I was telling a friend that I had just finished watching the 3h 30min version of The Irishman. He replied:

Can you summarize it for me? I want to make an informed decision before making any commitments.
_{(That’s how my friends talk)}

At first I thought it was a joke, but then I decided to try it for real, because this is kinda of my job.

A common misconception about data analysts is assuming our job relies solely on the technical skills to convert data into visualizations. It doesn’t. If it were just a matter of creating a dashboard, we would be software developers. If it were just a matter of designing good-looking graphs, we would be designers. The real challenge lies in understanding how useless or misleading numbers can be. That requires cumulative knowledge of statistics, data interpretation, database architecture, and, most importantly, understanding how two people can come to different conclusions based on a single graph.

There are infinite ways to summarize a 3-hour movie. Do you want to know the main events? Do you want to follow the perspective of one specific character? Do you want to feel the tension of the narrative? In any case, A LOT will be left out. When summarizing the movie, I had to keep two things in mind:

1. The person I was talking to.
2. What they intended to do with that information (decide whether to watch the movie or not).

So, not only did I need to focus on elements that might convince him, but I also had to leave enough out to avoid spoiling the experience if he decided to watch it. Overall, I think I did a good job for that specific task. However, if he were taking an exam focused on the plot, my summary would have been a major failure.

A dashboard is exactly like this summary. All data contains so much story and context that no single small set of graphs and tables can properly tell it all. Choosing which elements your summary will contain is extremely hard since everything feels important. When working with journalists, I would always provide a report covering all important aspects, caveats, and nuances, because in journalism, precision might be the difference between getting sued or not.

It is often said, “There are three kinds of lies: lies, damned lies, and statistics.” Ironically, that’s a lie. Statistics can’t lie; they serve as a frame. A data analyst’s job is to make sure they are framing what is relevant for the specific person using the data. That’s why it shouldn’t be taboo to ask, “Ok, but why do you want me to include this number?” when asked by a stakeholder. The wrong frame might render the numbers as useless for decision-making as rolling dice.

Basing a decision on numbers does not automatically make it more precise than “following your heart.” Making sure the numbers are framed in the correct way to answer your specific question will. And that’s a Data Analyst’s job.