Everything You Need to Know About AudioEye’s Auto Remediations
Summary: This is the first part of a three-part series on AudioEye’s solutions. Read part two: Everything You Need to Know About AudioEye’s Manual Testing and Remediation and part three: Everything You Need to Know About AudioEye’s Active Monitoring and Issue Reporting.
The internet is vast. As of January 2022, it was estimated to comprise at least 1.55 billion indexed web pages filled with every conceivable tool, diversion, entertainment, asset, and resource. It is at once a library that catalogs almost every idea and emotion ever expressed by humankind and a workshop or laboratory capable of generating innovation and advancing the human condition.
The internet, which did not exist 40 years ago, is a place created by human beings for human beings. It is a singularly human enterprise, and yet not everyone can use it. Among those billions of web pages are trillions of errors — man-made mistakes or poor design — that make the internet’s myriad benefits and possibilities inaccessible to billions of its intended users worldwide.
Fortunately, 100% of those errors can be fixed. But it takes time and effort.
Every website accessibility problem has a solution. Some fixes are easy and others quite difficult, but there are no website accessibility knots that cannot be untied, whether by making the existing content more accessible or by creating more accessible alternatives.
The biggest challenge in accessibility is not figuring out how to repair all the problems. It’s about acquiring the resources to get the work done. More than that, it’s about getting resources that provide an ongoing and scalable solution, to accommodate for the dynamic nature of websites. And while there are just too many accessibility issues for human beings to fix right away, people need to navigate the internet right now. They don’t have the luxury of waiting.
One of the tools AudioEye uses to help speed up this process is automated remediations. In this post, we’ll cover everything there is to know about them. In addition to walking you through their capabilities and limitations, we’ll show how we use the technology at AudioEye now and how we hope to see it used in the future.
What is Auto Remediation?
"Remediation" is the term website accessibility experts use to describe the process of removing the barriers to accessibility in a website's code and making websites accessible to every user.
When a remediation is done by a human being writing customized code, it's called manual remediation. When it's done by a machine programmed to make those changes using pre-existing code, it's called automated remediation.
AudioEye uses four main types of remediation. In addition to auto remediations, which are purely code, and manual remediations, which are written by human beings, there’s a hybrid guided remediation type that requires human input but does not require that person to be an expert in coding or website accessibility. The fourth type is user-controlled remediations.
Completely automatic code-based tools capable of alleviating website accessibility problems in a fraction of a second.
Guided remediations combine the speed of code with the power of human judgment. Especially useful because they do not require an expert’s skill set.
Human interventions remain the gold standard for remediation, but they are not as fast or cost-effective as automated solutions.
User-controlled remediations let users implement certain fixes and adjustments, like those mitigated by using AudioEye’s Visual Toolkit.
You may be surprised to learn that tools that help identify errors on websites aren't new. In fact, this technology has been around for about as long as there have been websites. The first version of HTML was published in June 1993, and the first HTML validator followed just a few weeks later.
“The amount of manual work that's required to detect all of a site's accessibility problems and to make everything on it completely accessible is tremendously high. It's just a massive, unattainable mountain of manual labor. If we are ever to have a hope of trying to make the entire web accessible, we have to leverage the power of automation.”
— Jon Page, Director of Data Science and Engineering at AudioEye
Since that time, developers and web accessibility specialists have used automated tools in different forms to find and fix all kinds of website accessibility issues. Some of those problems have always been able to be corrected by automated means, while others have become fixable as technology has progressed. Unfortunately, despite numerous advances, it's not yet possible to rectify everything with automated remediations.
How Does Auto Remediation Work?
If you're not a technical person, it might help to think of auto remediation as a sort of accessibility "spell checker" for the code running behind the scenes of a website. Like spell-check, auto remediation is a helpful and reliable tool that's capable of catching and fixing lots of mistakes. Of course, like a spell checker, auto remediation has its limitations. It can't catch everything, and there are some nuances it's simply unable to detect (at least for now).
The Invisible Fix
All the fixes made by the auto remediation process are completely invisible to users who do not require the use of assistive technology to access your site. That's because auto remediations do not adjust a site's source code. Rather, they inspect the code that is delivered to a visitor's browser and apply fixes to it that enable assistive technologies to correctly interpret the content for that user.
Where auto remediation excels is where there are identifiable patterns. A good example of a pattern would be HTML heading elements that are not in the correct order. The proper pattern should outline the ideas being introduced, starting at heading 1 and proceeding in order. When the auto remediation code detects an out-of-order element, it can make the fix automatically. Other common patterns would include factors like fields having labels or links having descriptions of their destinations.
At the most basic level, these auto remediations work on an "if X, then Y” basis. If the code detects a heading level 2 element followed by a heading level 4 element, then it applies a remediation to adjust the levels and put the headings in proper order. The code becomes more advanced in cases where the “then Y” component requires more information. For example, “if a form field is missing a label, then figure out what the label should be.” That “figure out” component requires automation to run through a series of conditionals until the right conditions are met. For instance, if the words “first name” are nearby, automation can determine that “first name” is the probable missing label.
