Why webfont services are the future of fonts on the web

The debate about webfonts just reached its peak at a panel discussion on this year’s TypeCon conference. (You can listen to the 2 hours audio at Typographica and read summaries at i love typography and NiceWebType.) The discussion left a lot of people puzzled about how the future of fonts on the web might look like. Here is my outlook  …

As a type and web designer, webfonts were always of great interest to me. As a type designer I was shocked, when Safari started to support linking TrueType and OpenType fonts (now usually called “raw fonts” in this context) with Safari 3.1 in march 2008. Web designers could now upload my commercial fonts to a public web server where everyone could easily download them. One year later almost all type foundries seem to remain in this shock.

But since I am also a web designer I began to see it from a web designer’s perspective. I thought about ways to improve the protection of raw fonts and also about new ways of licensing, for example web services that would host the fonts and deliver them whenever your website is being viewed. Apparently I wasn’t alone with such thoughs. Currently several webfont services are being developed. Meanwhile the percentage of users who can actually see webfonts keeps growing. Firefox 3.5 has implemented the possibility to link to raw fonts and so will Opera with version 10. So all major browsers support webfonts now! But there is also catch: There is currently no single webfont format, that all browsers and type foundries can agree on. To understand the current dilemma we need to look at the history of webfonts and the CSS “@font-face rule” that makes webfonts possible.

Webfonts history

Webfonts are font files embedded in websites using the @font-face rule. This technique was first available around 1997 in Netscape Navigator 4 using Bitstream’s TrueDoc standard and in Internet Explorer 4 using Microsoft’s Embedded OpenType format (.EOT). But both formats were hardly used, mostly because the render quality was rather poor these days and only well-hinted system fonts looked good on websites.
Today the render quality in all operating systems and browsers has improved dramatically and webfonts get a second chance. But while Microsoft sticks to its EOT format since version 4 of Internet Explorer, the developers of Safari, Firefox and Opera went a different route and implemented @font-face in a way where only raw fonts (TrueType/OpenType) work. So if web designers want to use webfonts they need to include both EOT and TTF/OTF to make it work across all major browsers.

Raw fonts and the intellectual property

For web designers using both EOT and TTF/OTF may not be perfect, but its not too hard either. The problem is that almost all type foundries don’t allow the use of their commercial fonts as “raw fonts” on a web server. They are afraid…

  • that there customers might upload their print fonts to a web server without a proper web license.
  • that anyone can download those raw fonts from the browser and install the fonts on the local system, without having paid for it at all.

So the foundries are hoping for a new solution, a new web-only font format. But it’s somehow ironic that at the very moment, when webfonts become finally useable across all major browsers, the foundries are trying to invent a new format that will currently not work in any browser at all and will take probably 6 to 8 years until it reaches the same level of support that EOT and TTF/OTF have today. Or as someone on a German blog put it: “The type foundries are standing on the platform arguing about the seating arrangements while the train has already left the station.”

.webfont format

One such new webfont format proposal is .webfont by Tal Leming and Erik van Blokland. It’s a zipped file that includes the font data and an XML file with meta data, such as permissions on which website the font can be used. Many foundries support this proposal, but it will certainly take years until it is included in all major browsers.

OTW and EOT Lite

The foundry Ascender proposed a new web format called OpenType Web (OTW). It is based on regular TrueType/OpenType fonts with some additional information and some changes in the format, which make it impossible to install this font on a local machine.
But Ascender realized that it may take too long to implement such a format in all major browsers, especially in Internet Explorer. That’s why they created a new proposal called EOT Lite, which is based on Microsoft’s EOT format but without the compression and the URL-binding, because these were features the developers of Opera/Firefox and Safari rejected. EOT Lite already works in Internet Explorer and would only needed to be included in the other major browsers. But then again: URL-binding and compression are the main reason for the existence of the the EOT format. Removing these features from a webfont creates a bad compromise. “And so, as I see it, an EOT Lite file is nothing more than a TTF file with a different file extension.” Quote from Richard Fink

The status quo

So there we have it: Webfonts are now supported in all major browsers, but most foundries don’t agree to the use of their fonts as raw fonts on web servers, so the web designers can’t use them. New formats are proposed, but they will take years to be included in all browsers. So what do we do? Use only freeware and open source fonts?

