I'm writing a CSS book

JavaScript archives Compatibility Thidrekssaga QuirksBlog Donations CSS book Politics Mobile About COH

contact

IE survey results

I wanted to know which IE versions my readers test on, so I held a survey. The results are below.

Last week I restarted my browser testing, after far too many months of working on the book. I wrote a bunch of new tests, but also wen over my browser test arrays once again. I’m getting tired of old IEs, so I considered removing data for IE7 and below from the desktop tables.

Before doing that, I wanted to make sure this aligns with the interests of my audience. So I created a survey and asked two questions:

1. What is the lowest IE version you test most of your projects in?
2. What is the lowest IE version you tested in at all in the last year?

IE8 is clearly the new baseline. 45% of respondents indicated that IE8 is the lowest version they test their projects in. 30% is on IE9; 16% on IE10 or 11; and 9% on IE6 or 7. So IE8 should definitely stay in my tables. I never doubted that, but it’s good to have the data to back me up.

Decision

Meanwhile I’ve taken my decision; partly based on the survey results, and partly based on the fact IE7 doesn’t run the latest version of my test harness.

I will cease testing in IE7 and below. However, I will keep historical data for these browsers around in a single "IE7 and lower" column.

This only really goes wrong with new tests: if IE8 supports a method or property I never tested before I have to guess if IE7 and lower also support it. In general I assume they support the Microsoft-invented properties, but for others I will occasionally have to add a "Don’t know" entry. If IE8 does not support something I never tested before, I assume IE7 and lower also don’t support it.

Data

Here are the raw data. First Google Drive’s nice graphic:

Then the data table:

How to read: Of all respondents, 705 (45%) test most of their projects in IE8. Of those 705 respondents, 218 (31%) tested in IE7 at least once in the past year.

The IE11-IE5.5 columns don’t add up to 100% because a few respondents were confused and reported, for instance, testing all their sites in IE9 but IE10 being the lowest IE they tested in at all.

Discussion

There are a few bits of data that merit calling out:

1. 54% of web developers with an IE8 baseline never test below IE8. However, 53% of the IE9-baseline developers occasionally test in IE8. This is an interesting difference. One explanation is that many IE9-baseline developers have ditched IE8 in the past year.
2. Of web developers with a baseline of IE9 or higher, about 70% still occasionally checks in a lower version. See the previous item for a possible explanation.
3. It seems likely that most web developers with a baseline of IE7 or 6 work on corporate Intranets. In fact, a fat slice of IE8- and maybe even IE9-baseline developers will suffer from the same condition.

Bonus stats

Here are today’s actual IE usage stats, courtesy of StatCounter; IE as a whole stands at 21% of the desktop market (though other sources disagree on that):

1. IE11: 10%
2. IE8: 5%
3. IE9: 3%
4. IE10: 3%
5. IE6: 0.24%
6. IE7: 0.15%
7. IE5.5 or lower do not appear in the stats at all