Engineering
Data storytelling matters as much as your model
Six simple habits that keep a well-scoped report from falling apart the moment your audience starts forming questions.
The meeting arrives. You are ready, the report is well scoped, and everything is going to go fine. Then the moment comes and, with every sentence out of your mouth, a question forms on your leader’s face. Why?
Six things that help
Still in the spirit of How to Lie with Statistics, here are a few simple things that can help:
- It is not always about the data, it is about how your audience understands it.
- “So, to make sure I follow, you mean that…?” saves you from a lot of misunderstandings.
- The obvious needs to be said, whether it is the main problem, how to solve it, or even what causes it.
- A concrete example can be worth more than a chart full of numbers. “1 in 5 proposals pay off” says more than “20% of the revenue forecast landed”.
- Whenever you can, simplify. If a number needs three sentences to be understood, a chart or an analogy is probably more effective.
- Connect the dots. Use small samples to illustrate, but always frame them against the whole.
In the end, adapt and test again
Adapt your communication for different audiences, and test again. Data storytelling is every bit as important as your linear regression model running in Python.