Learning from Nonsense

[sdadesky]

I think the best way to understand Google's algorithm is by using little known search terms. If almost no searches have ever been performed under a given term,
and little or no manipulation of the term is likely to have been undertaken, the results can be very revealing.

Pursuing that line of thought, I entered a friend's name in Google, who run's a travel web site in Germany.
So I Googled his name, Guenther Insam:
http://www.google.com/search?q=Guent...en-US:official

Not surprisingly the top listing was for his company. In fact, it's one of those canned "Sorry we've got an error" pages.

Then I started digging further. Why that page? I looked at the content, expecting to find his name mentioned one or two times. But there is no mention of his name anywhere.


So I thought, maybe his name appears constantly on the bottom of many pages linking to this one. But I find no backlinks for this page... Nor is his name actually mentioned on the main travel jungle site, anywhere I can see.

My tentative theory is :

- Google ALWAYS looks up the name of a company and matches it back to ownership. It is a matter of public record that Guenther Insam owns the company "TRAVELCLUB.DE GMBH" mentioned on the page.

If that is so it adds credence to the theory that Google always matches up web registration with ownership when trying to weed out site spam.

Anybody else have a different take on this?

[dburdon]

Are Google using Whois to gather this information? They can get the registration data from there.

[Faglork]

Originally Posted by sdadesky

Anybody else have a different take on this?

Yes. If you look closer at the SERP: The result shown contains the text from the cached version. The cached version includes the words "Geschäftsführer Günther Insam".

No magic here ...

faglork

[dburdon]

Essentially the test for any hypothesis on search should in the more competitive categories. Its where more algorithm factors will come into play.

[aaron2005]

I use godaddy which has domain privacy, so when you check whois it says "contact godaddy" or whatever. Can this hurt a website?

I am seeing noticeable better results in my sites that have google sitemaps and the bar code (as I like to describe it) in root.

Anyone believe this helps?

On another note, I also believe having a name and address on the bottom of your pages is also a trust key as well, though I can not prove.

Good find sdadesky, it's nice seeing someone who goes out there and looks around. ;)

[nipplecharms1]

Originally Posted by aaron2005

I use godaddy which has domain privacy, so when you check whois it says "contact godaddy" or whatever. Can this hurt a website?

On another note, I also believe having a name and address on the bottom of your pages is also a trust key as well, though I can not prove.

Aaron,

Point 1 - I too mask my ownership with my hosting company. It has cut down on Spam. But I am wondering, like you, if maybe I am shooting myself in the foot, or some other appendage!

Point 2 - I purchased one of the "trust" logos for the site. You really do have to jump through a bunch of hoops to get credited.

We did ok without it. Unfortunately, we purchased it and loaded in on the site during the "Mick Effect." Our sales tanked. Was one or both of these phenomena responsible? I have no clue. I am guessing it was Jaggerizzle. I wanted to see what the site would do after the trust logo was installed. I think I will have to see if we get back to the first page and then look at sales.

Just my buck three and a fifth.

Michael

[brian.mark]

I don't think this will harm you rankings at all. Google is a registrar, and as such has access to any private registrations as well as public ones. What they're looking for is 10,000 registrations to the same person (which may accidentally happen in the ISP style of protection mentioned by nipplecharms1), but more specifically I've heard that the idea was to discount links between sites owned by the same person / group of people.

This brings up another interesting thought with links. Reciprocal can be detected easy enough. Site a to b, b to c, c to a can be detected with a little more effort, but how about a to b, c to a where the registrant of b and c are the same? Could be interesting.

Brian.