It’s kind of hard to prove that they actually do that in search results when the tilde isn’t there though – you can get a bit of a better idea that they’re doing it when it produces incorrect search results though.

At the moment, if you search Google for Nintendo Entertainment System Data you get electionstudies.org ranking in first and second place.

The reason is because Google has associated the query “Nintendo Entertainment System data” with the term “NES”, which it has then matched up with the American National Election Studies site, because of it also thinks that that is related to the term “NES”. It’s kind of true, both terms are related to “NES” – they’re just not related to each other which is why the search result doesn’t work.

You can see the same effect with the search Game Boy Advance Association which brings up other sites that are related to the acronym “GBA”, including the Girls Basketball Association and the Gay Business Association.

There’s a great example of the same thing when you search for British Home Stores society, bringing back completely unrelated results like the British Horse Society, the British Hypertension Society and the Belgian Hematological Society.

Of course, all of these terms are unlikely to be searched very often, I’m not suggesting that Google’s search results are broken because of it – it’s just interesting to see how Google connects different keywords from time to time. Basically, what’s happening (it would seem) is that Google is connecting together two terms that aren’t really related, usually acronyms, because they’re all connected to one main term. That sounds a bit complex when it isn’t really – so I’ll explain what I mean with a crudely drawn image. The term “Nintendo Entertainment System” is correctly connected to “NES”. “National Election Studies” is correctly connected also to the keyword “NES”, but it looks like Google has connected the terms “Nintendo Entertainment System” to “National Election Studies” incorrectly because of that.

While it’s obviously not worked on the results mentioned in this post, it’s probably produced better results overall – and it’s interesting to see that Bing and Yahoo don’t return results like this when you search for these terms. Maybe they’re not connecting terms together like Google is?

Flickr image from DizDau.

Is Google Connecting The Wrong Keywords? is a post from: Shark SEO. Have you played The Search Game?

It looks like Google is starting to roll out updates to the Google UK design now. The US redesign has been seen here and there in the wild, but this is the first time that I’ve seen it for Google.co.uk. The new design (for US results at least) makes local listings more important, as they fill up more of the screen – you can read more about Google’s new design here.

New Google Design For UK is a post from: Shark SEO. Have you played The Search Game?

In the comments he mentioned that this was mostly likely because of how merchant and affiliate links have worked in the past. Matt Cutts recently mentioned that Google will try to prevent affiliate links from passing link value, in most cases. Is Dave right? Does anchor text not flow through some 301s, and is that because Google’s trying to stop merchants from benefiting from affiliate links?

Flickr Image from Dan Coulter & Sprantaclaus

Testing Out 301 Redirects is a post from: Shark SEO. Have you played The Search Game?

This page on Google is actually indexable, because their robots.txt doesn’t appear to be set up quite right. Get that page indexed and, in theory, you should be able to get links directly from Google.

I’m not sure if these links are passing juice (and in truth, if the page isn’t linked to internally then it’s debatable how much juice can be passed) but I haven’t yet found any reason for why juice would be blocked. No nofollow, it’s crawlable and able to be indexed. Bizarre, eh?

Don’t expect this to last too long, by the way

Flickr image from Stevacek

Getting Links From Google is a post from: Shark SEO. Have you played The Search Game?

Speaking of which, Google have also started testing something slightly ridiculous. For a while now they’ve been testing out having a blank homepage, except for the logo and search box (which I like). When you move the mouse, the rest of the search buttons and links appear. Worried about confusing people though, they seem to have now added a little piece of text that says “This space intentionally left blank”.

Sesame Street Google Logos is a post from: Shark SEO. Have you played The Search Game?

This idea fits in exactly with what Chewy from Google recently mentioned, where he suggested that users refining a query generally indicates that they’re not happy with what they’ve seen, and Google sees that as a failure. As Richard Baxter has pointed out before, Google might also want to make sure that they get it right the first time because each query refinement will cost them in bandwidth. Getting the query right means that their cost per query goes down. When you scale that to something the size of Google, that could potentially be an enormous saving.

Lastly, the Vince update appears to have been rolled out across search terms that have a very high search volume (like “car insurance”, “loans” etc), but I’m going to guess that Vince mainly applies to terms that are searched for a lot, but that also have a huge percentage of those queries that then get refined. Another reason that Google might do that is because they have no idea what to show people when they search for something as generic as “loans”. I wouldn’t know what to return to people either – are they looking for a loan? Looking for advice on taking out a loan? Do they want to lend money? Do they want a definition on a loan? Statistics on loans? Google isn’t sure either, so a benefit of Vince is that they can let users dictate what they want to see. If vastly more people want to see Money Supermarket and Tesco Finance instead of Wikipedia’s loans page – why not display them instead?

The interesting thing about Vince is that it looks like the related searches show the refined queries – so if people search for “car tax” after they search for “car insurance” – that’ll be shown in the related search. Last year the related searches only displayed semantically related terms (like if you’d put those terms into the Adwords Keyword tool). Now it doesn’t. Check it out:

Awesome.

Flickr image from Marc Shackman

Vince & The Related Searches is a post from: Shark SEO. Have you played The Search Game?

In the iCrossing post I mentioned that the related searches now seem to be driven by user data, so when people search for “cars” and then search for “Honda”, “Honda” may become a related search for “cars”.

