Web Accessibility and Web Analytics
Web Analytics Offer Services to measure and analyze a big amount of data which comes from the interaction of a visitor with your website. But can we use Web Analytics Data for issues of web accessibility?
There are two possibilities to use the Tracking for Accessibility Improvements:
- the tracking of a published website
- the tracking of a usable prototype, for example in a closed beta.
- the usual Tracking
- Statistic significance
- Beta Testing
- A/B and multivariate testing
- measurable success
- Read more
the usual Tracking
All big Analytics providers offer the tracking of special actions or the Analyse of click paths for example for a buying process. This tracking contains mouse tracking, formular tracking and click tracking.
With a heat map you can see, where persons did click, if there was nothing to click; you had found a problem to solve. Or you can watch the time which persons need for a special target, if it takes to much time you have to analyze the reasons.
Another important indicator is the bounce rate, the count of persons, which do not finish the buying process. When a measurable count of persons bounces on a certain point you have to find out, why.
You may argue that the bounce rate is more an indicator of Usability than Accessibility. And you may be right. But I think the borders blur at this point. However, if you have a significant bounce rate short before the user should press the "Buy-Button", you should hire an expert who can analyze the problem for example with a heuristic walkthrough.
Another point to take a further look on are flyout navigations. Most persons with disabilities and not only have they had problems with that, because this kind of navigation demands a lot of Feinmotorik from the user. With mouse tracking you can watch the dropout rate.
At this point I have to deal with the main problem in this area: Statistic significance. Web Analytics does not allow recognizing if a special user or a couple of users have a disability. The statistical aggregation let disabled users disappear in the mass of data. Today there is no possibility for the service provider to find out, if his visitor uses assistive technology or not.
Let’s assume that 10 % of our users have a disability. That does not automatically mean, that they have a problem or that they have all the same problem with your site. A visual impaired user may have no problem, whereas a blind user cannot use your website.
Nevertheless the lack of accessibility can be one issue for example for Form breakup or problems with an interactive player. A high bounce rate is an issue which affect abled and disabled users.
Another kind of accessibility is the accessibility for users with low computer skills or old equipment. This is for example important when you offer things which are often bought by older or poor persons. Indicators for such issues you can find in the statistics for operating system, user agents, low Internet speed and so on. persons who still use Windows XP are either technisch uninformed or they cannot change. Many of these users may have a disability because their software or assistive technology runs only on windows XP On their special configuration, or they do not get help to update their systems.
An alternative possibility to the usage of existing data is to test the web site or prototype in a closed beta area. The advantage compared to a test in a usability lab is that the testers can use their own technology at home. In most cases this will improve the quality of the results because the testers are more relaxed in their usual environment.
Another advantage is that you can invite more testers for example with a special disability and can combine different approaches like Web Analytics and loud thinking.
As a result you have only the data of disabled persons and you can make a lot of substantiated assumptions.
A/B and multivariate testing
A very helpful but ebenso complex possibility are A/B tests or multivariate tests. A/B testing means, that you test two versions of a page, multivariate tests means that you test three or more versions.
This is helpful if you are not sure of the cause of the problem. Do persons not click on "Buy" because they do not find the button nor don't they like their pay options? Is the contrast to low or is the text to complex? The only way to answer these questions through web analytics are A/B-Testing’s.
In this case you need many visitors to get valid data. Especially for multivariat testing you need enough persons who have seen the different versions to say which causes the lowest problem.
Another problem should be mentioned: With these procedures you only find the issues of the tested versions. It still can be that you oversee important issues because you look on the wrong part of the page. The users may have problems to log in and these users do not care about the placement of the "Buy"-Button because they never reached this point.
The best thing of Web Analytics should be mentioned at last: otherwise than in other actions you can measure the success straight after the change.
But you should be careful. Websites with low visitor counts may take a few weeks until the changes are present in the data. For websites with visitor rates a few procent can increase can be useful.
Web Analytics is not useful for all issues of web accessibility, but can be in some parts useful. You should consider using it for tasks of quality management and insurance. This is especially helpful in website areas with high interaction rates. Other areas must be controlled with other tools.