Note: I am not a Microsoft expert/insider, these are just datapoints compiled together by an interested amateur. Greg Whitworth of the Microsoft Edge team replied here!
It's widely known that IE11 is scheduled for End of Life in 2025.
It's less widely known that this calculation is tied to the end of life of Windows 10, which is indeed in Oct 2025.
It's much less widely known that Oct 2025 is the date for extended support of Windows 10, and that mainstream support for Windows 10 ends on Oct 13 2020.
This means "Mainstream End of Life" for IE 11 is this year, not 5 years from now. In fact, it is exactly five months from now.
Mainstream vs Extended?
Not being a Windows sysadmin, this was news to me too. But it makes sense on deeper inspection. Windows 10 was released in July 2015, so a 5 year "mainstream support" period with free upgrades sounds about right.
Enterprises pay extra for an additional 5 years of "extended support", which covers security updates as well as nonsecurity updates, but not new features.
Beyond that, you might be able to buy "Extended Security Updates" for 3 years. I could not find Win 10's pricing, but Windows 7 cost up to $200 per device per year.
An additional complication to consider is that there were 3 Enterprise LTSB/C versions of Windows 10. The 2025 Extended EOL date corresponds to that first version released in 2015, but there are two other versions, with EOLs going out to Jan 2029. So, any death of Win10/IE11 will be staggered rather than immediate.
However this is still true: The first end of Mainstream Support of Windows 10 is in Oct 13, 2020. This year.
End of Life
I'll confess that "Mainstream End of Life" is a term I made up. Microsoft doesn't recognize it. In fact, as far as I can tell, Microsoft doesn't even use "End of Life" anywhere on their documentation. They have "End of Sales" and "End of Service", and "End of Support". No End of Life. It's a colloquial term.
When developers talk about a "Windows 10 End of Life", they seem to mean End of Support. I write this blogpost to point out:
- End of support comes in two stages: Mainstream and Extended
- Mainstream Support ends in 5 months
- Non Enterprise users do not get Extended Support.
This is what I mean by "Mainstream End of Life" of Windows 10, and therefore IE 11.
Why the Timing is Right
Microsoft's own approach to EOLing IE11 has been more subtle:
- Microsoft itself has argued against using IE for everything since 2019.
- It then announced IE compatibility mode in Chromium-based Edge, which began rolled out on Jan 15 this year.
- That release was manual, but the full rollout of "Edgium" is underway. In February, rollout via Windows Update began. This rollout happens gradual stages until a full, wider release happens in the summer (aka, before Oct 2020)
Microsoft also has had precedent in dropping support early for IE - and Edgium with IE11 mode may pave the way for end of Extended Support to come sooner than 2025.
The coronavirus pandemic also had an unexpected benefit to this movement - because most Win10/IE11 installations are on older office machines, working from home means that overnight, millions of Win10/IE11 users realized they could move without significant impact. (I'm sure there are plenty of Win10 laptops so... idk)
IE11's market share
I'll preface this by noting the obvious: browser market share varies widely depending on what your customer demo is. You care about your customers more than you really care about global averages. But just to round things out where they stand right now:
- Statcounter has global share at 1.41%
- W3counter has (global?) share at 1.75%
- The US Government has share at 3.7%
- You can see browser share by your country here (for some reason Saint Helena, Christmas Island, the Vatican, and Korea randomly love IE)
With the facts already laid out, it's not hard to see that this year will be the first year Edge overtakes IE:
Recommendations for Developers
The development cost of IE11 support is high. Enumerating that cost is out of scope for this blogpost, but at a minimum, you should be employing the module/nomodule Differential Serving approach for serving legacy JavaScript. This ensures you don't penalize the ~98% of your users with modern browsers for the 1-2% of IE users.
However, even that approach doesn't account for CSS support and IE11 browser bugs/quirks. Those will still take up code weight and developer time. The key thing to realize here is that all code and developer time spent on IE11 support is effectively subsidized by other users on modern browsers. In theory, they all pay you (roughly) the same revenue per user, but you spend disproportionately more resources supporting legacy browser users. Hardly seems fair.
