SEO and Web Standards
Many of us who understand the importance of web standards have thought that search engines grant merit to web sites that adhere to web standards. Not because we conform, but because it has less clutter and is semantic. But we would be wrong. For the past year, I have been working on some experiments to help determine if developing web sites that adhere to the W3C Web Standards influences SEO ranking. The results are surprising, from a standards point of view. It appears that developing sites to W3C specifications receives inconsistent weighting within each major Search Engine and very inconsistent between the minor Search Engines.
My experiment
Make up two un-indexed keywords. Develop a page to offer eight equal links.
Each link will go to a new page with the file naming convention, representing each of the different types of HTML markup. Each page uses similar copy as possible, except for a unique content difference.
My experiments consisted of two layouts: a simple two column layout and a more complex three column layout.
The Simple layout included pages with the following HTML markup types:
- CSS semantic
- CSS not semantic
- Tables semantic
- Tables not semantic
The more Complex layout included pages with the following HTML markup types:
- CSS semantic
- CSS not semantic
- Tables semantic
- Tables not semantic
I tested how each development technique ranked over the course of the year. The research has mixed results.
In real-world SEO practice, I have witnessed improved ranking of websites that use highly sought after keywords such as "Vegas" and "Las Vegas". These websites have used semantic XHTML, and CSS and JavaScript. It seemed reasonable to expect that the results were partly a result of using well-formed markup. However, when I set out to determine if CSS vs. table layout and combined with and without semantic XHTML tags, I did not get the expected results.
My experiments have given inconsistent results not only from Search Engine to Search Engine, but also with their own results. Some do offer better rank to semantic code and that is consistent but the inconsistency lies in CSS presentation vs. tables or vice versa.
If I were to draw a conclusion and to offer my advice across all three major Search Engines, solely based on what type of development offered the most consistent results...
CSS with semantic code had the most consistent results, mostly second positions. Tables with no semantic code did the best most often, but also the worst at times. So I think that I will advice this. Do not build you site for Search Engines, but do not prevent them from accessing your site. Build for human visitors and deny nobody. Think of speed, accessibility re-use, and ease of maintenance. It should be obvious, you will find the best way to develop a site, since as far as SEO is concerned, there is no distinct favourite.
One thing the testing has shown: there is no benefit to using techniques that the web standards community has long since disowned. Few competent developers would now consider using tables for layout, even under the most trying of circumstances. None would ever use framesets.
The benefits of adhering to Web Standards are vast: accessibility, smaller files sizes, more effective code reuse, separation of content, better support on older browsers and non-compliant browsers, better mobile browser support, and it is easy to read for humans. (Though only those who need to edit the code will truly benefit from that.)
Why it SHOULD matter? It matters because you can put tags in context. This helps every type of visitor to your site and it should help the Search Engine type of visitor too.
So, why isn't Google offering a benefits to semantic (X)HTML?
Please note, I lost the reference to a Matt Cutts quote that I looking for about why it is not possible to penalize sites that are not standards compliant. Please bear with me.
Everyone can agree on this: there is no reason to penalize sites that do not follow standards. However, I do feel that some weight should be given to semantic sites by nature of the algorithms. The keywords will likely be found in the URL, the <title>, the <h1> and possible other <h?> tags and of course <p>, if in fact the document is about the keyword. How is this structure different from any other document, like a Word document for instance? Chances are if the filename, the title, headings and content all have the topic-specific keywords in the document. Using headings is just an intelligent and intuitive way to structure a document. Why should a web page be any different? An intuitive document should be more likely to be of value to a Search Engine user.
Offering a Web Standards adhering document to a Search Engine spider should only make the job easier for the spider. Tables break up the content flow, thus potentially removing content out of context. This is not advantageous to any user. Adhering to the Web Standard specifications, not in terms of building error free coded sites, but in terms of keep documents semantic should help the weight of a document in the search.
Next I am plan to change the order of the links and see if the results change. The order of my links was favouring the tables and no semantic code.