Moscow Rules


A quantitative exposé



By Dr. Eduardo Miranda

Carnegie Mellon University

Pittsburgh, PA, USA


This article analyzes the performance of the MoSCoW method to deliver all features in each of its categories: Must Have, Should Have and Could Have using Monte Carlo simulation. The analysis shows that under MoSCoW rules, a team ought to be able to deliver all Must Have features for underestimations of up to 100% with very high probability. The conclusions reached are important for developers as well as for project sponsors to know how much faith to put on any commitments made.

Keywords: Agile planning, release planning, requirements prioritization, feature buffers, MosCoW method


MoSCoW rules [1], also known as feature buffers [2], is a popular method to give predictability to projects with incremental deliveries. The method does this by establishing four categories of features: Must Have, Should Have, Could Have and Won’t Have, from where the MoSCoW acronym is coined.  Each of the first three categories is allocated a fraction of the development budget, typically 60, 20 and 20 percent, and features assigned to them according to the preferences[1] of the product owner until the allocated budgets are exhausted by subtracting from them, the development effort estimated for each feature assigned to the category. By not starting work in a lower preference category until all the work in the more preferred ones have been completed, the method effectively creates a buffer or management reserve of 40% for the Must Have features, and of 20% for those in the Should Have category. These buffers increase the confidence that all features in those categories will be delivered by the project completion date. As all the development budget is allocated by the method, there are no white spaces in the plan, which together with incentive contracts, makes the method palatable to sponsors and management.

Knowing how much confidence to place in the delivery of features in a given category is an important concern for developers and sponsors alike. For developers it helps in formulating plans consistent with the organization’s risk appetite, making promises they can keep, and in calculating the price of incentives in contracts as well as the risk of incurring penalties, should these exist. For sponsors, it informs them the likelihood the features promised will be delivered, so they, in turn, can make realistic plans based on it.

To this purpose, the article will explore:

  1. The probabilities of delivering all the features in each of the categories: Must Have, Should Have and Could Have, under varying levels of under and overestimation of the features’ development efforts
  2. The impact of features’ sizes, dominance, number of features, and correlation between development efforts in said probabilities
  3. The effect of budget allocations other than the customary 60/20/20 on them.

To calculate the probabilities of delivery (PoDs) we need to make suitable assumptions about the distribution of the efforts required to develop each feature since the single point estimate used in the MoSCoW method are insufficient to characterize them.

In this article, those assumptions are derived from two scenarios: a low confidence estimates scenario used to establish worst case[2] PoDs and a typical estimates scenario used to calculate less conservative PoDs.

The potential efforts required and the corresponding PoDs, are calculated using Monte Carlo simulations [3], [4] to stochastic ally add the efforts consumed by each feature to be developed.

The rest of the paper is organized as follows: Section 2 provides an introduction to the MoSCoW method, Section 3 introduces the Monte Carlo simulation technique and describes the calculations used for the interested reader, Section 4 discusses the two scenarios used in the calculations, Section 5 analyzes the main factors affecting the method’s performance, Section 6 discuss the method’s effectiveness in each of the scenarios and Section 7 summarizes the results obtained.


To read entire article, click here

Editor’s note: Second Editions are previously published papers that have continued relevance in today’s project management world, or which were originally published in conference proceedings or in a language other than English.  Original publication acknowledged; authors retain copyright.  This paper was originally in Proceedings, 23rd International Conference on Agile Software Development, XP 2022: Agile Processes in Software Engineering and Extreme Programming, Copenhagen, Denmark, June 13–17, 2022. Proceedings published by Springer.  It is republished here with the author’s permission.

How to cite this paper: Miranda, E. (2023. 2022). Moscow Rules: A quantitative exposé; originally presented at the 23rd International Conference on Agile Software Development, XP 2022: Agile Processes in Software Engineering and Extreme Programming, Copenhagen, Denmark, June 13–17, 2022; republished in the PM World Journal, Vol. XII, Issue III, March 2023. Available online at https://pmworldjournal.com/wp-content/uploads/2023/03/pmwj127-Mar2023-Miranda-Moscow-Rules-a-quantitative-expose-2nd-ed.pdf

About the Author

Dr. Eduardo Miranda

Pennsylvania, USA


 Dr. Eduardo Miranda is Associate Teaching Professor at Carnegie Mellon University where he teaches courses in project management and agile software development at the Master of Software Engineering Program and at the Tepper School of Business. Dr. Miranda’s areas of interest include project management, quality and process improvement.

Before joining Carnegie Mellon, Dr. Miranda worked for Ericsson where he was instrumental in implementing Project Management Offices (PMO) and improving project management and estimation practices. His work is reflected in the book “Running the Successful Hi-Tech Project Office” published by Artech House in March 2003.

Dr. Miranda holds a PhD. in Software Engineering from the École de Technologie Supérieure, Montreal and Masters degrees in Project Management and Engineering from the University of Linköping, Sweden, and Ottawa, Canada respectively and a Bachelor of Science from the University of Buenos Aires, Argentina. He has published over fifteen papers in software development methodologies, estimation and project management.

Dr. Miranda is a certified Project Management Professional and a Senior Member of the IEEE. He can be contacted at mirandae @ andrew.cmu.edu.

For more, visit the author’s website at https://mse.isri.cmu.edu/facstaff/faculty1/core-faculty/miranda-eduardo.html

[1] These preferences might induce dependencies that need to be addressed by the team, either by incorporating lower preference features in the higher categories or by doing additional work to mock the missing capabilities

[2] Worst case, means that if some of the assumptions associated with the scenario were to change, the probability of delivering within budget would increase