – this step is simulating the completion of a single sprint
  3. Repeat until no work remaining
  4. Add up how many sprints were needed to complete the work

The simulator runs this process 10k, 20k, 50k, 100k, or 250k times. The percentiles of the results are presented in the table.

Additional Notes

  • Sprints and story points are used above as an example to appeal to the masses. The simulator can work just as well if you swap out the word “sprint” with days, weeks, or months and/or use work item count instead of points.  
  • 50% confidence is akin to the flip of a coin – 50/50 chance
  • The forecast is based on historical data, which represents your current/past ways of working. The forecast does not consider any potential improvements you may introduce to the team, nor can it predict catastrophic events. 
  • This is not a crystal ball or the silver bullet of planning.

About Ian Carroll

Ian consults, coaches, trains and speaks on all topics related to Lean, Kanban and Agile software development.

Connect with Ian at the following


Further information...

The content of this article is just the tip of the iceburg. To dive deeper into any of these case studies or concepts join our 2 day Portfolio Kanban live online course.

Application for a free Agile Coaching session

I would like to speak with an advisor