SEO Experiment on HTML code
February 4, 2006 – 3:46 pm
A great article written over at Mike Industries called the Roundabout SEO Test.
An experiment done to see how rankings fare depending on coding methods used. Below is a summary followed by my own conclusion/notes
For a quick summary - one keyword was used
1. H1 tag beat a b tag and also a b tag plus another mention in plain text
- H1 is more dominant
2. Nested tables does not create a major negative effect on your rankings
- CSS is still the better choice, but tables are not *so* bad
3. A valid table was indexed, two invalid tables were not even indexed
- Valid code should be used to be certain of being indexed. Invalid code can be indexed as I've seen it, but if certain tags are left unclosed etc. such as in this experiment you could be completely ignored.
4. H1 tag after a paragraph still ranks higher than a H4 tag at the top
- Semantics plays a part and the H1 tag is more dominant. Thus semantics and hierarchial coding should be observed.
5. External link text is more important than title text, then filename and then a H1
- link text is often out of your control but if you can, get description anchor text used to link to you, then ensure that descriptive page titles and descriptive filenames are used - again observing semantics.
Bear in mind this is from the SEO point of view, not a development point of view. It proves that certain coding methods out rank others, well written code will always help. The PR of a few sites were also taken into account and PR does not play much of a part in ranking at all.
From this experiment you can take the created proof that the priority list of creating a web site from an SEO point of view should be:
External anchor text - a nice keyword/phrase to be used (look at who's at the top of Google for "miserable failure" ;))
Page titles - descriptive and well written
Filenames - don't go spammy, but either one or two word filenames, use a hyphen to seperate them
H1, H2, H3… - Use hierarchial order and don't go mad. use the number after the H to denote how often to use them ie. H1 once, H2 twice, H3 three times. You could probably go further on a longer page eg. my own blog has a H2 tag per post. And there are 10-20 posts on the front page. But there is a lot of content separating each title ie. it's a long page, not a spammy one
And ensure your code is valid!
Other parts of HTML code was not experimented on here such as alt and title attributes, strong and em tags, lists and internal anchor text. It's an idea that I may try myself when I have time to expand on it further.


One Response to “SEO Experiment on HTML code”
"4. H1 tag after a paragraph still ranks higher than a H4 tag at the top
- Semantics plays a part and the H1 tag is more dominant. Thus semantics and hierarchial coding should be observed."
thats a new one to me - better go and fiddle with thet 3col css layout!
By pest on Feb 4, 2006