Webfont services

I think these first examples of commercial webfonts show very clearly how commercial webfonts can enrich websites.

Professional web designers want professional typeface for their work and corporate clients want to use their corportate typeface on their websites – and there is a solution that works today: Webfont services like Typekit, Fontdeck and Kernest. As I said in the beginning of this article, it’s the solution from the web designer’s perspective. An external service that you embed in your website like a route plan from Google Maps. You don’t need to worry about different font formats, browsers version, compression and subsetting – the webfont service will automatically deliver the right CSS and the right font file to the visitors of your website. And these webfont services can even include some level of protection, so the font foundries might agree to the use of raw fonts with these services.

Using a webfont service is certainly a very different way to license a font and many are being sceptical about this concept. Here are some common concerns:

What if the service fails and the fonts are missing?

Don’t worry, the developers of those webfont services are well aware of this issue and will host the files on Content Distribution Network (CDN) that can guarantee the constant delivery of the fonts all over the world.

But I don’t want external services, I want all files on my server!

Yes there was a time when a website was nothing more than a folder with lots of files in it. You could stick it on a CD-ROM or upload it to webserver, it would always work. But today there is hardly any website left, that doesn’t use a permanent database connection or external services like Google Analytics. Today a website must work on a device that is online and therefore external services are a commonly accepted tool.

Don’t forget that webfonts are a piece of software that needs maintanance. Your page might get a new domain and you need to update the URL-binding in the fonts. You site might be translated to Russian and you need to change the subsetting to reflect that. A new browser version might handle some @font-face descriptors or OpenType features differently and the fonts need to be updated …

And that’s where you can rely on a webfont service that will handle this automatically and generate all the necessary files on the fly.

So now I have to pay forever to use a font instead of just once?

Yes and no. Sure, you now just rent an access to a certain font file. But that has also some big advantages for the users. Traditionally fonts are licensed for a flat fee and only after your payment do you get a chance to to use the fonts. So you are forced to buy a pig in a poke. It may not work for you or your client might reject the font. So even if you never use it, you have paid the full price.

But with webfont services it would be really easy to offer a trial period. You could play with several fonts, present them to you client and when the trial period is over you could decide whether you stop using the font or whether you start paying the monthly fee. And if you, let’s say after 6 months, don’t like the font anymore, you could switch to another one or just cancel the subscription and you have probably paid much less that you would with traditional font license.


I asked Jeffrey Veen from Typekit and Veronika Burian from Type Together about their opinion on webfont services. Here are their answers …

RH: Web designers often say they want full control over the fonts and don’t like external services. What do you think are the most important advantages of using webfonts as a service?

»I think the tide is changing a bit. I see many sites linking to third-party Javascript frameworks like jQuery, or using Amazon S3 for image hosting.
There are really two things to consider: reliability and performance. Typekit is hosting fonts on a global content delivery network with redundant servers in North America, Europe and Asia. Most organizations can’t afford to host with providers like this, so the reality is that Typekit fonts will be served more reliably and with lower latency than most of our customers’ sites.

The second thing to consider is how browsers request content from servers. Many people assume that adding third-party servers will slow them down, but the reality is it actually helps quite a bit. Browsers have limits on how many simultaneous requests they will make to individual servers. By hosting parts of your site on different domains, the browser will open more connections and load things like fonts and stylesheets much faster than if they were hosted on the same server.«

RH: Type foundries are used to license their fonts to commercial design studios and other corporate clients. Webfonts might reach a much wider audience – do you agree? Which customers do you want to address with Typekit?

»We’ve identified a number of audience segments that could benefit from webfonts, and Typekit will have multiple ways of addressing them. Working designers who maintain brand identities only scratches the surface of what’s happening on the web today. Social media has blurred the distinction between publisher and consumer. There are an estimated 184 million blogs in the world today, all of them building a unique voice and identity. Some of us may see the vast unwashed masses on MySpace as a design disaster, but it also represents a first step towards people expressing their online identity visually. Webfonts will accelerate all of that.«

RH: What’s your opinion on the debate about webfont security and the ongoing discussion about better webfonts formats?

