Similar or related sites - is the list in rank order?

[Hissingsid]

I was just looking at the similar link for a few of my competitors and was wondering what other folks feelings are regarding ranking of the related sites.

Are the related sites ranked by most similar or highest ranking in organic results that are also similar?

In my top 10 for one particular 2 word term all but one of the sites has a closely comparable list of similar sites. Their similars are similar ;-) One stands out because its similars are a different profile to all of the other sites in the top 10. From a "common knowledge" point of view this is what you would think, it's kind of obvious that this site is not similar to the others. It makes you wonder why it is there. Surely if the other 9 are correct as an "answer" to the question then since the other one is not similar then it is incorrect.

Which brings with it another thought. Should Google be using similarity as a ranking factor? Alternatively is Google throwing in one (or perhaps more) screw ball result in some important serps as a real time test of correctness. In other words if users select that odd ball and don't bounce straight back does that tell Google that perhaps the others are not such a good "answer" to that question after all.

I think this topic would make an interesting topic for discussion if nothing else.

Cheers

Sid

[tedster]

I remember you started a discussion about the changes in "similar" or "related" websites [webmasterworld.com] a little while back. No doubt something shifted for the better then, but I still haven't see any evidence that related: sites gives any view into ranking parameters.

There certainly might be something we can glean from this area, I just haven't been able to isolate anything. I'm thinking about query taxonomies [webmasterworld.com], which is another blurry area.

That's an interesting observation - that one of the ten first page results shows a very different "related:" profile than the other nine are showing. Again, it makes me think about the whole area of automated taxonomy assignments and "forcing" variety into the top ten.

If I had to guess right now, I'd think that related: sites is a result of some factor, rather than a cause. There could still be a correlation, and that might yield useful insight. I'll be doing am in-depth study of some top tens this week, and I'll fold the related: factor into that study to see if anything falls out.

[Hissingsid]

Hi Tedster,

The term I was looking at her is one with very different semantics each side of the Atlantic.

It is like <<widget service>>. The 9 sites provide "widget service" the other majors very strongly on widget and has service as a very minor element. In Webmaster tools the new "Keyword Significance" tool shows my site as 100% service and 66% (ish) widget. I wonder if the other site is 66% service and 100% widget.

Cheers

Sid