As a brief introduction for the non web-designers out there, Cufon is basically a way that we can use fancy fonts on the web without being restricted to the favoured few (eg: Arial, Times New Roman, Verdana etc..). For many years those stock fonts were the only choice we had and designers craved a way to ‘pretty up’ their sites and have the same freedom that print designers have.
A few solutions arose thanks to much hard work (probably unpaid!) by the founders of Cufon (Simo Kinnunen) and sIFR (Mark Wubben). These two achieve relatively similar results visually, but behind the scenes they work in very different ways, and each has their pros and cons. Both are based on Javascript and utilise the jQuery library, however sIFR creates Flash files (no support for iPhones / iPads then), whereas Cufon uses HTML5 < canvas > elements. Anyway, that’s enough background for now, this isn’t meant to be a debate between sIFR and Cufon, as there are a few more modern alternatives worthy of consideration (Fontdeck and Typekit) nowadays.
For various reasons we have deployed Cufon on quite a few sites (including our own!) and are pleased with the results. They take a bit of tweaking, but generally you can get the visual look you are after fairly easily. We normally use Cufon just for headings, as it would be easy to go overboard and use some funky fonts for normal paragraph text, but then users wouldn’t be able to select or highlight those bits of text if they wanted to.
So what is the problem then?
Recently I had a discussion with another SEO company who considered that Cufon has a detrimental effect on search rankings, and that they had seen other clients drop in the rankings because of use of Cufon for headings. Based on the way these font replacement tools work, I couldn’t see a reason why they would actually have a negative impact on rankings. If you ‘view source’ for a site that uses Cufon, the existing heading tags <h1>, <h2> etc.. are still contained in the markup, and so the search engine spiders would see these heading tags, just like normal headings. It is only after the page is loaded that the heading tags are replaced by Javascript with the Cufon <canvas> elements in the user’s browser. The heading tags still actually exist in the source, they haven’t gone anywhere!
My only concern was that as Google and the other search engines are becoming more advanced, I wondered whether they saw font replacement as displaying different content to users than what is presented to the search engines, which is a big no-no in SEO-land. In general terms cloaking of content can have serious results, as it even resulted in BMW Germany being banned from the Google index recently! Another loose theory I could think of when trying to see the other SEO company’s perspective was that Google looked at the code that was created after the Javascript ran, rather than the raw source code, but this just didn’t seem right from how the search engines work. So, I thought I would create an experiment to see whether it actually does have any impact.
The test…
I decided the definitive way to test this out was to create a basic page that uses Cufon and one that doesn’t, and then see which one ranks more highly for a made up keyword phrase that doesn’t exist in the index. Then to try and scale this up a bit I tried it with three different keyword phrases, resulting in six different pages, three that use Cufon and three that don’t. The picture on the right shows one of the pages. Obviously there are a lot of variables at play, however here is how I tried to address the main issues:
- I only used the keyword phrase once on the page, as the first word in the main <h1> title tags on all pages.
- I varied the text in the main paragraphs and heading so that duplicate content wouldn’t be an issue.
- The wordcount of the paragraphs and heading tags on different pages were kept roughly similar so that this wouldn’t have an impact.
- I still included the Cufon Javascript file in the <head> of all pages, even on the ones that didn’t use Cufon, so the file sizes and hence load times of the pages would be as similar as possible, as we know that Google takes page load speed into account.
Here are the different pages and their keyword phrases:
- Keyword: “Tstkwdsites”. Cufon Version. http://79.170.44.81/cfontest1.com/
- Keyword: “Tstkwdsites”. Non-Cufon Version. http://79.170.44.81/cfontest2.com/
- Keyword: “Blablajstatst”. Cufon Version. http://79.170.44.81/cfontest3.com/
- Keyword: “Blablajstatst”. Non-Cufon Version. http://79.170.44.81/cfontest4.com/
- Keyword: “Reydgbfgmsg”. Cufon Version. http://79.170.44.81/cfontest5.com/
- Keyword: “Reydgbfgmsg”. Non-Cufon Version. http://79.170.44.81/cfontest6.com/
What are the results?
Well it took a couple of weeks, but all of the six pages were included in the Google index, and all of the keyword phrases were now returning two results rather than the zero before.
Just to recap, there were two pages created for each of the three different made-up keywords, one using Cufon and one just using regular headings. As you can see on the picture on the left for the keyword ‘tstkwdsites’ just my two results were appearing, and the version that uses Cufon was appearing above the page that doesn’t use Cufon.
When I checked the other two keyword variations, these also had the Cufon utilising pages above the non-Cufon pages in the rankings.
The SEO Conclusion
As in all three keyword variations the pages that utilised Cufon appeared above the non-Cufon pages, I think it is definitely safe to say that Cufon doesn’t actually have a negative impact on search rankings and SEO.
The fact that all of the pages that use Cufon appeared above the non-Cufon pages is probably a coincidence, perhaps because the Cufon version was always on a lower number folder – eg: /cfontest1.com and /cfontest2.com folders. It would be jumping to conclusions to say that Cufon aided search rankings! I think if this was to be taken further then extending the sample size to test say 20 different made-up keywords would be a good idea, as would having them on separate domain names, not just sub-folders as I did.
I hope this has dispelled some myths banded about by some SEO companies, and that SEO-minded web designers can be free again to use Cufon without worrying about the ranking consequences.