Therefore my recommendation is this: By IE11 Mainstream EOL (13 Oct 2020),
- Explicitly drop IE11 support for all non-Enterprise users
- Offer paid IE11 support for all who need it past mainstream EOL (be careful to price it correctly for its true cost) (or use https://quik.dev/)
- Publicly notify a registry of peers in your industry who have also decided to stop support, like Dailymotion, GitHub, Stack Overflow, Zendesk, Slack, Atlassian, YouTube
Point 2 helps you justify this move as a potential revenue stream - if you are doing this work anyway, you might as well get properly paid for it instead of stealing time from other users.
Point 3 is there because this is a collective action problem - this becomes more successful the more people do it, particularly within peer groups. I considered setting up a change.org petition, but have not done so because I don't consider myself a fit leader for this movement - I am just pointing out facts for people who may not know them. But if you tweet at me I will collect you and your company's vote and help whoever wants to take up creating this coalition (is that you?! please be you!).
Mike Sherov has also advocated actively blocking IE11 based on lessons from the Deaths of IE6 and IE8. Since we started getting automatic upgrades to Edge in February, this is finally a feasible IT policy you can justify.
What Now?
I know I'm writing for a very narrow audience. I expect developers reading this post to be in 3 camps: EITHER you already don't support IE11 (supports es6-module is a good browserslist target, according to Jason Miller), OR you work at a BigCorp and you can pry IE11 from their cold dead hands, OR you work at a company that basically supports IE11 because your competitors do and because, for now, IE11 share is still above >1%.
If you fit that last description, I wrote this for you. We have a window, right now. You have 5 months to advocate internally, with all the stakeholders, and help make the Internet get a tiny bit safer and our industry a tiny bit simpler. You can also nag your users to update their browsers.
I'm rooting for you.
Top comments (12)
Glad to see Chrome's success on the Windows Desktop! It's a great browser for development.
The fact that IE has crashed and burned couldn't have happened to a nicer company. You know the one who repeatedly, for years; threw their own adopters under the bus.
Thanks for the article Shawn.
But Google didn't make IE 😉
I would love this to be true but unless Windows 10 now automatically comes with Edge, IE11 will need to be a fallback, we could all slowly add messages to our sites with links to firefox, chrome and edge to get people to download.
I love to drop IE11 however I have clients who still have users using it and to them that money.
i dont understand. Edge is the default browser... and they're shipping Chromium Edge with IE11 compatibility via Windows Update.
I might be mistaken, but when you install windows 10 you get ie11 by default, has microsoft completely changed that via an update and would that be all windows 10 users?
Its underway, but in stages. Msft rolled it out to Release Preview insiders in Feb (gadgets.ndtv.com/apps/news/microso...). The next step is Home and Pro users, planned this summer. zdnet.com/article/microsoft-kicks-... It seems this does not affect Enterprise/Education customers though :(
Freelancer here, but will wait in line to sign up that petition when it becomes available.
Two points:
Don't web bundlers do most of the fallback things. I mean make separate bundles for IE11. With CSS it doesn't really matter anyways. If your user is using IE11 let the user suffer the horrible ui layouts.
With JS errors, it's a concern. The app becomes unusable. This can be totally avoided using webpack or rollup.
Angular has a nice build system where it only brings up a polyfill if the first script it sent says what kind of browser it is. Now for CSR apps this is an extra http call and user ability will be slow.
For SSR you get the user agent anyways. Can serve bundles accordingly. I think optimal use of module bundlers is what can solve the big problem regarding IE11
To add to that. Android development also of suffers from the same problem. If you are building apps like Whatsapp or Facebook you have to support builds all the way to Android 4. That's horrible for a developer.
But implementing fallbacks are what defines developers top. Find the automated solution to a problem is what makes development interesting for me.
Hey - what kind of users are "Enterprise users". Is it businesses ... or could it include schools / councils etc. I'm particularly curious to understand whether schools (who seem reluctant to upgrade) could keep IE 11 going for longer ...
Most people overestimate what they can do in one year and underestimate what they can do in 5 years 5 months and 2 days.