Where auto remediation technology falls short today is when context is required. Like a spell checker that can’t tell whether you meant to write “his” or “hers” in a particular sentence, auto remediation may be unable to fix an issue if the underlying code varies greatly from patterns known to be inaccessible.
AudioEye’s technology delivers more than a billion remediations daily, and is well-versed in identifying these patterns. Luckily, there are many patterns that can be picked up on, and new patterns are constantly emerging. Identifying these patterns allows us to continue to learn and build new technology, pushing us forward on a path to solving more issues and complex challenges for web accessibility.
Auto Remediation at AudioEye
AudioEye embraces a hybrid approach to website accessibility that takes advantage of both automated and manual remediation techniques. We want our auto remediation tools to fix as much as possible, because automated fixes are faster and more affordable than manual options, and allow every website owner to make their site more accessible regardless of their resources or skill.
Although we understand that auto remediations can't solve every website accessibility problem today, we are devoted to improving our capabilities. And the reason for that is simple: Every time we're able to solve an issue reasonably well with automation on one website, we become capable of fixing millions of instances of the same issue across the internet. Currently, AudioEye has more than 70 thoroughly tested auto remediations that are helping to resolve billions of issues automatically each year.
Issue Reporting Dashboard
As we mentioned earlier in the article, accessibility is an ongoing effort that requires continuous monitoring. So we built an Issue Reporting feature that allows even non-technical users to continuously detect and monitor accessibility issues on their websites. With its user-friendly interface and custom views, Issue Reporting dashboard provides detailed information on issues, including the impact these issues have on site visitor experiences, which issues have been fixed automatically, and which still need to be fixed. These insights help to prioritize high impact issues and provide full visibility into AudioEye’s work.
Who Creates AudioEye Auto Remediations?
AudioEye believes in the power of auto remediations and continues to invest resources toward the development of new solutions. Among the contributors to the innovations dreamt up by the auto remediation workshop are highly skilled experts, including:
Web Developers and Software Engineers
AudioEye accessibility experts may be certified in the use of screen readers, alternative input devices, and other assistive technologies, and they may even hold a professional credential from independent professional organizations like the International Association of Accessibility Professionals (IAAP).
AudioEye’s accessibility experts, who are AQA testers, web developers, designers, solutions engineers, accessibility evangelists, account managers and salespeople, assist with the development of remediations in several important ways. They help to:
- Determine which accessibility issues to create auto remediations for.
- Provide best practices for accessibility, including what code works best across different assistive technology and/or browser combinations.
- Highlight potential edge cases to ensure the remediation works across sites with a high degree of accuracy.
- Test the auto remediation using assistive technologies in both a simulated "playground" environment built specifically to challenge the remediation and in staging versions of actual websites.
- Approve the final product before it is released.
- Monitor auto remediations after release to determine how an auto remediation could be improved.
AudioEye A11iance Community
The AudioEye A11iance, a group of accessibility testers who are members of the disability community, provides feedback for product development and QA at AudioEye. A11iance members use assistive technology in their daily lives, and their perspective is invaluable to our processes. Through their QA testing, the A11iance helps us detect issues with current auto remediations and identifies the need for new ones.
Data scientists use computer science, programming, modeling, mathematics, and statistics to drive insight and innovation regarding some of website accessibility’s most vexing problems. They apply their considerable skill to find answers from data sources no matter how vast or minute.
During the process of building new auto remediations, data scientists look at monitoring data to identify the type of problems we should try to solve next. They also monitor the data on existing auto remediations to ensure they are performing optimally.
Data engineers are close colleagues of data scientists. They specialize in building pipelines that transform huge quantities of information into formats data scientists can use. Data engineers build and maintain database systems and data warehousing solutions, and frequently apply their skills with programming and machine learning to solve problems.
Because their data scientist colleagues require access to so much information, data engineers work to develop ways to organize data sets so they can be optimally delivered and optimally stored.
To focus our efforts, we rely on mountains of anonymous data (AudioEye does not require or track individual user information for this work). We constantly collect and monitor data that helps us identify the most prominent problems and most urgent needs. Once we zero in on a challenge, we can apply our know-how and resources toward finding a solution.
Our goal is to improve our capabilities. When a manual fix cannot be avoided, we provide no-code solutions like guided remediations, so people can make the fix easily and without having to understand the technical details of the issue. Of course, when a problem is just too complex for automated remediations as they work today, we have the expertise and experience necessary to create custom accessibility solutions.
“At AudioEye, our long-standing mission is to eradicate every barrier to digital access for individuals with disabilities. We are pushing the envelope when it comes to the reaches of automation to identify and fix issues of accessibility.”
— Sean Bradley, AudioEye Co-founder
We hope you found this article useful. To learn about the state of digital accessibility and AudioEye's innovative approach to helping businesses make their sites accessible, download Building for Digital Accessibility at Scale white paper.
Share this post