Why you should operate digital Accessibility on Data

Digital accessibility has so far been processed in a rather unstructured manner. Problems are found through manual testing, feedback, or more randomly. This article is about why you should develop digital accessibility based on data and how you can implement it.

Many websites are very complex in terms of the number of subpages, participants and components. Content is changed, forms are inserted, social media content is embedded, components are exchanged. Manual monitoring for accessibility can hardly be implemented.

In addition, many decisions today are based on data: be it search engine optimization, measuring the success of the content or the user experience. So far, digital accessibility has been the exception.

This can have real disadvantages: If you don't know where you stand with accessibility, you can't measure success or deterioration. Then it is difficult to justify the means to the superiors.

As mentioned above, it is difficult to check a complex web project for accessibility, especially when many different people maintain the website. Let's assume that a colleague accidentally does not make a form accessible on a subpage. With a large web presence, such problems are hardly noticeable. A testing tool would test the site as soon as it is published and could give hints.

Accessibility measurement tools

Automated inspection tools can find up to 35 percent of accessibility issues on a website. We assume that artificial intelligence will enable the tools to find more and more problems in the future.

There are numerous programs, all of which have a web-based interface. The more well-known applications include Siteimprove, Silktide, Wave, Deque ax Monitor® or WebAIM Wave. Most of the tools we know are based on axe-core, an open library for automated accessibility tests. The test scope is therefore relatively similar, although each provider can of course also expand the functionality.

Limits and dangers of automatic tools

As said above, about 35 percent of problems can be tracked down automatically. This means that about two thirds of the problems have to be found in other ways. A simple example is captions: so far, a tool can only recognize that alternative texts or form labels are present. The tool cannot judge whether these are useful. An automatic checking tool can lull those responsible into a false sense of security. It can only ever be part of a comprehensive accessibility strategy.

Data is especially useful when you work with it. It's good to know that there are problems. It is even better to work through them systematically. This means that an accessibility specialist should look at the problems, prioritize them and use them to create requirements for development, design or editing.

It makes sense to have this task done by an accessibility specialist. The problem is that these tools also show problems that actually aren't.

The automatic check cannot completely replace other processes. Editors still need to know, for example, how information graphics are designed, texts are structured or images are described. A checking tool can only be used in addition to a good accessibility workflow.

Data and Science