»We’ve written about this on the Typekit blog, both in our support for Tal Lemming’s and Erik van Blokland’s .webfont proposal, as well as our in-depth explanation of how we’re serving and protecting fonts.

I see a couple issues here. First, the type industry is fortunate to be able to learn from recent history. Media companies have struggled for a decade with the balance of protecting their assets while enabling as big an audience as possible. They’ve tried different DRM schemes, which have pretty much all failed. They’ve tried strong-arm legal tactics and learned that suing their best customers is a short-term solution at best. There’s a lot of precedent there, and I think we can skip over a lot of earlier mistakes.

But I also believe there are unique qualities to fonts that make the analogy to music, photos, and video a bit weaker. Unlike, say, stock photography, fonts are pieces of software used to create visual communication. Not only that, but the same software used to create a work is needed to view the work. So type designers are having their entire source code sent to every person who views a web page. That’s the reason why so many of the foundries are being cautious and proposing new formats. There are parts of this new world that have no precedent and we all want to make sure we think about this very carefully.«

RH: Webfonts are uncharted waters for type foundries. Why do you decided so early to team up with a webfont service to bring some of your fonts to the web?

»We think it’s a great idea to bring finally good typography to the web. We understand, that web designers have been long keen on having access to a broader range of typefaces. Various workarounds exist, but never took off on a big scale. So, when TypeKit asked us to participate in their new scheme, we thought that it is a very promising business model that is fair to both parties, the web and type designers. It has the potential of promoting good typography in a legally safe environment and at the same time bring some revenue to the type designers, who spent long hours designing good fonts. It might also push type developers to add extra value to their fonts for the web, by preparing them for on-screen viewing, i.e. hint them properly.«

RH: What are your expectations for this new market, its growth and customers? Are your customers requesting webfont licensing?

»Yes, we have been asked a few times about our policy on licensing for the web. At the moment, we allow for font embedding, but not font-linking, except via TypeKit. We think, it’s a growing market with a lot of potential. The demand for fonts on the web seems very high. However, it’s also a competing market and the players (browsers, foundries, W3C) are not unified enough and don’t have a common opinion or interest in the same issues. I fear this slows down the process.«

RH: What is your opinion about the whole issue of webfont security? Are you satisfied with the way your fonts are obfuscated through the Typekit service?

»It’s impossible to completely avoid font piracy. Our fonts have been hacked and are already available illegally on various sites. Personally, I am not so bothered unless the fonts are redistributed and/or used for a clients project with monetary gain. As far as I can tell, that hasn’t happened yet. TypeKit is using several techniques to make it harder for users to gain access to actually working fonts. Although it’s possible to download the font files from the source code of the website using the fonts, it takes some effort and technical font knowledge to make them into fully working fonts again that can also be printed. I seriously doubt that any decent designer would use them for a client project and risk legal prosecution. So overall, I believe TypeKit is doing a good job in regards to webfont security.«


If webfont services manage to find a pricing scheme that works for the users and the foundries I predict them to be extremely successful in the next couple of years and webfonts can be a huge an unprecedented market for the font foundries. A new webfont format like .webfont might emerge over the next years and can then be available for conventional licensing (i.e. without renting). Webfont services can easily include such a new format, but it will take a very long time until it can make the current use of EOT and TTF/OTF superfluous.

Related links:

  • Webfonts.info
  • Webfonts.info RSS-Feed
  • Follow me on Twitter


  1. Yannick 2009/07/29 at 8:55 PM #

    How can a font format allow users (the web readers) to download a font to display a web page but not to use it on their computer?

    It’s impossible, at least with a standard used by open source browsers.

  2. ralfherrmann 2009/07/29 at 8:57 PM #

    You can check the Typekit blog. They mention some of the possible techniques to achieve this.

  3. Bil Hill 2009/07/29 at 9:23 PM #

    A good summary of the story so far.

    Ironic that your page displays on my machine with large areas of machine-skewed “imitation italic” which is unreadable…

    I also see the need for offline support of content, in which case the fonts would need to be cached locally.

  4. ralfherrmann 2009/07/29 at 9:56 PM #

    Bill, you’re in a position to improve the quality of the rendering of “your machine”.

