July 6, 2009
a browser prediction: the modern era of the web
There's been a lot of talk the last few days about Internet Explorer's drop in usage share. I'm still waiting on the Net Applications June numbers which I think more highly of than the StatCounter numbers, but one thing is clear, IE 6's downward trend isn't changing.
When Microsoft shipped IE 7 in October of 2006, it moved a lot of IE 6 users.
As you can see from the chart, the Windows Update system did a pretty good job of moving users in the first few months. Then things slowed for a bit. Seeing the slowdown in migration, Microsoft decided in October of 2007 to lift the WGA requirement and you can see the line get much steeper again
But by October of 2008, after a year of WGA updates and then a year of non-WGA updates, Internet Explorer 7 was no longer pulling in large numbers of IE 6 users. At this point, I think it's a safe bet that anyone who was going to update from IE 6 to IE 7 had done so.
And not only had the remaining IE 6 users stopped moving up to IE 7, they didn't move to IE 8 either. By the time Internet Explorer 8 was released, IE 6's trajectory had stabilized and even a major release from Microsoft couldn't budge it.
So what is driving IE 6 share down at this pretty consistent rate of 9/10ths of a point per month? My theory is that this is a combination of two factors. The first, and larger factor, I think, is the PC upgrade cycle. Every month, some number of PCs get replaced with newer PCs and so IE 6 installs are disappearing from the Web at least at that rate. I think that the second factor is overall growth of PCs on the Web.
Last year, about 300 million new PCs were sold. About 200 million of those were replacement PCs, upgrades of older computers. About 100 million of those were "new to the Web" computers. That is, they were not upgrades that replaced an existing Web connected computer and those represent the actual real growth of users on the Web.
As PCs are upgraded, we see some direct movement from IE 6 users to IE 7 or IE 8. As the total Web population grows, the IE 6 installed base represents a smaller portion of the pie. Taken together, I believe these factors are the only influence on the drop of IE 6.
That's some bad news and some good news. The bad is that there's nothing any of the browser vendors can really do to speed the withdrawal of IE 6 from the Web. Releasing new versions, the big lever we have to move the Web forward, just doesn't reach those IE 6 users. The good is that the PC upgrade cycle and the growth of the market are both relatively predictable and steady. That means that we should be able to predict pretty accurately what IE 6 usage is going to look like in a month or in a year.
Here's my IE 6 prediction. It looks to me like we could see Internet Explorer 6 finally fall under the 10% mark by the end of this year. If you extend that line, IE 6 would be essentially gone from the Web by the end of next year.
But I can do a little better than that :-) Because IE 6's trajectory is so stable, it's not really going out on a limb to make predictions 6 months out. I like going out on limbs :-)
The chart here looks at more fine-grained tracking information and so the trajectories of the newer and more active versions of Internet Explorer are a bit easier to discern. Still, six months is a long time in the browser world and seeing the trends sooner, as is possible with the more fine-grained data, doesn't mean seeing the trend better.
Having said that, wouldn't it be neat if at the end of this year, IE 6 and IE 7 were both under 10% share? With IE 6 and 7 quickly disappearing, will we look back at 2010 and say "there began the Modern Era of the Web"?
Data through May 2009 from Net Applications Browser Market Share Report. All newer plots are my own predictions
reactions, thoughts, comments, etc.
Posted by: Aaron | July 6, 2009 7:34 PM
The sum of your predictive lines doesn't look right Asa.
The gradient of IE6 is steeper than the gradient of the IE total. But IE8 isn't going up more sharply than IE7, so the gradient of the IE total should therefore be steeper than IE6... but it's not.
Posted by: Damian | July 6, 2009 10:38 PM
It seems to me that your IE6 slope is to steep. Not even from oct-08 til feb-09 was it that steep and it seems to have been leveling off since feb-09. Where are the IE6 users by country or by reverse domain lookup / ip-range (probably harder to find out)?
Posted by: AndersH | July 6, 2009 10:39 PM
hmm... from your analysis, it seems that for the past year or so, the non-IE browsers (Firefox, Safari, whatever) are mostly only stealing shares from IE6, not IE7/IE8. From what you have plotted, it looks like it's those IE6 users who are suddenly switching to non-IE browsers like Firefox, but IE7 users are mostly only migrating to IE8, not to other non-IE browsers.
That looks very strange, since it should be those IE7/IE8 users who are more likely to accept new and better things thus switching to Firefox. If an IE6 user is reluctant to even upgrade to IE7, I'd think it's even less likely for him to suddenly switch to Firefox.
