Making Sense of Data: the Human Side of Data Analysis Danyel Fisher Microsoft Research How do you get from a sheet of raw numbers to a good visualization? How do you decide which visualizations, interactions, and linked views make the most sense for your data? Most visualization classes assume that you know what you want to see; data science scenarios all-too-often assume that users already know what they expect . In this talk, I will outline the process of “data counseling,” which is a structured way of working from your business needs down to concrete visualizations. I will discuss requirements gathering; how to operationalize those requirements into questions that your data can answer; and how to construct a visualization that addresses these questions.