    Just so you can take a look how the vast majority of my readers see this page:

  5. kupfers 2009/07/29 at 10:47 PM #

    Graublau needs a great deal of spacing improvement

  6. Tor Løvskogen Bollingmo 2009/07/29 at 11:39 PM #

    “But today there is hardly any website left, that doesn’t use a permanent database connection or external services”

    That’s a bold statement.

    Btw. does the EULA of regular typefaces cover usage on websites via font-face? If not, we’d be free to use the raw files.

  7. baxter 2009/07/30 at 10:00 AM #

    I’m running Firefox 3.5.1 and I don’t see that font. Custom fonts elsewhere on this page seem to be working though.

  8. ralfherrmann 2009/07/30 at 10:19 AM #

    Btw. does the EULA of regular typefaces cover usage on websites via font-face? If not, we’d be free to use the raw files.

    The license agreements always prohibit redristribution – and this is what you do, when you upload a font to a web server to be used with @font-face.

  9. make-up artist 2009/07/30 at 11:15 AM #

    Baxter, same here!?.. strange

  10. Tor Løvskogen Bollingmo 2009/07/30 at 11:21 AM #

    Depends, you can also see it as usage. It’s not redistribution as in selling the font or sharing it for the purpose of giving it away for free. Just as posting an image on the web, it too can be stolen - but it’s main purpose is for people to look at, not redistribute it.

  11. ralfherrmann 2009/07/30 at 12:41 PM #

    Depends, you can also see it as usage.

    The usage is subject to the terms of the End User License Agreement. Uploading commercial software to a public server is not allowed in any such license.

  12. BrightBold 2009/07/30 at 4:52 PM #

    Great article - I have already recommended it to people looking for a good summary of the current state of web type.

    However I’m not seeing the font as you intended either. In FF 3.5.1/Win (with ClearType enabled) I appear to be seeing p.interview-answer in Graublau Web Regular - definitely not italic. Looks OK but a little thin to be highly legible. In Safari 4.0.2/Win I get an ugly machine-imitation bold.

    Yikes - I’d better triple-check the sites I’m working on with @font-face! Webfont services can’t get here fast enough.

  13. Richard Fink 2009/07/30 at 6:06 PM #


    Thanks for the mention. Although I don’t agree with you that the new EOT lite is a “bad” compromise. It’s garnered a lot of support.
    Frankly, I find the whole Web Fonts situation so muddled, all I can say for sure is that font services are certainly a part of the future. As will be experimentation and confusion.
    Developments over the next couple of months will tell us a lot.


  14. John Daggett 2009/07/31 at 10:53 AM #


    Looks like a font loading bug in Firefox. I’ve logged this and will take a look at it as soon as I get a chance.


    As for the thinness of the face on Windows, that’s unfortunately how OpenType PS fonts render on Windows. They render with grayscale anti-aliasing, rather than subpixel antialiasing and that’s why it looks thin. I’ve been pestering the Microsoft folks about this and trying to get Adobe to do the same. For comparison, in Safari change the font smoothing preference from the default “Windows Standard” to see what it should really look like!! That renders using the Windows port of the Mac OS X font rasterizer rather than the Windows default rasterizer.

    John Daggett
    Mozilla Japan

  15. Richard Fink 2009/07/31 at 7:43 PM #

    Ahh, rasterizers… John Daggett wrote:
    “For comparison, in Safari change the font smoothing preference from the default “Windows Standard” to see what it should really look like!! That renders using the Windows port of the Mac OS X font rasterizer rather than the Windows default rasterizer.”

    One of the things web fonts is going to highlight for all of us is just how deeply cross-platform rendering differences go. The “web safe” fonts have largely shielded us from these things, so far.
    And then there’s the question of how display technology fits into all of this?
    This is the road we have to travel down now. Web Fonts are just a part of the readability mix. It’s going to take awhile to sort this stuff out.
    It’s just the beginning.

  16. BrightBold 2009/08/04 at 2:53 AM #

    @John Daggett - Thanks, that is a big improvement in Safari. It’s still not italic, but other than that looks like Ralf’s screenshot.

    @Richard Fink - Just the beginning indeed. A lot of complexities here. The more I read the more I realize how incredibly complicated this is; I’m grateful there are lots of very smart & knowledgeable people working on it.

    - Kelley


Leave a Reply


More in Web Design & Webfonts (15 of 27 articles)