Related pages on google. What's the logic behind?

[skuba]

Does anybody know what logic is behind the list of related searches on google?

If you search for example related:widgets.com, you will find results like:
otherwidgets.com
relatedwidgets.com
etc..

But how do they build that list? How could a site that is not listed there, get listed?

Thank you so much

[skuba]

Does anybody know?

[Crush]

sites that are on the same page as your site or you link to them. They got all my sites in the list. Some people were saying it was the toolbar but that is BS.

[g1smd]

When looking at the "related" list for your site, you'll find that they are some of the sites that some other site that links to you also links to...

[larryhatch]

I just checked my "related sites". These are mostly:

a) Sites that I link to
b) Sites that link to mine.
c) Sites that rank near mine in the SERPs.

As for the c) group, there are a few I would rather not see there.
Offhand, it looks like a service provided to browsers to widen their options.
I doubt much of their resources are devoted to "related sites",
nothing like G's massive SERPs positioning algorithms and hardware. -Larry

[skuba]

I am not sure that a) and b) is a rule.
If you search for a famous store, like related:gap.com will see that the first ones are oldnavy and bananarepublic, but that's ok, they are the same company. But all other ones are competitors, they are not linking to each other.

It could be based on stats from the google toolbar. This is how it works in Alexa.

[g1smd]

I looked at the "related" list for one site, and by chance I recognised the list as simply being an almost exact copy of the links page on some other site that linked to the site that I was looking at.

[mjtaylor]

When I go to related sites for some of the sites I manage or own, I see lots of sites that don't link to me or vice versa, but what they seem to have in common is that they are in the same regional directory of DMOZ.

Which does, indeed, make them related geographically, if not topically.

FWIW.

Cheers, MJ

[g1smd]

.... so, the "related" list for your site, is simply a copy of all of the links on a single page of an authority site, where that page on the authority site also includes a link to your site ....

What else do you deduce from that, then?

[mjtaylor]

so, the "related" list for your site, is simply a copy of all of the links on a single page of an authority site, where that page on the authority site also includes a link to your site ....
What else do you deduce from that, then?

Well, as I said, they are related geographically ... some of those "related" sites are in Big Pine Key, Fl ... others are indeed related because they link to me and my site to them ...

I guess I deduce that Google considers DMOZ to be an authority site ... and that Google considers geographical location to be relative.

Makes sense to me ... it also makes sense if related sites included other sites with the same topic as mine, as it does ...

What do *you* deduce?

Cheers, MJ

[g1smd]

I deduce that by careful searching using this feature on various website URLs you can uncover several other sites that Google considers to be authority sites, and it is those that it might be a little bit more valuable to ask for an incoming link.

[mjtaylor]

I deduce that by careful searching using this feature on various website URLs you can uncover several other sites that Google considers to be authority sites, and it is those that it might be a little bit more valuable to ask for an incoming link.

Very interesting idea. Thanks for suggesting that ... I will look at the similar pages now with a fresh point of view!

Cheers, MJ

[skuba]

I don't see any of these relations.
When I search for related:mysite.com , I can't see any connection between all the results. And don't see all of us being listed at Dmoz or any other site.
I still suspect it could be the toolbar.

[g1smd]

>> I will look at the similar pages now with a fresh point of view! <<

The authority site isn't always in the list of similar pages, it is the site that also links to most or all of the sites that are found in the similar pages list.

[mjtaylor]

The authority site isn't always in the list of similar pages, it is the site that also links to most or all of the sites that are found in the similar pages list.

Another good insight, thanks ... Lots of digging there, unless you know a quicker way.

Gee, you are helpful. That's what I love about forums like this, and people like you ... it keeps the spirit of sharing information alive. That's what fascinated me the most about the Internet when I first logged on in '95 ... sheesh! A decade ago!

Thanks again!

MJ

[Tami]

Hello All this is my first post here!

I have been having problems. Lost my SERPS ranking end of 12-2004. Have made changes and submitted a site map to google so I am indexed but not in SERPS.

When I do a related:mysite.com I get "did not match any documents." What does this mean?

Tami

[hutcheson]

The site was recognized as being truly in the heart of a "bad neighborhood". In other words, few "authority" sites linked to the site, and sites that DID link to the site, tended strongly to list to other sites that Google had already recognized as being artificially promoted.

[Tami]

Thanks hutcheson.

I just discovered that if I do a related:mysite.com on Alexa I get plenty of listings. But in google nothing.

Tami

[Tami]

I just wanted to add that I have never knowingly linked to any bad neighborhoods nor did I ask to be included on any bad neighborhood sites. But I do know that an awful lot of spam sites, link farms have links to my site.

What do you do in this instance, do you actually email all the "bad" sites that have links to you and ask them to remove your link? Would this help?

Tami

[mjtaylor]

To the best of my knowledge, a site is not penalized for the quality of inbound links. Since a webmaster can't control whether spam sites link to them, it would not be logical, IMO. And if that *were* the case, it would be possible for a competitor to get a site penalized.

I had a site that dropped from Google, and I emailed G, got the standard reply about quality and practices outside their guidelines. I checked each and every outgoing link and found that, indeed, that a plump handful of partner sites had degraded to "bad neighborhoods." Once I removed my questionable outbound links, Google reindexed my site. Those "bad neighborhoods" do still link to us but the site remains indexed and well ranked.

I am now vigilant about checking link partners on all sites regularly, and I am even more careful than ever about which sites I link to.

Cheers, MJ

[g1smd]

If you link to a bad neighbourhood, as I have done several times as a test, what I see is that the cache date for the page, that was previously updated dailiy, freezes for several days and then reverts to a date from 10 to 20 days ago, and stays there.

When you remove the link, nothing happens for several days, and then the cache date updates to one from several days ago, then the next day to one from yesterday, and then resumes the daily update. I have tested this many times during the last year, and it seems to be repeatable.

[hmatisse]

Google has patented a couple of methods for determining "related" pages that are based on links. Whether or not they are using these methods is, of course, a different question.

Basically, in both methods they are first considering documents that link to document in question, and then look at the set of documents to which they link, and then also consider a third set of documents to which the second set links.

It is covered in "Techniques for finding related hyperlinked documents using link-based analysis".

[glengara]

When G first started it was called something else, but always purported to show "similar" pages.
As it's always been based on linkage, I think we can determine that ab initio Google assumed links would be topical ;-)

I always check them as they give a good crosscut of a sites' linkage...

<added>

Here we go, Google Scout....

[internetnews.com...]

[RockyB]

Well, it seems rather random. My site is apparently realted to W3C :/

[seppyr]

We set from Domain1 (which is not authorative) two footer links to Domain2 and Domain3.
Domain2 and Domain3 doesn't have any common. They are not releated in any way, e.g. different page topic, different Ips, different country ect. and no one set links to both domains in any way on there pages.
After a while, the best match for related:domain2 was domain3. So it seems that the link neighbourhood is a strong criteria for beeing related.