Tested and success!

I love this forum!

About 3 months ago I saw Google SERP's ranking climb for a keyword for a client. We had zero keywords as a link coming in for this keyword.
I figured that it must have come from a word within a paragraph, e.g <p>Car parts companies like vvvvvv.domain.c0.uk have a great selection</p>.
You will notice that the domain does not have anchor text.
It looks like Google see's the link and then looks at the paragraph, strips it of standard text and assosiates the url with the keywords left within the paragraph.
I started testing this and it looks like it works. Infact it does work!

Has anyone else come across this?

How do you know that this page doesn't have other, normal, inbound links?

It was tested on a new page.

So that wasn't an outbound link, but just an unlinked domain in the text?

Hi CommonDavid,

The link within the paragraphs was like this ( <a href="http://www.domainname.co.uk">www.domainname.co.uk</a>)
There was no anchor text yet Google picked up a keyword from within the paragraph and assosiated it to the linked site.
Weired i know.
I found out by adding a new page for a made up product name and tested by adding 10 incoming links like above with the made up product name within the paragraph and hey presto, it was ranking in Google.

Unfortunately this was tested on a clients site otherwise i would give you the full details.

I don't know if this is related, but I looked recently at images related to Japan in Google Image search and found a photo that did not fit the word/subject I was searching for.

I went to the site and the photo was sitting in the middle of someone's description of their trip to Japan. The image, the photo's file name and the alt tag had nothing to do with Japan.

And, the word/subject I searched for was not in the surrounding text, but it was a related word.

So it looks like Google made an association to the photo based on the proximity of the text, and chose to make the association using a related term that does not exist on the page at all.

I have to conclude that Google makes simular associations on non-image elements as well.

Ed

This is very interesting. If what we have discovered is correct then buying a link with your favourite keyword can now be devalued by the search engines.
If you think about it a single text link within a html tag (<p>, <td>) etc or between a <br> is completely unnatural and the search engines can easily devalue these.
By choosing a relevant keyword to tag onto a separate link within a paragraph looks like this could be the way Google is heading. In fact it may not even be a linked domain; it could simply read the www. etc as text and also attribute the keyword to it.
Maybe this is why Google are contacting sites to get sitemaps up and running so they can discount these as paid links.
Just a though.

IMO Google has always placed some weight on the surrounding text for some time. Even back when reciprocals were king the description you used for your site would have an effect on the strength of the link.

How many people actually link the keyword to the website in natural usage. Not many. But if you a sentence in either direction the link's purpose comes into focus. This shouldn't be too difficult for Google to figure out but It's good to see some direct evidence! Don't give away all of our secrets!

If a link is out of context, i.e. the surrounding text does not support the use of the link, then it's pretty easy for Google to discount it. I bet it makes no difference if it's a high PR page, if the link is out of place, don't count it.

In my example above with the image search, the photo stood out from the rest like a sore thumb. Since Google can't interpret the photo, it made an indirect association based on the surrounding text. The funny thing is, they associated a related word and did not use a word from the actual text.

I think this reveals a lot. Maybe an opportunity for someone to build a tool to help write text to maximize the juice behind links.

Ed