Here are some thoughts of mine about The WebAIM Million’s report (opens in new window) for 2021 that runs an automatic accessibility test on “top million sites” (top based on Open PageRank Initiative and Alexa Top sites). It is important to notice that this is based on automatic tool that is only able to find around a third of all WCAG errors. So in reality we would have to do a manual audit on all of the sites to find realistic status of accessibility. Automatic audits are still useful and can also catch serious barriers, so it can still be worth referring to them.
Better accessibility than previous year, but still far from conformance
I would say that 15.6 percent decrease of accessibility errors is quite a good sign. We are still far from accessible internet when we consider that sites still have about 51 accessibility errors per page, but compared to almost 61 in 2020 it is quite a progress.
This is a positive trend but when we examine conformance to WCAG only 2.6 percent of all home pages conform to it. It is an improvement of 0.7 percent and it means that only about 26000 out of a million pages conform to WCAG 2 as far as automatic tool can tell.
And once again – we are not counting the errors that can only be discovered with manual audits and those can weight much more from user perspective.
Time will show but it looks like there are some positive trends going on already. Including accessibility early in the process is the key and everybody that makes the web should do their best to add to accessibility. Passing automatic test tools can help a bit and should be a side effect not a goal. Testing with users should be the preferred way.
Design should learn more about accessibility and implement it early on
Well, this is just my personal opinion based on the types and trends of failures, but it seems that designers should really learn more about accessibility, so that developers can do their part as well.
Most common types of WCAG failures that are worse than last year and are could very likely be fixed at the design level;
- Low contrast – 86.4 percent of home pages have issues with low contrast, previous year it was 86.3, this means about 4000 pages more,
- Missing form input labels – 54.4 percent compared to 53.8 percent previous year. So forms are now worse for about 6000 pages more than previous year.
Developers can not fix poor contrast without designers. Otherwise they can introduce inconsistencies in the overall design, sometimes even making page look ugly. It should not be their responsibility. If design took care of this in the start it would not be a problem, at least in most cases.
Sometimes the responsibility also lies on the content side, but an experienced designer should also define border cases and content creators should also be thought of best practices regarding accessibility.
Missing form input labels is most likely also a design issue, especially in forms that are made of inputs with placeholder-only labeling. Designers should make labels in their design and developers should code them appropriately. If developer gets a design that is missing form labels then they would have to define them together with the designers.
So once again – it is probably obvious by now – if team would have to fix the issues they would lost some time discussing the fixes and implementing and re-testing them. Much less efficient comparing to a design that has such simple but important issues defined at the start.
Developers should use ARIA with caution and learn more about it
I can imagine the motivation behind using ARIA from a developer’s perspective. It seems perfect for the task. Let’s just add ARIA and make it work, right?
Well – usage of ARIA code increased by 25% compared to previous year. 68.1 percent of the million home pages used ARIA. But ARIA is by no means sole indicator for better accessibility. Especially when considering that home pages with ARIA had on average 41% more detectable accessibility errors compared to those that did not use ARIA at all.
First rule of ARIA is to not use ARIA at all if we can use “native” semantic HTML elements. And it is the first rule for a reason. Bad implementation of ARIA can do a lot of harm to accessibility. ARIA needs to be understood before just used.
Solution is empathy and knowledge
In my personal opinion we all need more empathy and knowledge. Then we have to apply the knowledge early on and test with real users. Some testing can be done by us but to really make our products accessible we need to involve users of assistive technologies.
Applying accessibility early, from the very beginnings is the only viable and efficient solution.