Data Science: The Hard Parts

Chapter 5. Building Business Cases

Learning to write a business case for a model or experiment is a critical skill that data scientists ought to develop. Not only can it help you quickly learn whether a new project is worth your time and effort, but it can also help you gain stakeholder buy-in. Moreover, it is consistent with the type of extreme ownership that will make you shine.

Business cases can be as complex as you want, but many times you can come up with good-enough estimates. In this chapter, I will go through the fundamentals of business case creation.

Some Principles to Construct Business Cases

While every business case is different, most can be built using the same underlying principles: you compare making a decision or not, calculate costs and benefits of all options, consider only incremental changes, and many times you can account only for unit economics.

Decisions

Business cases are most commonly built to evaluate a new decision that is under consideration, be it a new campaign, a change in a lever, or any other decision.

Costs, benefits, and breakeven

Most interesting decisions have trade-offs. A critical starting point is to enumerate the main costs and benefits derived from the decision. The business case will be built around net benefits calculated as the monetary difference between benefits and costs. Breakeven is synonymous with having zero net benefits and serves as the limit case, or worst-case scenario, for your decision.

Incrementality

A good business ...