Similar "related:" much improved

[Hissingsid]

Has anyone else noticed that the similar link, under the listing in SERPS, is returning a much more accurate list of sites/pages that are actually related?

In the past one of my sites had "similars" that fairly accurately reflected what the site is about but had one or two oddballs. Now it lists all of my main competitors with no erroneous ones. It is spot on.

Some competitors (the ones that are not very cleaver in their choice of sites that they purchased links from) used to have quite a few outliers in their similar results but now their similar profile is very accurate from a market watcher perspective.

I've checked other sites I have and they all have a much better/almost perfect similar list.

Any comments?

Has this already been discussed and I missed it?

Cheers

Sid

[tedster]

Totally agree - it's been a long time coming, but related sites are now much more related in topic, rather than some spastic collection of loosely interlinked sites in an odd cluster that does not give a user any value.

[julinho]

I wonder:
1) how Google improved this part of the algo (maybe a deeper analysis of searchers' and surfers' behaviors, rather than a simple link graph analysis);
2) how this improvement may affect rankings (e.g., the ranking power passed by a link could be proportional to how closely related the linked pages are).

[Seb7]

Has Google also gone and reduced the amount of links shown on the 'link:' page?

[physics]

Wow, I may actually start using that feature instead of just laughing about it :)

[remove]

I have to say, I'm impressed by this improvement.

While they're at it, they ought to fix "link:" next, because that's long returned pathetically incomplete results.

[tedster]

Google's not going to "fix" the link: operator. They give much fuller link data in your WebmasterTools, but that's for the verified site owner - and that's intentional. They are just not going to surface large amounts of link data for people to research any website at all.

[Hissingsid]

tedster They give much fuller link data in your WebmasterTools, but that's for the verified site owner - and that's intentional. They are just not going to surface large amounts of link data for people to research any website at all.

Even what is shown in Webmaster tools is only 30% of the links in my case.

What would be interesting would be if we could work out on what basis the selection for link: and in Webmaster tools is selected. Are those selections indicative of something?

Cheers

Sid

[aristotle]

In my Webmaster Tools, the lists of external links used to show about 60% of them, but the number has been steadily dropping in recent months and is now down to about 20%. There doesn't appear to be any logic as to which links are selected to be shown, but it isn't a random sample because typically half of the links to a given page could be from a blog roll.

I suspect that Google is intentionally restricting this information even to the site owners themselves, to prevent them from using it to evaluate the success of their various artificial link-building efforts.

[Hissingsid]

I've always analysed my top 10 competition backlinks and on page profile. I'm now wondering if I should be doing the same for my top 10 related sites.

If the #1 result deserves its place as I think it is in my case ;-) then IMO the related: search actually produces a better top 10 than are returned by standard SERPS. It could be a good way of producing a new search engine technique. First find the best site/page for a search then find others that are similar. Someone from Google recently let slip that they would like to move towards providing one correct answer to a search. This would be better, one correct answer and 30 similar ones, take your pick.

Cheers

Sid