In December 2011, I conducted a quick analysis of the Alexa Top 100 web site home pages. The results showed notable issues across these sites. Now, over 5 years later, I thought I’d re-evaluate those same sites to see if and how things have changed.
As in 2011, the WAVE browser extension was used to conduct the evaluation. While the extension has changed a bit, the results give an indication of change over the last 5 years.
The following table shows WAVE errors from 2017 compared to those identified in 2011 (both values exclude contrast errors).
Site Data
| Site Name | # of Errors in 2017 | # of Errors in 2011 |
|---|---|---|
| ehow.com | 334 | 1 |
| espn.com | 238 | 18 |
| usatoday.com | 208 | 105 |
| aol.com | 177 | 15 |
| myspace.com | 158 | 10 |
| huffingtonpost.com | 123 | 9 |
| allrecipes.com | 119 | 15 |
| kohls.com | 117 | 51 |
| hulu.com | 115 | 50 |
| washingtonpost.com | 111 | 43 |
| nytimes.com | 96 | 58 |
| tmz.com | 88 | 30 |
| macys.com | 88 | 2 |
| slickdeals.net | 77 | 306 |
| amazon.com | 76 | 6 |
| nfl.com | 73 | 107 |
| about.com | 71 | 26 |
| twitter.com | 71 | 5 |
| reddit.com | 61 | 24 |
| msn.com | 60 | 5 |
| dailymail.co.uk | 59 | 22 |
| pch.com | 57 | 0 |
| ups.com | 55 | 24 |
| yelp.com | 54 | 3 |
| cnet.com | 53 | 20 |
| foxnews.com | 49 | 220 |
| youtube.com | 48 | 8 |
| go.com | 47 | 44 |
| toysrus.com | 47 | 30 |
| bing.com | 47 | 0 |
| match.com | 43 | 26 |
| imdb.com | 42 | 20 |
| vimeo.com | 42 | 3 |
| salesforce.com | 37 | 22 |
| pof.com | 36 | 17 |
| sears.com | 36 | 15 |
| latimes.com | 34 | 67 |
| godaddy.com | 33 | 16 |
| wsj.com | 33 | 16 |
| foxsports.com | 32 | 19 |
| jcpenney.com | 30 | 1 |
| pandora.com | 28 | 44 |
| cnn.com | 27 | 89 |
| imgur.com | 25 | 19 |
| fedex.com | 24 | 23 |
| gap.com | 23 | 22 |
| photobucket.com | 23 | 18 |
| newegg.com | 23 | 6 |
| shopathome.com | 22 | 8 |
| answers.com | 21 | 61 |
| barnesandnoble.com | 21 | 29 |
| warriorforum.com | 21 | 16 |
| verizonwireless.com | 21 | 2 |
| homedepot.com | 17 | 90 |
| capitalone.com | 16 | 55 |
| coupons.com | 16 | 35 |
| att.com | 16 | 30 |
| mywebsearch.com | 16 | 1 |
| adobe.com | 15 | 2 |
| target.com | 14 | 12 |
| facebook.com | 14 | 8 |
| wordpress.com | 14 | 8 |
| yahoo.com | 13 | 9 |
| americanexpress.com | 13 | 4 |
| rr.com | 12 | 39 |
| swagbucks.com | 12 | 33 |
| constantcontact.com | 12 | 28 |
| bestbuy.com | 11 | 6 |
| comcast.net | 11 | 2 |
| reference.com | 10 | 9 |
| netflix.com | 10 | 1 |
| cbssports.com | 9 | 17 |
| tumblr.com | 9 | 10 |
| drudgereport.com | 8 | 21 |
| wellsfargo.com | 7 | 2 |
| etsy.com | 7 | 1 |
| craigslist.org | 6 | 2 |
| weather.com | 5 | 28 |
| linkedin.com | 5 | 18 |
| ebay.com | 5 | 12 |
| google.com | 5 | 6 |
| stumbleupon.com | 5 | 6 |
| paypal.com | 5 | 3 |
| wordpress.org | 5 | 1 |
| blogspot.com | 5 | 0 |
| chase.com | 4 | 51 |
| pinterest.com | 4 | 2 |
| bankofamerica.com | 4 | 1 |
| ask.com | 3 | 11 |
| groupon.com | 3 | 11 |
| usps.com | 3 | 5 |
| indeed.com | 3 | 2 |
| wikipedia.org | 3 | 1 |
| bbc.co.uk | 2 | 21 |
| microsoft.com | 2 | 11 |
| live.com | 2 | 5 |
| flickr.com | 1 | 25 |
| apple.com | 1 | 0 |
| walmart.com | 0 | 29 |
| thepiratebay.org | 0 | 2 |
Analysis
The number of errors found has increased 60% over the last 5 years – from an average of 25 errors in 2011 to 40 errors in 2017. The median number of errors increased from 15 to 21. 67 of the 100 pages have more errors now than in 2011.
A similar analysis of these pages from only one year ago showed the average errors to be 35.5 (compared to 40 today). This would suggest that the pace of accessibility errors is steadily increasing over time.
Because home pages are often dissimilar to content pages, these data should not be used to cast a sweeping judgement on a site. However, home pages are also typically the entry point for users to a site and often the most utilized page of a site.
Contrast Errors
The average number of WCAG 2.0 AA contrast failures detected for home pages of these popular web sites is 70! 91 of the sites had WCAG contrast failures. Contrast data was not collected in 2011, so comparison data is not available. I believe this represents a significant barrier for users with low vision (though perhaps it also makes one wonder about the relevance of the WCAG 2.0 contrast thresholds for modern design).
Conclusion
Users with disabilities face notable difficulties if home pages for the most popular sites on the web average 40 obvious accessibility errors and 70 contrast issues! These are only the errors detectable through automated analysis – the non-detectable errors have, presumably and based on our experience, increased at a similar pace.
At WebAIM, we try to keep an optimistic view of web accessibility. But these data paint a rather dismal picture. It’s difficult to define why sites seem to have gotten so much worse in just 5 years. Web sites (particularly home pages) have generally become more complex and heavier in recent years. At the same time, I believe that awareness of web accessibility has also notably increased. I think much of the decline in accessibility is due to the increased usage of frameworks and libraries that do not fully support accessibility. Regardless of the causes, it’s clear that much still needs to be done to promote web accessibility.
I’ll be sharing additional data about the types of errors that are most prevalent on the web in future blog posts.
I’m curious how you got the number of errors you got. For example:
1. ehow.com = 146 errors and 150 alerts, but you say 334
2. espn.com = 307 errors and 51 alerts, but you say 238
I haven’t checked any more by entering the email address in the WAVE tool, but would you share how you came to those numbers? I was looking as I wanted to know what kind of errors the tool was finding.
I know the number of errors can vary because of the day the test is run (these seem to be sites that update often), but can you give a little more insight into how you came to your numbers?
Thanks and this is a really fascinating test you’ve done!
Holly –
These were evaluated on my computer using the Chrome toolbar. As you note, these types of pages can change significantly in short order. It’s possible that some aspects of my system would cause differing content to be presented, such as additional Facebook widgets based on my Facebook log in status, etc.
Interesting numbers Jared. I’m curious, since 2011 analysis did not include contrast errors, were the 2017 contrast errors removed from the data set prior to determining the 40 error average/60% increase?
Jim –
The contrast errors are not included in the table or values to measure the increase.