If we look at the expanded related searches for some of the terms that have been most affected by the Google UK glitch, quite often the related searches are fairly surprising.

“Tennis courts for rent” is one of the biggest affected search terms, it comes back with a load of prangfastical results for Singapore, Chicago, the Phillapines, Russia and more. But look at the related results when expanded:

It seems unlikely that people in the UK are going to be searching for results in Singapore, Canberra, Auckland, Melbourne, the Philippines and more. It looks like Google has globalised (if that’s even a word) all of the searches for “tennis courts for rent”, rather than used searches from just the UK.

“Family homes Kent” also displayed international results, with a lot of US results mixed in (thanks to @ismepete for finding that example). If you look at the related searches you get this:

Mostly the results focus on UK searches, but there’s a couple of US searches mixed in there that appear to be affecting the UK results somehow.

In Matt Cutts comments on his Google UK SERPs post the very clever Yoshimi mentioned that a search on “tennis shoes” resulted in the US tennis-warehouse.com ranking above relevant UK results. If you check the expanded related searches, you can see that the search tennis warehouse is listed in there:

And they all seem UK based, except for tennis warehouse. Why is it listed there? Do people actually search for it in the UK? If so, I guess that goes some way to further reinforcing the ol’ Vince theory but that should possibly be ironed out. And if people don’t search for it in the UK, why is it there?

This doesn’t sharkboggle work for all of the examples that we’ve looked at, because sometimes it’s impossible to tell if it’s a UK/US/international related search – but if you can find any other examples, please leave a comment (because I have no idea what’s going on).

(Flickr image from Furryscaly)

Is This Google’s UK Glitch? is a post from: Shark SEO. Have you played The Search Game?

Google wants relevant results. That’s it’s strength, and anything that can give Google better results is worth considering.

(flickr image from Danny Sullivan)

Sentiment Tracking

At the moment it looks like Google values backlinks massively (as long as they’re dofollow). It’s one of the things that makes argumentative linkbait work, where you create controversy that causes people to link to you. If they link to you, even though they don’t agree with you, that link can still in many cases help you to rank. Look at Habitat when they started spamming Twitter with the #iranelections hashtag. They got an absolute ton of criticism, and quite rightly. But at the same time, Habitat have benefited from getting a number of juice passing links, as shown by Yahoo. There’s thousands of blogs out there that talk about how much they hate Ryanair, for example. Thousands. Yet the strength of their links is, from where I’m sitting, letting them rank 2nd for the term “flights”. A while ago the UK furniture company MFI announced that they’d gone into administration and their website displayed a notice of this on the front page. They picked up a ton of links from UK news sites and for a short time even ranked on the first page for “furniture”.

Personally, I think it would improve Google’s results if they checked the sentiment of the linking page. If the page that was linking to Ryanair, Habitat or any other site also included a lot of negative phrases, and Google was able to determine that the page was referring to the link negatively, maybe that link shouldn’t pass quite as much value. Maybe it shouldn’t pass any value at all. In theory, this would prevent sites from benefiting (or at least, not benefiting as strongly) from getting negative press. I think it makes perfect sense, users aren’t likely to want to see unpopular brands appearing in the search results, just because they’re newsworthy.

Focus Less On Anchor Text

Anchor text is a huge clue to search engines what the topic of the linked page is about. A link that says “search engine optimisation” currently reinforces very, very strongly to Google that the linked page is highly relevant to “search engine optimisation” and is a good contender to rank for this term. The problem with that, though, is that most people don’t link to things like that. We do as SEOs, of course, but there’s a significant amount of people that don’t. There’s a surprising amount of people that link with just the URL as the anchor text, and there’s still an unbelievable amount of links out there that say “click here”, “more info” and “find out more”. In a lot of cases that I’ve seen, the anchor text doesn’t always relate to the page being linked to. That’s a huge problem for Google because all of those “click here” links that should pass on a decent benefit aren’t as strong as they could be. A site that should be on the first page, or ranking first, might not be. When you scale that idea to something the size of Google’s index, it becomes a problem. If Google wants to progress then it might help to focus less on anchor text and more on…

Topic of the Linking Page

Instead of having such a strong reliance on the anchor text, Google could instead focus more on the theme of the linking page. A page that’s all about cars, when linking to another page that’s all about cars, should (technically) pass more value. If a page that’s all about cars is linking to a page that’s all about badgers, that link shouldn’t pass as much value. I’ll be honest in that I suspect Google may already be doing this, but I have absolutely no idea to what extent they’re doing so. While it’s not really possible for me to prove this, I think valuing links more if they’re from related pages will lead to better search results. If I link out to SEO ROI, SlightlyShadySEO and UXBooth (all really useful sites by the way) then those 3 links should pass the full level of juice. If I link out to sites that have nothing to do with this site or post, like The Badger Preservation Society then perhaps that link shouldn’t pass quite as much value. In turn this could lead to more relevant results – and it has the added bonus of potentially negating quite a large chunk of spam.

Anticipating changes that Google might make can give you a bit of an advantage when it comes to getting your site ranking both now and in the future. Even if Google doesn’t start to change by tracking sentiment or focusing less on anchor text, you’ll still benefit massively from being talked about positively, and from getting links from relevant sites and pages.

(flickr image at top from Qualien)

How Can Google Improve It’s Search Results? is a post from: Shark SEO. Have you played The Search Game?