Q-Phrase on Google's Orion: "Kiss Snippets Goodbye?"
Yesterday I said that the reason I've been so obsessed with this Orion purchase is that Google "may be deliberately integrating 4th-place (in usage) Ask's functionality into how it enables users to navigate SERPs."

Based on an email I got from Q-Phrase's President, Andy Miller and its Chief Software Architect, Danny Espinoza I've begun to rethink that position.

I think there's an even larger issue here: the usefulness of the SERPs themselves.

This is from Q-Phrase's email to me:

"Problem: the sum of the content on one SERP does not have enough data, in and of itself, to reflect the concepts mentioned on that page, and thus the true relevance of that page is unknown."

Both Q-Phrase and Ask have approached this issue in different ways - Ask with Zoom and Q-Phrase with ConceptQ.

Google has not - to my observation (maybe the "did you mean" function?) - addressed this issue in a significant way in its SERPs.

Orion SOUNDS LIKE a method for more meaningful SERPs organization. Whether it is or not is pretty much irrelevant, as Sullivan pointed out in comments yesterday: "this was probably more about hiring the guy and having to sort out his patent which may not even get used."

So, now, I'm attempting to reframe this discussion as a "problems with SERPs" issue. Because there ARE problems.

Jupiter Research and iProspect recently measured searcher behavior (pdf). Though the study did not seek to measure whether or not these users found relevant information, or investigate HOW the SERPs themselves could be influencing searcher behavior, I find this fact telling:

"41% of search engine users who continue their search when not finding what they seek, report changing their search term and/or search engine if they do not find what they are looking for in the first page of search results. A full 88% do so if they do not find what they seek in the first three pages."

Changing search terms (not to mention changing search engines) is VERY significant user behavior. I think the very need to rethink and reexecute search queries (a search refinement behavior that search engines are TRAINING into searchers... and will have to train OUT of them later) indicates a problem with the SERPs.

Q-Phrase's solution to getting more out of SERPs plays out this way:

1) Ask ConceptQ Pro to search for "american revolution" on Google.

2) The first 100 SERPs returned by Google are crawled automatically by ConceptQ Pro and the text on every page is data mined out of the HTML (note that this ignores all meta data on the page, and of course ignores content which is no longer on the net, since it won't find it).

3) The most statistically significant topics are then computed from the entire data set, extracted, filtered and collated by word count and displayed in the window.

In my post yesterday I asked for help: "Sullivan, Garcia, (anyone!), based on your understanding of the Orion algo, am I grasping at straws here with my conceptual zoom reading of the Orion algo?"

Q-Phase replied to this:

You've got a handle on this for sure, but I think Google's motives go beyond a simple "zoom" feature.

Once the conceptual topics have been computed, it is possible to calculate the relevancy of every sentence in a SERP page to the original search. And once that is completed, you can kiss "snippets" as you know them goodbye.

INTERESTING.

SERPs without Snippets! For organic search marketers that'd mean making sure clients fit into the larger story surrounding a given conceptual space relating to client business goals (which, incidentally, is how you should be thinking anyways - that's the difference between optimizing for presence and optimizing for rankings).

NOTE: The likelihood of snippets being entirely eradicated is unlikely - especially any time soon. SERPs experimentation and evolution is an inevitability though, and search marketing preparation starts with thinking about and discussing likely futures.

I'll leave you with an illustrative anecdote from Q-Phrase:

"...search "IPTV" on Google and the first result is for "Iowa Public Television". According to PageRank this makes sense. Now, search for "IPTV" using ConceptQ (which will read the first 100 SERPs returned from Google): the conceptual topics discovered on all those pages include: internet protocol, iptv technology, video on demand, etc.

"If you open the "Sources" window in ConceptQ, you will see a list of those SERPs re-ranked by their relevance to the conceptual topics. You'll find that Iowa Public Television has sunk to the bottom, and a blog on "IPTV trends, deals, software" has shot to the top."

And that, in a nutshell, is what Concept Rank could look like, and one possible (and strong) vision for the future of the SERPs.

For the trigger happy amongst my readership - this is ALL SPECULATION. Though if you ask me it's certainly well reasoned and right ;)

Questions on the future of SERPs? 919-433-3139.

And now on to Yahoo Maps and the Google Calendar. It's going to be one hell of a day :)

(PS: Q-Phrase + PreFound would make an interesting little couple.)
Garrett French - MarketSmart Interactive | permalink | +marktd |