Recently, Quirk, in conjunction with Rafiq, decided to run a little test using an obscure search term. Two domains were created and set up with identical web content, optimised for the surprisingly unpopular term Purple Pot of Golf. I’m quite shocked that some niche marketing guru hasn’t jumped on the monetisation potential of this sure-fire traffic driver ;-)  I guess that’s the Internet for you; awash with ripe opportunities.

Anyway, the aim of the experiment was to test whether search engines rank pages higher when their content is placed above their navigation.

So two sites were set up on separate domains (hosted on the same server), with just one key difference in the websites’ structure;

-          The first one, hosted at http://stokes.za.net/  and with a page title Purple Pot Of Golf | Rafiq, has its navigation positioned before the content – let’s call it the Stokes Page from here on

-          The second, titled Purple Pot Of Golf | Lyle and hosted at http://ybc.za.net/, has navigation positioned after the content - we’ll call this one the YBC Page

With all other factors being equal, we hoped to isolate this single variable and determine its effects on ranking.

Both pages were given a link from the Quirk homepage, titled test 1 (Stokes Page) and test 2 (YBC Page) – effectively this meant there’d be another variable but one of the links would have to be first so we saw no way around it. So, the moment of truth. How would the big G rank them?

A search for the keyphrase, (we first just typed in the phrase and later repeated the search using quotation marks) ranks the YBC Page first, damning the Stokes Page to supplemental hell. Of course we expected that one of them would have to get flagged as duplicate and the question was really just which one it’d be.

Checking the cached times we see that the YBC page got indexed at 17 Aug 2006 12:20:28 GMT while the Stokes page got cached at 17 Aug 2006 00:13:44 GMT. -  So while the Stokes page got indexed fully twelve hours earlier, it’s still considered to be the duplicator!

Interesting indeed.

Then, we checked what Yahoo! would do.

For the same search, Yahoo! ranks the Stokes page first and considers the YBC Page the duplicate. We don’t have the exact times of indexing but my assumption would be that the spiders found them in the same order as Google’s spiders. Running the search again using quotation marks, we’re surprised to find that now only the YBC page ranks, with the Stokes page supplemental.

These are just our initial findings, and we’re still investigating to be able to draw proper conclusions but at this point what I find interesting is:

-         Google appears to favour the YBC page, which has the navigation after the content.

-         This might suggest it considers the page more relevant

-         Using quotations to force a search for the entire string yields pretty much the same result

-         It takes Google twelve hours to go from the test1 link to the test 2 link…unless it came back an indexed the YBC page a second time and ignored the Stokes page.

-         On Yahoo! the Stokes page is favoured.

-         I’d speculate that this is because that page was indexed first

-         Searching the exact string on Yahoo! finds the YBC page again

-         Perhaps this suggests something about how Yahoo!’s duplicate content filter works

In hindsight, one variable we’d overlooked relates to the titles of the pages. The YBC page uses “Lyle” in the title. Since Lyle is an employee of Quirk, his name appears elsewhere on the site and this may mean the link to that page gets a boost of relevancy over the Stokes page.

We’re working to find any other variables that we may have missed and will hopefully be able to draw some solid conclusions from all of this. As always, we’d love to hear what you think about it.

Keep watching this blog for updates…