The Curious Challenge of Accessible Data Visualisation

Airbnb is a vast product. Its various facets provide writers, designers, and engineers with a veritable playground of projects. One of the most interesting design challenges across all roles is the task of making data visualization accessible.

What is data viz, really?

“Storytelling itself is an activity, not an object. Stories are the closest we can come to shared experience…”

Harriet McBryde Johnson

Data visualization is commonly defined as the representation of information in the form of a chart, diagram or picture. In essence, it’s a way to communicate the story of a dataset to users in a form that’s easier to absorb than a table of numbers and information.

For most people, the easiest way to absorb information is visually. The visual cortex — the area of the brain tasked with interpreting sensory input from the eyes — is an incredibly powerful problem-solving tool. It translates rays of light that bounce from objects near us into objects with depth and definition; neutral input is transformed into a tangible environment. Were sight not available, that area of the brain would continue working to make sense of the world, instead using its energy to interpret the input from the other senses, such as hearing and touch.

With this in mind, the term data visualization is really a misnomer. Visualization is problem-solving in the form of storytelling — most of us are just so attached to our eyesight that we can’t fathom how that story could be told in another way. This implicit bias has led designers to traditionally focus mostly on the visual experience of users.

As technology for creating stunning UI has advanced year after year, designers have been able to build more complex and dynamic visualizations to capture users’ attention. Unfortunately, design around other modes of presenting data has fallen short. While sighted users are provided an elegant presentation of their hosting insights via eye-catching imagery, blind and low vision users are left to navigate through broken components to decipher indistinguishable elements or long lists of numbers. Their assistive technology treats the graph component as an unwieldy object, rather than an interactive and harmonious tool. In many ways, these users are left to fend for themselves, never knowing if they’re drawing inaccurate conclusions from the data presented to them.

What does this mean for Airbnb?

“It is very rare, as a disabled person, that I have an intense sense of belonging, of being not just tolerated or included in a space but actively owning it.”

S.E. Smith

From a business standpoint, Hosts who don’t have access to sight are at a significant disadvantage when it comes to understanding their strengths and areas of improvement. They aren’t being empowered to grow their listings in the same way sighted users are. For blind and low-vision folks, the employment rate is about half that of non-disabled people, so empowering these users to make money through alternative means is incredibly important. So what are we doing at Airbnb to close that gap?

Step one is making what we already have accessible with keyboard controls. Airbnb is currently adding the capacity for users to interact with graphs using keyboard controls, so their screen readers can announce individual data points to them. But without proper forethought into that keyboard interaction, we can unintentionally create a situation in which a user is forced to tab through hundreds of data points before exiting a graph, leading to frustration and keyboard fatigue. After all that effort, they get basically the same experience as reading out a table of information. The data is there, but is a story being told? Certainly not.

Exciting possibilities in the works

With the understanding that non-sighted users have all the power of the visual cortex at their disposal just waiting to interpret data, designers should challenge themselves to offer rich and complex auditory and touch interactions in addition to what’s offered visually.

There are a few creative solutions already out there, such as data sonification and 3D printed models, but they come with their own disadvantages. Data sonification allows users to hear data points as different pitches that change over time — but for users who were born blind and cannot translate that sound experience to the visual concept of a line graph, the experience still isn’t accessible. Another option, 3D printed models of graphs, while an effective alternative to visual graphs, are not a scalable solution for a web product.

Off to a promising start

The truth is, there isn’t a perfect solution to this design challenge right now. But that’s what’s so exciting, there’s so much space to grow!

In the short term, we can create innovative keyboard shortcuts that allow users to skip through data more quickly, or type in a search for a particular point of interest. We should challenge ourselves to come up with more descriptive graph labels, so that if the visual itself isn’t accessible, the story it’s trying to tell isn’t lost.

As we embark on the journey of making data visualization fully accessible, we can build on our understanding of data as a story, and trust in the powerful visual cortex to interpret stories told through touch or sound rather than sight.

The first step is bringing awareness to the need for a better experience. As more designers and writers turn their efforts towards this gap in accessibility, solutions — at first imperfect, and then more refined — will be revealed.