When to act on data?

A cool diagram


There's a diagram Nick Tordoff and I use in our Data Conversations course, and which I've now started using in my Using Data to Make Business Decisions course. It shows what it might feel like to be a decision-maker who has to make a decision (or take an action) based on data, and it's a representation of how they'd decide whether or not to trust the data, and how they'd evaluate the risks and benefits associated with any action they might take.


Source: When to act on a correlation and when not to

The diagram was created by David Ritter, who is—or at least was—may still be—a consultant working for the Boston Consulting Group. It's basically a scatterplot, where the x-axis measures risks (more risks on the left) versus benefits (more benefits on the right), and where the y-axis measures the strength of the correlation in the data. High correlation at the top; low correlation at the bottom.

The further to the right and to the top of the diagram a decision-maker finds themself, the more likely they are to make an actual decision—take an action, even—based on the data.

This diagram is useful for data analysts because it enables decison-maker empathy. If you're a data analyst and you're preparing some data that will be used to make a decision or take an action, then you can be more systematic about predicting how a decision-maker might critique your data.

It's a great diagram, I find it extremely useful, but I want to make David Ritter's y-axis a bit more generic. I'd prefer it if the y-axis measured confidence in the data rather than confidence in a relationship. I've taken the diagram from an article in which Ritter was talking explicitly about correlation; but in the context in which I want to use this diagram, I'd rather it said: "Do we have confidence in the data? Has the right measure been chosen? Has it been defined correctly? Is the data accurate?"

So you can now imagine that—as a data analyst preparing your data for a meeting—you can go through various iterations of your data exhibits, each time trying to nudge the dot on the Ritter scatterplot further to the right on the x-axis, and further to the top on the y-axis.



[27 February 2023]


If you want to get notifications by email of new content on the Kurtosis website, please email info@kurtosis.co.uk and ask to be added to the email subscription list.