Posted by: soshimi | July 7, 2009 4:50 AM
The modern era of web browsing is when I stopped using lynx from the terminal. Mosaic was great!
KW
Posted by: Kevin Wright | July 7, 2009 7:13 AM
Just to note that there is something that can be done to make IE6 market share fall quicker. That is to design kiiler web applications not functional in IE6.
Posted by: Ivan Ičin | July 7, 2009 7:16 AM
soshimi, it's actually not as simple as that. If you assume that the loss in IE share is a direct result of people upgrading PCs or purchasing PCs for the first time, 100% of those new machines come with Internet explorer 7 and 8 or Safari. So, for any of those people to become Firefox users, they must move from the newer IE versions of Safari. Also, it's quite possible that they're not moving to Firefox, that they're sticking with newer versions of IE and Safari and it's people who aren't in the upgrade cycle moving to Firefox.
We don't know what the details of the churn are. We just know that IE overall is falling and IE 6 is falling at a pace independent of other browser releases. The rest is speculation. I think it would be extremely unlikely if all of the IE 6 users were migrating to Firefox either directly or through new machine purchases and immediate migration afterward. I think it's a lot more likely that many IE 7 and 8 users are moving to Firefox and are being back-filled by IE 6 users upgrading to newer equipment with newer IE versions.
Ivan, that might have a small impact, but it's really Intranet applications and not Web applications that are holding things up for some (many?) of those IE 6 installs. I don't think there are enough killer Web apps out there (unless you include "search") to actually move enterprise users dependent on IE 6 off of it.
For those IE 6 at home, as opposed to at the enterprise, I suspect many of them are stuck on ancient hardware and don't have the desire or the means to upgrade their hardware in order to get a newer browser in order to get fancy new Web applications.
That doesn't mean that a large-scale campaign to get Web apps to abandon IE 6 (and 7) and start taking advantage of the full feature set of at least CSS2.1, and whatever else is supported by that modern browser set, wouldn't be a fine idea.
AndersH and Damien, well, how about this. You guys make some predictions, draw some charts, and we'll come back here in 6 months and compare. Who ever gets closer gets bragging rights :-) (of course the lines aren't perfect. they're just my predictions. and yes, I should have had the IE total line sitting about a point lower at the end of the chart -- the angle an alignment were slightly messed up in photoshop but I'm not gonna bother to fix it.)
- A
Posted by: Asa Dotzler | July 7, 2009 8:12 AM
I'm sure the predictions will be farely close, just the sum of your 3 browsers does't look right and as it stands could only be acounted for by a small increasing usage of a 4th IE browser.
I'm not having a go at how close the predictions are, just that the sums seem to be a little off.
Posted by: Damian | July 7, 2009 8:17 AM
Soshimi, you are incorrect.
You assume that IE7 is better than IE6.
For the web-developer, this certainly the case. It supports standards better.
However, for the user IE7 is really a terrible browser. I am a rather conservative person. I used FF2.0 for developing a website, but I still used IE6 for browsing. After a while, I stopped developing.
So, there was no need to run IE6 anymore. I upgraded to IE7 (a year after launch or something like that). I also upgraded FF to 3.0.
IE7 is really a hasty product from MS. It contains many bugs and irritating things. It has tabs, but that is the only positive things for the user. It is really a disaster browser from user perspective (I am not talking about web-developer perspective). It crashes, it has placed the button on random places, visited links disappear after relaunch and many more very visible bugs. For me, that was the reason to go to FF3.0.
I had the same experience on my company. They still use IE6 (due to some apps) but they offer also IE7 for people that really want. You can also install FF locally (flash won't work), but that is not really supported. So, conservatively as I am, I switched to the browser the company offered, IE7. But after a few weeks, I was totally annoyed by IE7 and installed FF3.5.
The good things about IE6:
- It is very stable. Probably the most stable browser of all.
- Rather simple, without too many features.
It is rather difficult to compete with those two properties for the average user.
However, if some websites stop supporting IE6, then abandoning IE6 may accelerate. In such case, the property 'very stable' is not true anymore, from a user perspective.
Also companies may start dumping IE6. They are not like old grandma. Conservative, but they don't want a minority browser.
Lucas
Posted by: Lucas | July 7, 2009 8:22 AM
The IE8 line has a 'kink' at Aug 09 in the bottom graph. Why are you extrapolating the other lines in a linear manner, but assuming that IE8 will slow down in Aug 09? The computers that are joining the web for the first time are typically going to be using IE8.
