In this article:
1. Introduction to the "Monte Carlo - When" Simulation
1.1. Dataset Configuration
2. How Does This Chart Work?
3. What Is That Chart Telling Us?
4. Controls for This Chart - Customize Your View
You can find more information about the Analytics module in the dedicated article.
1. Introduction to the "Monte Carlo - When" Simulation
The "Monte Carlo - When" simulation does the opposite of the “Monte Carlo- How Many” simulation. The "Monte Carlo - When" chart aims to tell you when you can expect the team to finish a specific number of tasks that you are yet to start working on.
Based on how you have performed in the past, the "When" simulation can estimate how long it will take you to get a certain number of items (for example 100) done in the future. For example, you can use it if you have a product update scheduled for March 30th and you wish to know how many features you can finish by then.
The "Monte Carlo When" simulation consists of three charts:
- Throughput basis (1)
- Throughput Navigation (2)
- Monte Carlo (3)
The horizontal axis of the "Throughput Basis" (1) is a representation of time, while the vertical one shows the daily throughput. The high points of the charts represent the maximum throughput of any of the days shown in the chart, while the low points mark the minimum number of tasks that were completed. Hovering over the dots will give you summarized information about the number of tasks that have been completed that day. The line between the dots visualizes the rises and drops in your team's productivity.
The Throughput Navigator (2) allows you to zoom in and out of a specific interval within the selected time frame. This way you can generate different Monte Carlo simulations without resetting the general time frame each time.
Monte Carlo (3), the bottom chart, is a direct visualization of the results of the simulation in the form of a probabilistic distribution. The horizontal axis visualizes the total work items that will probably be completed within the selected future date. The vertical axis shows how many instances of a certain result occurred during the trials.
Dataset Configuration
To filter your data, use the Dataset Configuration menu on the left side of the simulation chart. You have to select at least one of the following fields — Start Date, End Date, or Created at, and at least one workflow.
There are two additional settings you can enable:
- Ignore the cycle time configuration for the selected workflow(s) — enable this if you want the system to disregard the cycle time settings for the workflow(s) you have selected.
- Ignore the block time in the queue columns — enable this if you don't want the system to account for the time cards were blocked while they were in queue columns.
2. How Does This Chart Work?
It has an identical structure to the "Monte Carlo - How Many" simulation and works in the same way.
(1) Imagine you have a backlog of 100 items. You have to enter this number in the dedicated field.
(2) Activate the percentiles to get the probabilistic view of the third chart.
(3) Run the Trials (up to 1 million).
3. What Is That Chart Telling Us?
The third chart will be able to tell you with a probability of 50%, 70%, 85%, and 90%, an approximate end date when your team will complete 100 items. The simulation shows you that there is an 85% probability that your team will complete the items by July 4th. If you want to be more certain go to the 95% percentile, where the date is July 7th.
Assign a probability to your estimations or deadlines. If you communicate the date without the probability you commit to a fixed date delivery and in knowledge work that is very often unrealistic.
Practical Tip:
- Note that if the characteristics of the team that generated the historical data changes (for example, a new team member joins the team), the simulation produced from the data might be no longer valid.
- You may doubt the results if your work items are different in size or some of them are too complex.
That's why it is so important to keep your cycle time smaller. You have to do right-sizing and try finishing work within a certain period of time. If a task is too big, break it down into a smaller chunk of work and flow it through the system. If you do this consistently, the forecast will be more accurate and your process much more predictable.
4. Controls for This Chart - Customize Your View
You can use the controls to change the "Monte Carlo When" simulation view and to apply additional filters. Please, check the short video below.
-
Simulation Control - select the date for which you want to run the simulation and the number of items that you need to complete by this date. If you anticipate changes that will most likely affect your daily throughput (e.g. an upcoming national holiday or new members joining the team), you can downscale or upscale the throughput to account for those changes. Allowed values are 0.1 to 10. The lower the scale value, the longer the duration. For example, a value of 1 assumes the following:
- All child cards are “normal,” meaning they are likely to take the same time as your historical data.
- The child cards are worked sequentially, one after the other. Learn more about the scale parameter.
- Throughput Chart type - choose between a line or a bar chart.
- Percentiles - select or deselect percentiles to get the probabilistic view of the chart.
- Layout - you can remove some of the charts and make only the simulation visible. Moreover, you can add a calendar that visualizes the start and end dates for your prediction.
- Item Filter - apply additional filters to run the simulation for tasks that have a specific location or a certain property: tag, type, etc. Choose values to select which items are available for the prediction of this chart. For example, let’s say that you are about to launch a new marketing campaign and have already made the breakdown of the tasks that need to be completed. You’ll probably want to know when the content for the campaign will be ready. To find it out, select the specific card type in the filter and the platform will exclude the rest from the simulation, thus giving you a more precise prediction based only on your previous content performance.