The world’s leading publication for data science, AI, and ML professionals.

How To Avoid Data Vomit

A checklist to consider before telling any story with data

At work, I illustrated a story using Data that a supplier’s rates were expensive. I explained the story through a summary supported by a detailed view of the largest price variances.

I spent 80% of my time analyzing the appropriate visuals and constructing the storyline for my stakeholders. I presented it and forgot components for a clear and compelling story – an axis title, a clear headline, and appropriate use of colors.

I felt stupid, forgetting these critical steps. And I remembered in the _Checklist Manifesto_ that __ humans are not naturally detail-oriented, especially as problems become more complex. It was evident in healthcare and aviation that pilots and surgeons would miss "killer" steps despite heavy training. So checklists became necessary to avoid mistakes and produce positive outcomes.

With Storytelling with Data, I’ve created a checklist that helps me stay consistent and defend myself from trivial mistakes before presenting any data to stakeholders. I’ve designed this checklist to apply to any medium: a single slide, a whole slide deck, a memo, or a Jupyter notebook. But, this checklist is not a how-to guide to tell a story effectively. Below is the checklist, and I’ll explain the rationale.

☐ A thesis with a call to action or critical insight

☐ Remove chart borders and gridlines

☐ Graph and axes titles are gray with only the first word capitalized

☐ Avoid legends and label data directly with consistent colors

☐ All text and labels are upper-left justified

☐ Use color, size, shading, and bold text to highlight points of interest

☐ Add text callouts in your visual to reinforce the story

☐ All slide/section titles together tell the story


Story

☐ A thesis with a call to action or critical insight

☐ All slide/section titles together tell the story

Data is just numbers. Humans have to tell the story, so tell your audience what you want them to know or do. Aim to distill it into a clear thesis statement for clarity. You can apply it to a single visual or the entire presentation.

Each slide/section contributes to your overarching story. When you collate your titles together, is your story progressing towards what you want the audience to know or do? Your titles should represent a concise way of explaining your message.

Clutter

☐ Remove chart borders and gridlines

☐ Graph and axes titles are gray with only the first word capitalized

☐ Avoid legends and label data directly with consistent colors

☐ All text and labels are upper-left justified

Like a clean room helps with a clear mind, an organized presentation creates simplicity for our audience. Your audience will remember the single takeaway, so remove the complexity starting with the design.

You don’t need borders and gridlines in your chart. It doesn’t add value. Similarly, labeling your data directly, instead of a legend, simplifies where your reader needs to look. Design principles also apply to text.

We read from left to right starting from the top, so let’s maintain that alignment unless you have a particular case. And words in a title do not have to all be capitalized. Write Normally Instead Of Like This. Many capitalized words compete for the audience’s attention – we want our audience to focus on the message, and the color gray helps reduce that competition.

Let’s start with this graph:

When I use the checklist, I produce:

It looks more appealing, but the orange makes Portfolio 1 the focal point. I have to improve on that.

Focus

☐ Use color, size, shading, and bold text to highlight points of interest

☐ Add text callouts in your visual to reinforce the story

It is useful when your data is accessible. We achieve that by guiding our audience using strategic focus. We inject ‘life’ into our visual by providing meaning through contrast and text. The graph above doesn’t display any critical insight, except portfolio 2 being higher. Let’s improve that shown below.

Based on my hypothetical analysis, I know that Amazon (AMZN) was added to portfolio 2 for 2018, so I decided to use a line and bold text to flag this. I added a text callout to explain to the reader the change that happened in 2018. If you noticed, I changed portfolio 2’s color to blue and made portfolio 1 gray as portfolio 2 is in focus. These are simple but essential items that solidify my story.


I’m forgetful at times. I can be deep into my analysis, incorporating exciting statistical techniques but forget basic rules, so I hope this checklist helps you provide food for thought instead of vomit.


Related Articles