Users using IE7 will typically move to IE8, because they have previously moved to IE7. The IE7 line has flatlined, so I believe IE8 will flatline at the same point.
You can't draw the same assumptions as the move from IE6->IE7 because there was such a large time span between the versions.
I think IE total will reduce, but it will flatline at some point, maybe 60%.
seems like a troll to me, especially with the bent IE8 line.
Posted by: shane | July 7, 2009 10:03 PM
Sorry I have to agree with some of the others. I don't think Asa fought off his impartiality hard enough for this one :)
I agree that IE 7 will fall like a brick and IE 8 will climb proportionally. But after Windows 7 is released, IE 8 will spike, and it may even take other browser's market share with it.
IE 6's decline will taper off. There will be some holdouts (those who can't upgrade because of pirated copies of XP and/or older PCs) and the share will probably level off around 5-10% for a while. It'll probably be years before that number approaches 0%. But 5% is low enough for even larger sites to stop supporting it, so... we'll see :)
Posted by: Jose Fernandez | July 8, 2009 9:22 AM
Lucas, you are incorrect. I have not said anything about IE7 being better than IE6, it's your (groundless) assumption. you are putting your own words in others mouth, which is wrong.
I was just saying that for those users who are even reluctant to upgrade to IE7 from IE6, it's strange that they'd go straight from IE6 to Firefox, as what those graph suggest.
The strange thing is that from those graph, it looks like the IE7/IE8 combined share remains constant for quite some time, while the non-IE browsers (Firefox, Safari, whatever) are mostly only stealing shares directly from IE6.
And here we are not talking about installed base, it's about usage share, and from a user's perspective, the switch from IE6 to Firefox is much more drastic than the switch from IE6 to IE7. So it's hard to imagine that an average IE6 user who refuse to switch to IE7 for so long, would decide to go directly to Firefox, in terms of browser usage.
Your own particular experience is irrelevant here, since you are already very familiar with Firefox, while the average IE6 users are those who have only used IE before and know little about non-IE browsers. For those die-hard IE6 users, it'd still be easier for them to upgrade to IE7, rather than going directly to Firefox.
Posted by: soshimi | July 8, 2009 5:38 PM
soshimi, while it may seem like the simplest explanation of the graph is that IE 6 users are moving to Firefox, I don't thnk that's the case. I think it's much more likely they're moving to newer versions of IE and at the same time some different users of newer IE versions are moving to Firefox.
We can't actually know what's going on with individuals or even cohorts inside of those percentages, but I think it's far more likely that some IE 7 and 8 users are moving to Firefox "making room" for IE 6 users to move to 7 and 8 without 7 and 8 lines growing from it.
Does that sound reasonable?
- A
Posted by: Asa Dotzler | July 8, 2009 8:21 PM
Asa, yup, I do agree that sounds much more reasonable, but still it's kinda strange that the percentage of people moving from IE6 to IE7/8 is almost exactly the same as the percentage of people moving from IE7/8 to Firefox, that's one hell of a coincidence. But then I guess the world is full of strange coincidences :P
I just find it looks kinda strange for the IE7/8 combined share to be able to remain so constant in a time like this.
On another note, I don't think there are any new machines currently for sale out there that comes pre-installed with IE8 yet? I don't know any PC vendor that cares to do Windows update before selling their Windows machines.
Posted by: soshimi | July 12, 2009 2:56 AM
IE6 market share also doesn't move because corporations will not update the browser if any vendor app they use needs to run on IE6 and won't run on IE7 or IE8, nor will corporations allow users to run alternate apps like firefox. So if a production machine needs IE6, it won't get approval for any updates unless a newer version works for everything and they are willing to roll it out. Firefox also doesn't provide Enterprise level zone security/patching features either that work with other tools. Corporations have to test all software for all apps, which can take months.
More and webbased apps use IE6 much better than any other browser. MS SSRS, VS, Outlook based Webmail apps, MS SQL Server, HighJump, Rockwell Automation, etc, etc, the list goes on for apps that are incompatible with anything but IE6.
Until vendors change, IE6 marketshare in the corporation will not change.
Posted by: cuz84d | September 21, 2009 1:47 PM
Another factor that is beginning to come into play is that web developers are no longer actively trying to support IE 6. Case in point: Just this year, I finally convinced our company to no longer specifically support IE 6 in newly developed web applications.
So, when these stubborn IE 6 users start seeing their browser fail to work in the next few years, I can see movement to other browsers accelerate.