Why design matters in Data Visualization

In this post we will tell you how we work at Mercado Libre in order to improve our understanding and communication of millions of data and thus make better business decisions.

At Meli we have + 1TB daily data stored in our data lake! A large number of dashboards, extracts and queries are generated that are consumed by thousands of users in 18 countries within our entire ecosystem: Mercado Libre, Mercado Pago, Mercado Envíos, Mercado Shops, Mercado Libre Publicidad (Publicity) and Mercado Crédito.

Finding meaning and correctly communicating this enormous amount of data is a great challenge and it is here that our Data Visualization area acts as a strategic link.

Our strategy in data visualization is aligned with a narrative that is consistent with our business needs. We focus constantly on striking a balance between content, visuals and narration to meet our primary goal: the clear and compelling communication of information.

For accomplishing this task, it is paramount for BI teams to integrate profiles from the world of communication and design.

Design makes data visualization effective

The visual presentation of the information affects its readability, comprehension and memorization. That is why we have incorporated a set of design guidelines that ensure our compliance with these three pillars.

The following attributes are the starting points that build up our good practices:

• accessible
• adequate
• credible
• complete
• concise
• timely
• comprehensible
• valuable

Let us explore how this is done at Meli.

We define a clear objective

The most important step in creating an efficient visualization is understanding the types of data involved and knowing our storyline and purpose.

We take the user into account

Depending on our users’ needs, we choose how much data to show and what context to provide since not everyone will perceive the same information in the same way.

Each design choice is made with the purpose of improving user experience so that they are not overwhelmed with an excess of information.

We provide context

Data without context are not information. The more context we provide, the easier it is for users to know which content requires some action on their part; it is a key strategy for clarifying information.

We use headlines, colors, iconography and other visual cues to provide information and help users contextualize, focus, attract attention and interpret information quickly.

It is important to ensure all charts have their headlines as these constitute
our assumptions; here we anticipate to our users what they will be seeing
in the graph. This saves time since they will know what to focus on when looking at it.

We highlight the most important information

It is a proven fact that users do not read in order but start with what stands out most, looking for connections and meaning. Thus, we divide complex information into manageable and comparable bits; we do not put everything on the same page or in the same graph.

This process requires certain organizational strategies like hierarchizing and highlighting the most important information in order to facilitate reading as well as using design resources to direct attention to the most important points.

Another great ally is the color gray, used for the less important elements. This makes the featured colors — which must be set aside for the most important data — stand out even more.

Consistency and standards

Our visualizations have a visual and textual coherence allowing us to solve the same type of problem in the same way.

That is why we seek a very clear visual structure that allows the user to access information instantly and accurately.

We choose the correct graph

We must choose the appropriate visualization format since not all graphics serve the same purpose. Each graph is designed to show a type of information effectively, depending on what we want to communicate.

The way of presenting this information should not distort its meaning.

We make minimalist visualizations

It is important to eliminate visual clutter from the display by removing all items that do not provide information. Structural elements such as backgrounds, lines and borders can be dimmed and in some cases removed so that they do not compete with the data that we want to draw attention to. Part of this visual order is based on using blank spaces to facilitate the recognition of information units.

We are in continuous search for a balance between complexity and clarity.

We use tables as little as possible

We try to use as few tables as possible. Instead, we prefer to use charts that show information about relationships across data, patterns, trends, etc.

We use color meaningfully

Color is a great tool to quickly direct the user’s eye to important elements of the chart and to distinguish between data categories. Misuse of color can distract and mislead the user.

In order not to confuse users, these are some of the points we take into account regarding color:

• We use the same colors for the same data and thus increase comparability
• We make sure there is enough contrast between colors; if they are too similar, it can be difficult to tell the difference.
• To facilitate interpretation, we use up to five colors in the graphics. If there are more variables, it is advisable to group the contents into categories to reduce the number of colors.
• We explain the value or variable of each visual brand by means of references. The closer to the element they are, the better since they limit the path of the eyes.
• When choosing the color palette, we take into account its meaning in our users’ culture.
• We use light colors for low values ​​and dark colors for high values.

Thanks for your reading us!

. . .

Give us your applause 👏🏻 if you enjoyed the article. It will help us to continue writing about what we do.

. . .

My name is Vanina Ogueta, I am a Senior Data Intelligence Analyst at Mercado Libre.