THIS ARTICLE IS PART OF OUR SERIES EXPLORING INNOVATIVE NEW PROJECT MANAGEMENT SOFTWARE, PLUG-INS, AND PLATFORMS. IN THIS SERIES OF ARTICLES, MUSTARDSEED PROJECT MANAGERS HIGHLIGHT THE BEST NEW TOOLS THEY’VE DISCOVERED TO SUPPORT EFFICIENT, FORWARD-THINKING PROJECT MANAGEMENT.
Schedule risk analysis (SRA) is a statistical tool providing mathematical insight into project risks. At its most basic, an SRA allows you to predict the likelihood of certain outcomes within your project schedule. Whether you are a project manager (PM), executive leader, or simply an interested project stakeholder, a Monte Carlo simulation, the most common SRA model, provides distinct tools that can assist in decision making.
Before performing a Monte Carlo simulation, you’ll need a few things:
A fully networked project schedule with a duration of 6+ months. Running a Monte Carlo simulation on a short project timeline is not ideal, as shorter schedules typically do not contain sufficient statistical variation to assess large schedule risks. If you do run an analysis on a short project, you’ll likely find that discerning weeks or months of timeline delays is simply not possible. To avoid this pitfall, try running Monte Carlo simulations primarily on projects with a timeline of six months or longer. It is also important to ensure your project schedule is fully networked along a critical path. A simple list of tasks with dates will not be effective in an SRA, as the software uses task interdependencies to predict the likelihood of particular outcomes. It is not possible to accurately predict risks for standalone, non-networked tasks.
An easy-to-use software. Because Monte Carlo is a statistical technique, you’ll need a software package with the built-in capabilities to perform your desired analysis. One fantastic tool available as an add-in for Microsoft Project users is the Barbecana Full Monte program. Full Monte is easy to install, user-friendly, and comprehensive. There are also other project management tools with built-in SRA capabilities, such as Primavera P6 and Open Plan Professional; however, we will focus on Full Monte in this article.
A clear set of inputs. Work with your project team to answer the questions below. You’ll need answers to these questions before you can run your calculations.
Input #1: Look at the current duration in your project schedule. Is it possible you could finish the project faster than what is shown in your schedule? Or might it take longer to complete than what’s shown? Assign percentages to each of these three scenarios: (1) the most realistic duration, as compared to your project schedule (e.g., “I think it will likely take 10 percent longer than what’s in my project schedule, so my most likely scenario is 110%); (2) the best-case scenario (e.g., “If everything goes perfectly, I think it’s feasible to finish this project 10 percent faster than what’s shown in my schedule, so my optimistic scenario is 90%); and (3) the worst-case scenario (e.g., “If we encounter a number of problems and delays, it’s possibly this project could take 50 percent longer than what my schedule is showing, so my pessimistic scenario is 150 percent).
Input #2: What is an appropriate confidence interval for this project? A confidence interval measures how confident you are in the assumptions you provided above for the total project duration percentages. For most purposes, Barbecana recommends that this field is left at 100%. A confidence interval lower than 100% produces additional pessimistic values within the simulation, which sometimes may be necessary, however this makes defending the results more difficult. Leave this at 100% unless you are in a very specific scenario or are an expert user of Monte Carlo SRAs.
Input #3: What type of distribution would your team like to use? A distribution determines how your data is plotted and analyzed. Different types of distribution are more or less conservative. For example, a triangular distribution is the most statistically conservative, and the most used. If you aren’t sure which distribution to select, opt for triangular as a starting point, or run the analysis using a few different types of distribution and compare your results. Some other distribution options are normal, uniform, and beta, but these should only be used in specific scenarios which we won’t dive into for this article’s purposes.
Input #4: How many iterations would your team like to run on the analysis? You can think of this as a probability model, similar to flipping a coin – the more times you flip a coin, the closer you’ll get to having a distribution that is exactly 50 percent heads and 50 percent tails. Monte Carlo works by running through your schedule multiple times to predict the likelihood of different paths. The more iterations you run, the more sensitive your results will be. Barbecana recommends one thousand iterations as a starting point for basic calculations, but MustardSeed project managers have used as few as 500 iterations to as many as 2,000 for sensitive, executive-level reporting.
Now that you’ve completed the above steps, you’re ready to start running your SRA and identifying where the major risks lie in your schedule. Begin by uploading your project schedule into Microsoft Project. Navigate to the Add-Ins tab and click on the Full Monte program. Input your duration and confidence intervals, then run your simulation. While your results will look different than the images we share here, we’ll walk you through how to understand your results below.
After inputting duration and confidence intervals, the first output you’ll see is the histogram (Figure 1).
In this example, you can see at the top of Figure 1 that the mean generated in this simulation is April 18th, 2018. The mean shows the most likely date when your project is expected to finish, based on the results of the Monte Carlo simulation. On the righthand side of the graph, you’ll see the probability that the project will finish early (lower on the y-axis) or late (higher on the y-axis). Reviewing this histogram can help address uncertainty around project delivery.
Another output of the Monte Carlo simulation that may be helpful is the tornado diagram (Figure 2).
In reviewing Figure 2, you can immediately see that Tasks 5, 8, and 6 carry the most risk because they have the longest bars in the tornado graph. Using this information, you might advise your project team to allocate additional resources to reduce the risks on those specific tasks. The tornado diagram is built into Full Monte – viewing this information just requires a quick toggle over to another tab after running the initial analysis.
While a Monte Carlo simulation provides additional information and output far beyond what is detailed here, the histogram and tornado diagram are an easy entry point into the world of SRA. Both diagrams provide an easy-to-understand overview of schedule risk, enabling project teams to engage in conversations and decision-making driven by data.
Barbecana, Full Monte - https://www.barbecana.com/full-monte/