Browser Statistics By Rendering Engine
There is quite a bit of talk of browsers statistics as the browser wars heat up again. But what most folks following the statistics fail to realize is that it isn't just a browser war, it is a rendering engine war. Why would you want to think of it like that? Because, as web designers/developers, we want to ensure that our sites work for everyone. But, for many projects, we can't possibly test every single browser version of every single platform. It makes more sense to test rendering engines if time or cost is an issue, for instance. Having the rendering engine stats is also a great way to convince webmasters and business folks that perhaps they should consider writing more standards-compliant code and testing it on multiple browsers... thereby gaining visitors.
The Statistics (as of December 1, 2005)
Without further ado, here are the statistics (compiled at the end of the month listed):
|Rendering Engine||Nov 2005||Oct 2005||Sep 2005||Aug 2005||Jul 2005||Jun 2005||May 2005||Apr 2005||Mar 2005|
|MSHTML-Modern (Internet Explorer 6)||82.77%||82.51%||83.33%||83.26%||82.81%||82.19%||81.04%||80.49%||79.99%|
|MSHTML-Legacy (Internet Explorer 4/5/5.5)||3.76%||3.94%||4.34%||4.49%||4.88%||6.51%||7.77%||8.76%||9.23%|
|Gecko (Moz Suite, Firefox, Netscape 7, etc)||9.92%||9.97%||9.02%||9.00%||9.05%||8.31%||8.22%||7.84%||7.78%|
|Opera (Opera - all versions)||0.83%||0.83%||0.72%||0.72%||0.73%||0.68%||0.67%||0.65%||0.66%|
|Netscape (Netscape 3/4)||0.07%||0.07%||0.08%||0.07%||0.08%||0.10%||0.09%||0.09%||0.09%|
|Other/Unknown (Unknowns, spiders, etc)||1.02%||1.06%||1.22%||1.23%||1.26%||1.20%||1.21%||1.21%||1.31%|
Legend: Increase Slight Increase (≤0.05%) Unchanged Slight Decrease (≤0.05%) Decrease
Commentary - November
Still no update from Websidestory, so the numbers are getting increasingly innacurate (as Websidestory's numbers are from April 2005). If this is unchanged next month, Websidestory will be replaced with a more up-to-date source.
Commentary - October (from July)
OneStat has released their new numbers, resulting in more accurate, up-to-date compilations. The big winner over the last couple months has been Gecko (lead by Firefox) as it captures close to another point. KHTML (lead by Safari) has also picked up a bit under 1/2 a point. Opera picked up a few users as well. The big loser is still MSHTML-Legacy as people abandon the old versions of Internet Explorer. A bit of a surprise, though, was that MSHTML-Modern has lost users as well, primarily to Firefox. This is the first decrease in modern IE's numbers since I began tracking statistics by rendering engine. It should be noted that WebSideStory still hasn't released their new numbers. Most likley, we'll see Firefox comfortably over 10% when they do.
Commentary - August/September
Not much to report either month as no new stats from OneState or Websidestory were released. So, our overall stats are increasingly out of date.
Commentary - End of July
Unfortunately, neither OneStat nor Websidestory released new numbers this month, so not much meaningful movement will be seen from a business or large site prospective. TheCounter can still give us some useful information, though. We can see that the defection away from legacy Internet Explorer is continuing, with users mostly moving to both IE6 and Gecko-based browsers. We can also see slight increases in Safari and Opera again -- whether they are statistically significant is another matter. Hopefully OneStat and WebsideStory will release their new numbers soon.
Commentary - End of June
The biggest change again was the decreasing numbers of legacy IE browsers. The difference this month is that many seem to be moving to IE 6 instead of the even split between IE6 and Gecko we saw last month. Opera, KHTML and Netscape are all steady. Not much in the way of new stats numbers this month, just a new month of numbers from TheCounter, so look for a more meaningful update in the next couple weeks when WebSideStory and OneStat release new numbers.
Commentary - End of May
Between the end of April and the end of May of 2005 we can see that the biggest shift in the stats is users moving away from the Legacy MSHTML engine. As users abandon IE5 and IE5.5, about 1/2 seem to be upgrading to IE6 while the other 1/2 seem to be abandoning IE for a Gecko-based browser. (Of course, more of the legacy users could be upgrading than switching and some of the modern users could be switching to Gecko as well.) All other stats have remained relatively stable.
Rendering Engines Explained
For our purposes, we are doing a rather rough combination of rendering agents. Realistically, the major rendering engines are as follows:
- MSHTML-MODERN: Microsoft's current engine used in Internet Explorer 6.0
- MSHTML-LEGACY: The engine from pre-IE 6. (a rough combination based on necessity of the stats)
- GECKO: The engine used by Mozilla Suite, Firefox, Kmeleon, Galeon, Camino, Netscape 6/7, etc
- KHTML: The engine created within Konqueror and used in Safari
- OPERA: The Opera browser's engine
- OTHER: All the rest that don't fit into the above along with misidentifications, no useragent string, some bots, etc
Now, as I said, this is a bit of a gross oversimplification of things. Internet Explorer 4's rendering engine is quite different from IE5. And IE5 is quite different from IE5.5. Netscape 6 is very different from Netscape 7 (though this amounts more to bugs on an incomplete engine rather than vast engine differences.) There are also platform differences to take into account. IE-mac and IE-windows are really more the same in name than anything else. But there's only so much we can do in terms of splitting things up to keep things readable and operate within the statistics we have available.
Data Sources and Combining
To get a feel for things, I used 3 different data sources:
- WebsideStory (May 10, 2005): websidestory.com/products/web-analytics/datainsights/spotlight/05-10-2005.html
- OneStat.com (November 2, 2005): onestat.com/html/aboutus_pressbox40_browser_market_firefox_growing.html
- TheCounter (December 1, 2005): thecounter.com/stats/2005/November/browser.php
All three are imperfect, but together, they may cancel out some limitations. WebSideStory is very US business-centric. OneStat is more global. TheCounter is more geared to smaller sites.
Now, not all stats sources provide the detail needed for the above summary, so I interpret out using relative percentages from the other sources. For example, WebSideStory gives only a combined Internet Explorer percentage. So, I used the relative percentages of IE versions from TheCounter to extrapolate out a possible version breakdown. Not incredibly accurate, but still of some use to us.
As with *ALL* statistics, they should be taken with a grain of salt. There are lies, damn lies and statistics. WebSide Story and TheCounter seem to use a combination of IP address and cookies to count unique visitors (about as accurate as you can get from a server), but we still can't consider these numbers... even combined... as the definitive word on how many people are using each browser. It's simply not possible. We can, however, examine the trends in usage of these stats over time and gain some insight from that. Of course, the best thing for us web folks to do, is to build sites to standards and then add in the quirks for all the major rendering engines. This ensures a site that works with all the browsers out today and that will work with browsers in the future.
I'll be updating these results as the numbers change on the respective sources. If another good data source is being overlooked, please let me know.