Update: Google has confirmed that this is a ranking/UI change. See statement below.
Is it a test? Is it a bug? Is it a permanent change to Google’s search results? No one knows, and Google hasn’t answered our questions about it yet. What’s “it”? As Malcolm Coles describes in a blog post today, Google is allowing a single domain — from a well-known brand — to dominate the first page of search results on some brand-related searches.
Consider this screenshot below of a search for apple ipod:
(click for larger version)
The first seven of 10 organic/natural results come from apple.com. There are some shopping and news results mixed in, and other sites have the 8-10 spots. But to rank a single domain in so many results on a single page is unprecedented. The same thing appears to be happening on other brand-related searches like espn nfl and even search engine land articles. It doesn’t happen on other brand searches like “ford trucks,” “hanes t-shirts,” or “microsoft software.”
Ironically, just a couple weeks ago, Google’s Peter Norvig specifically talked about the need for variety in search results:
We haven’t figured out any way to get around majority rules, so we want to show the most popular result first, but then after that, for the second one, you don’t want something that’s almost the same as the first. You prefer some diversity, so there’s where minority views start coming in.
So, what’s going on here?
We’ve asked Google for a statement about these brand-dominated search results, but they haven’t replied at this point.
In the meantime, Bill Slawski has written a post today that refers to a Google patent about “How a Search Engine Might Assume a Query Implies a Site Search.” Bill explains it thusly:
The process in that patent may mean that if Google recognizes when a search query involves a particular entity, and if the entity can be associated with a specific web site, it might show multiple results for that site.
That seems to describe exactly what’s happening in these search results. If so, Google might argue that variety in search results isn’t as necessary when users perform brand-related searches.
We’ll update this post if/when we learn more from Google.
Postscript: A Google spokesperson has confirmed that the search results discussed above are part of a ranking/user interface change related to domain-based intent:
“We periodically reassess our ranking and UI choices, and today we made a change to allow a larger number of pages from the same site to appear for a given query. This happens for searches that indicate a strong user interest in a particular domain.”
Postscript #2 by Barry Schwartz: Google not only confirmed it but wrote about this in more detail.
Related Topics: Google: SEO | Google: Web Search
Matt – Here’s a variation on the same theme. If you google “wordpress seattle” or “wordpress websites seattle”, you’ll see results are dominated by subdomains of wordpress.com. The only reason these sites are in the SERPs – for these and many other keywords containing “wordpress” – is that they are on the wordpress.com domain. In other words, even more than in the apple.com example you cite, these SERPs are filled with less than relevant results. They are not just biased, they are polluted.
Very interesting. It is putting more of the onus for delivering relevance on the searcher. I’m wondering if this might not be a good thing … but it really feels experimental.
A search for “apple ipod” does, indeed, feature Apple.com listings.
A search for “apple ipod los angeles”, however, begins with a primary link to Apple’s retail store near downtown Los Angeles and exposes two secondary links to Apple.com’s store list and a link to the next-closest Apple store in Century City. The rest of the results are what you have become used to seeing … a smorgasbord.
A search for “apple ipod sale in los angeles” virtually ignores the location preference, returning a wide variety of links to sites where one could purchase an iPod, but no physical store locations in the Los Angeles area.
Using extremely rudimentary logic, the broader the query, the more brand-specific it may be, and it seems like certain keywords (“sale”) have a lot more weight than they used to, overruling location-sensitive keywords (“los angeles”) where previously location seemed to be an overriding metric.
The more specific the search (as usual), the more relevant the results. Perhaps Google is just getting tired of trying to figure out what people are thinking, and are now expecting the searcher to take a more active role in specifying their target?
Mark, your post certainly begs the question: What kinds of results were you seeking with such an open-ended query for which you received open-ended results? If that was a “real” person making the query, were they seeking a list of websites that use WordPress and are edited by people who live in Seattle? Or what? How could that search have been useful with *any* results?
Perhaps if you had been more specific, as normal people who are actually looking for something do, you would have been satisfied, instead of searching on an ill-defined query and receiving ill-defined results? In order to evaluate the effectiveness of a search service, you need to have an actual goal, and then tailor your search to accomplish that goal. Tossing out random phrases will return random results. It’s always been that way … even back to Archie and Gopher.
great post matt. i don’t think this observation is a result of a recent shift. about 1 year ago (9/09) myself and another well known seo consultant observed and documented the exact same setup on a query for the US brand “petsmart”. the same result set can be observed today which is identical to the apple example above. i agree with all your points, however it is worth noting that this change has been in effect for at least a year.
google would never admit it, but i’ve experienced them playing favors to those who pay for ads in their organic results.. at least it certainly appears that way in very competitive industries.
I think this is a similar scenario.
I have just noticed the effect here in New Zealand, where one of our country’s most popular visited sites – Trademe (an ebay auction type site -which has a least 1million monthly search queries, and 20, 000 – 80, 000 members online anytime).
It now displays six references of categories for the search term Trademe NZ. This has pushed my Trademe tips blog a bit further down, but still on page 1 – but a noticeable drop in visitors.
It doesn’t look as bad here in the UK. Just the first 4 organic search positions are from Apple. Will keep watching though.
No need to be a big brand. I try with a small world-wide brand “Creaform 3d scanner” and i got 8 organic search positions for Creaform. Interessting post :-)
Wow. I’m not a fan. If I wanted to see that many results for a particular domain, I would go to that domain as a user and search from there.
Thanks for the news. Interesting stuff.
Premium member since 01/2009
“No need to be a big brand.” — yes, that’s correct. Google is matching queries to domains. That’s why it happens on the Search Engine Land example I mentioned above. It also happens for “seomoz” searches (at least some). And neither SEL nor SEOmoz are huge brands in the way that Apple, ESPN, etc., are. :-) It’s domain-related, not brand related.
I was going to post the same thing Matt. It looks like Google is placing more emphasis on the URL and the keywords inside there. All of the Apple domains above include atleast 1 of the two words from the search, “apple ipod” (if not both).
aavallon said..
“It looks like Google is placing more emphasis on the URL and the keywords inside there.”
i think you hit the nail on the head. seems to correlate with the results i’ve seen as well.
either that or what i’ve said above.. which probably isn’t the case.
Premium member since 09/2009
Love it – this type of news writing is what makes SEL a go-to site for everything on search engines. Great work Matt!
Great post. I didnt think one domain could dominate the search results like that! ridiculous