Google Expands Snippets & Related Searches Word Relationships

Google announced that they are rolling out a new technology to better understand word relationships and extend their snippets on longer search queries.

Starting today, we're deploying a new technology that can better understand associations and concepts related to your search, and one of its first applications lets us offer you even more useful related searches (the terms found at the bottom, and sometimes at the top, of the search results page).

Note that they claimed that this is "one of its first applications." If they can improve relevancy with integrating this technology directly into the core search algorithms then it will lower the importance of on page optimization (since you only need to be close rather than use the specific words that were searched for). Such a change would likely decrease the traffic to low authority sites held up largely by strong site structure and on page SEO, while increasing the amount of traffic going to high authority sites and well branded sites that are poorly structured from an SEO perspective.

I am not sure if this sort of algorithm change would favor shorter or longer content pages. In most cases I would guess longer pages if they were kept on theme, and broken up to relevant chunks. The expanded snippets on longer search queries show a lot more information directly in Google's search results, which helps thicker pages show off their offering more than thinner pages, but cedes more control of the information over to Google as they can show close to 250 characters in the search results.

If the technology was applied to anchor text it might also limit the value of anchor text manipulation by boosting up the value of related phrases (if Google knows that the word Salesforce is relevant to CRM then they might count that anchor text more).

Greg Sterling noted that this change came from the Orion technology that was purchased by Google from Ori Allon in 2006. He also interviewed them:

I spoke yesterday to Google and Ori Allon. To the extent that I understood his discussion of the way Orion’s technology had been applied to refinements here’s what’s going on at a high level: pages are being scanned in “real-time” by Google after a query is entered. Conceptually and contextually related sites/pages are then identified and expressed in the form of the improved refinements. This is not solely keyword based but derived from an “understanding” of content and context.

It is hard to speculate if/when this technology will move from sideshow to becoming a big deal. The current usage is fairly trivial, but it could get much more well ingrained into many parts of the relevancy algorithms.

As search engines get more sophisticated with how they show word relationships (on branded and non-branded search queries) that is one more thing that can be optimized, though likely one that will require a holistic marketing strategy to optimize, because you will need to create a lot of co-citation (or some other signal of relevancy) across many pages on the web.

A couple years ago Lord Maurice Saatchi described their brand strategy as being built off of One Word Equity.

In this new business model, companies seek to build one word equity - to define the one characteristic they most want instantly associated with their brand around the world, and then own it. That is one-word equity.

It is the modern equivalent of the best location in the high street, except the location is in the mind.

Published: March 25, 2009 by Aaron Wall in google


March 25, 2009 - 6:54pm

There's something else going on though not mentioned in any of the articles about this last update.

I've noticed sites changing rankings multiple times a day since this update occurred. Sites going from page 3 to page 2 to page 1, and then back again in a loop. Almost like they're being rotated on some kind of schedule... But no one seems to be talking about it.

March 25, 2009 - 9:35pm

Ehinchman's on to something.

I've just had a quick check on a few rankings and they do appear to be yoyoing around pages 1, 2 and 3.

Hmmm... the plot thickens.

March 26, 2009 - 6:30am

No plot imo . There was a deep crawl @ 23-24th march and the temp yoyoing could well be related to that.

Aaron, do you think Google has started algo shift towards semantic search ?

March 26, 2009 - 12:12pm

This update has nothing to do with the ranking of sites yet.
The new related search elgo is based on user data aswell.

For example if you are searching for Xvideos then Videos then funny videos then Music Videos then Vids google will keep a record of all searches you did and starts showing you related search data based on the theme you were previously searching.

ehinchman i did not noticed any thing changing the rankings of the sites i monitor. Might be your site is too young to get any position in index. Are you checking many sites or this is the case with only one of your site?

March 26, 2009 - 2:23pm

Aaron, Thanks for the post!

This addition by Google is going to help the users to find the exact match what they were looking for. It’ll be beneficial for SEO companies like mine,( as it adds value to help the clients in getting more business.

Anyways thanks once again for highlighting various aspects of this update.

March 26, 2009 - 3:54pm

Well. It appears that the yo-yo effect may have died down.

@ soul-healer ~ I've been monitoring multiple sites, and yes, I noticed no change at all with some, and did notice the yo-yo effect with others. Friends of mine noticed the same thing and we've been trying to figure it out.

@ Aaron ~ Your blog doesn't notify me of replies to comments :/

March 26, 2009 - 5:19pm

@ehinchman when ever i see yo-yo effect it is with the new sites :) Lets see how it work out for you. keep me updated though

April 3, 2009 - 1:33am

Have you seen Amazon's keyword data? They have an inverse-document-frequency type thing for their books, where they show you keywords likely to be associated with the book's content, based on their rarity elsewhere in Amazon's index. (Nice data source, btw). Perhaps Google are doing the same, but combining it with co-citation?

April 3, 2009 - 5:57am

I haven't read a lot about the Orion technology...but it would seem that IDF and co-citation are too obvious signs of synergy. Also user search data (and query refinement) is a great data source.

Add new comment

(If you're a human, don't change the following field)
Your first name.
(If you're a human, don't change the following field)
Your first name.
(If you're a human, don't change the following field)
Your first name.