Will Google's Caffeine Update Really Change Search Results?
A few weeks ago, Google announced the beta launch of Caffeine, the company's next-generation search infrastructure. At that time, Google said that most of the changes in this update were under the hood and that users wouldn't notice a difference in search results. At its core, Caffeine is basically a major overhaul of the Google File System. There have been some discussions about whether this update will bring any other major changes to page rankings or the importance of certain categories in the search results. Summit Media, a UK-based digital marketing agency, compared search results for 9,000 keywords (PDF) in Caffeine and Google's default ('vanilla') search and, interestingly, didn't find any major differences between the two.
While the first version of the Google File System obviously scaled quite well, it wasn't built to support the blended search results Google displays today (images, videos, news, etc.). Even Google's own Matt Cutts argues that this update won't have any major effect on Page-Rank and doesn't constitute a change in Google's search philosophy and algorithm.
There has been quite some discussion about whether this actually turns out to be true, though. Judging from this new study by Summit Media, there really haven't been too many changes to the rankings in Caffeine, though the Summit Media team did notice some interesting differences.
There was no major difference between the kinds of categories of sites Caffeine prefers compared to Google's 'vanilla' search. Caffeine, for example, doesn't give any more weight to price comparison sites than Google's current search infrastructure. One slight difference that the Summit Media study noticed was that slightly more co.uk domain names now appear in searches that originate from the US. The difference there, though, was only minor.
Difference: News Sites
One area where Summit Media noticed a difference, however, was sites that fall into the 'Information,' and 'News' categories. Here, at least for generic search terms, Caffeine shows a bias towards more timely information and seems to punish sites that don't update often and mostly consist of archived material.
Does it Help to Have a Keyword in Your Domain Name?
Summit Media also tested another popular SEO theory: does it help to have your keyword in your domain or page URL? Summit didn't find a difference between the Caffeine and 'vanilla' search results and maybe more importantly, having a keyword in your domain name doesn't seem to make a real difference when it comes to search rankings. According to Summit, only about 6% of all search results on the first three search results pages actually contain the specific keyword in the URL. That's a very small number, and, as the report concludes, it is also "a demonstration that SEO needs to be based on actual research - rather than perceptions over browsing a few pages of results."
Facebook
Comments
Subscribe to comments for this post OR Subscribe to comments for all ReadWriteWeb posts
Just the other day I was talking with some other SEO folks about keywords in domain names. We were wondering if having a dash or not helps with multi-word key phrases. For example: drycleaner.com or dry-cleaner.com. Can Google pull out the two words when it sees drycleaner?
The research noted here seems to say that I wasted an hour of my life and I won't accept that.
But, my experience has been that the first ten search results in the Google's Caffeine search are generally quite different from the ones shown in regular existing Google Search. And, it is the first ten search results that matter the most. Not many persons bother to check the remaining search results.
Checking the first ten results yields no difference overall through the study.
The full study shows that there is more change on more generic terms but it's still not huge. I think a lot of SEOs have been checking very generic terms for a limited number of results and not considering how limited that dataset is - it completely ignores the long tail.
I think everyone is missing the shift towards brands in the current and Caffeine search results, google seems to want to push everyone down the page by using "local" maps for as many search results as it can, for example "hotels placename" I want a price comparison site not a directory enquiry type of listing of the names of hotels on a map, that doesn't tell me hotel star ratings or room prices. Google is losing its way, we can all see google news and how can the latest facebook news story appear above the facebook site when you search for it.
I wish they will hurry up and roll it out already.
Cute posting. Very detailed and informative. Good work. Thanks for posting.
Any idea when they'll be rolling this out? While nothing drastic, I've definitely noticed some marginal changes in the listings. The emphasis on newer content makes sense, as too many first page results are often loaded with pages that haven't been touched in nearly 10 years.
I have noticed some significant differences in the Caffeine SERPs, especially for keywords that aren't as competitive or are niche market keywords. The big boys won't notice sweeping changes, but many SMBs will see quite a bit of flux.
And as for the keyword being in the url ... from direct experience, it does have an impact. The exact same site on a www.nokeyworddomain.com and a www.keyworddomain.com will rank very differently. Put an domain alias on a site and see for yourself. In my experience it can make a 20-30 position difference.
Re: the 'dash' comment ... Matt Cutts has said that the dash does act as a separator, but that Google can usually differentiate the words. If you don't want to take any chances, the dash is useful; however, the prevailing opinion seems to be that your mileage will vary depending on how well the big G has already divided your terms on their own.
Just because the current set of SERPs does't have a high percentage of keyword domains doesn't necessarily mean that it's not important. That idea is a case of a poorly thought out conclusion based on anecdotal evidence, not good analysis of empirical data.
@Horse barns, you say 'a case of a poorly thought out conclusion based on anecdotal evidence, not good analysis of empirical data'.
Your comment contains an anecdote about keyword domains. The study does not, it contains .... data.
I think there's a significant problem with your argument.
Does anyone know when this update will take effect? Many of my sites will improve according to the sandbox, and I am getting eager
We have been tracking our results and comparing the regular google and caffeine pages. and, we do see some good changes in our restaurant industry. So just like everyone else, we're excited to know when will this new update go into effect?