If you can't beat 'em... join 'em
Well it’s January and to start things off; Yahoo wants your SEO job as well. You may remember last year (Feb 08) when we looked at Microsoft’s SEO Program and at the time it seemed a strange conflict of interest. There are certainly the same set of questions surrounding any search engines that offer SEO (at least Google sold off Performix) and this patent (should it become a service) does that once more.
Now it’s unclear if this is a stand alone service or tied to Yahoo Stores and other hosted services. Ultimately there is a broad enough stroke here for a few implementations. While much of the systems SEO methods are fairly standard stuff, they do have one huge upside… but we’ll get to that shortly… Here’s the patent;
Automated system to improve search engine optimization on web pages - Filed; June 28 2007 and Assigned; Jan 01 2009 - Authors; Ting; Chyr-Chong (Joseph); (San Jose, CA) ; Hyder; Adam; (Los Altos, CA)
Optimizing the page
In simplest terms the system would analyze and target primary keywords/phrases, related phrases (by popularity) and assess the web page to create greater prominence of the terms through;
- Page title; optimizing title tags
- a hyperlink; editing link texts
- a meta-tag (description); apparently still holds ranking value somewhere
- a meta keyword; did I miss a memo?
- a heading; you know… H1-5 tags.
- a heading tag;
- alt tag; another sensible item…
Now they actually do give some scoring hints to valuate the various elements;
“In one embodiment each of the editable information may be ranked or given a point value that is reflective of its significance for SEO.
For example, the point value may be from 0 (least relevant) to 100 (most relevant). The title may have a value of 100 because it is important for search engines and for SEO, while the meta keyword may have a value of 10 because it may be less important to search engines.”
Essentially the primary terms are more targeted to the higher value areas while the secondary terms on lesser locales.
All in all some sensible concepts and nothing one would really get too worked up about. They also discuss the option of a page-by-page SEO program or entire sites. While this all seems fairly logical, it does give some insight into what elements they feel are important (across all the major engines). I say that because they talk about optimizing for Google and MSN not merely their own offerings.
Query log mining
Where it does get interesting is how they are getting the data to decide which terms are important. You guessed it I am sure; search query logs. Yahoo can use it’s existing search query logs to analyze which terms (relating to the page) are most popular. Creating a list of primary and secondary terms which is then cross referenced with the page and elements listed above.
“The act of determining a plurality of popular units from a search query log comprises generating a unit frequency list, wherein the unit frequency list includes the search frequency for the units; wherein the units from the search query log comprise phrases from search queries stored in the search query log.” (‘units are KWs by the way)
Now that is a reasonably considerable advantage if you ask me. Now, any SEO worth their salt knows that some terms convert better, thus mere query analysis won’t identify anything more than popular terms, but I’d still LOVE to get my hands on data like that when working my programs.
And it doesn’t limit the system (ultimately) to even the query logs as noted with;
“Other data may include other searching or web browsing tendencies identified by the search engine that may reflect the popularity of a concept or query.”
This means Google trends could be used, Yahoo’s social web and pretty much any other accessible data.
Phrase research and concept identification
Of interest as well are the sections that cover the term research as they mention phrase relationships and discerning concepts and topics. This analysis is done on the web page as well as the search query logs.
“A unit is a phrase of common words that identify a single concept. As another example, the search query "Chicago art museums" may include two units, e.g. "Chicago" and "art museums." The "Chicago" unit is a single word, and "art museums" is a two-word unit. Units identify common groups of keywords to maximize the efficiency and relevance of search results.”
This approach to aggregating potential terms is somewhat reminiscent of the Google work on Phrase Based Indexing and Retrieval and other probabilistic learning models. Past Yahoo filings cited in the phrase/concept identification are;
System and methods for generating concept units from search queries
Systems and methods for search processing using superunits
Systems and methods for managing and using multiple concept networks for assisted search processing
Search processing with automatic categorization of queries
This is a dafacto keyword research system that combined with their query logs can go a long way to some effective SEO – as we know that KW research is a VERY important part of any SEO program. Some other concepts include;
- Breaking terms into related parts; ‘New York City law enforcement’ is broken into ‘New York City’ and ‘law enforcement’ to analyze popularity for the parts and the whole.
- Search term popularity over time – by looking at historical query data for temporal anomalies (search trends).
For more on targeting do review the patents posted above....
Putting it together
As I mentioned before they seem to be talking about hosted pages and applications. There is essentially an interface that pulls the web page, the research data and applies the changes to the page all from the same integrated system.
It would seem the system suggests potential on page improvements and then the end user agrees and the changes are made to the page. This is what really has me thinking this is for one of their hosted apps such as Yahoo Stores. Imagine having a CMS that had such power under the hood? Sure, the on-page adjustments aren’t cutting edge, but the term research abilities do have some advantages.
If we go beyond musing about the novelty of the approach and the power of query logs – it remains a bit of a problem (to me) when search engines essentially provide services to potentially self-influence the so-called ‘organic’ listings. It seems to me that not so long ago we were nothing more than SERP manipulators and now the engines are taking the, ‘if you can’t beat ‘em, join ‘em’ approach.
Are you concerned? Threatened? Considering there aren’t any link profile development aspects covered, will it really be an advantage? Would you get an account just to get at the query data?
Lot’s to consider here… I am some how sure we’ll get back to this one again soon…
Happy New year ya’ll!!
Here are some random thoughts on some of the points in the patent filing, (emphasis mine);
I did find this part a little ironic;
“However, the amount of information available online is so large that it may be difficult to find relevant information. …..”
“…search engines are responsible for directing a portion of web traffic to various pages, it is important for pages to appear within the search results for related search queries. ….. it is desirable for a web page operator to have his/her web page appear near the top of the list of search results.”
It’s kinda funny having a search engine offer SEO tools/services because, “we can’t deliver relevant results” – did they just legitimize the SEO community or what?
This part almost sounded like something I could plagiarize for my SEO services
“Search Engine Optimization (SEO) refers to actions by a web site to improve its presence and positioning for relevant search queries. The most popular search engines, including Yahoo!.RTM., Google.RTM., and Microsoft Live.RTM., may utilize similar techniques for determining the relevancy of search results and SEO may include a site's attempts to more accurately appear in those search results.
SEO may be performed by improving the labels, tags and information on a site. Accordingly, it is desirable to improve the SEO of pages, so that those pages appear in search results for relevant search queries related to the content of the pages.”
By design or coincidence, I liked the use of ‘units’ when describing target terms (keywords and phrases);
“…wherein the act of selecting the related popular units from the plurality of popular units comprises comparing the existing page unit list with the unit frequency list to identify the related popular units from the unit frequency list that are related to the existing units from the existing page unit list.”
Essentially we compare the mined terms from the page with the desired terms from the popularity list (query logs) for primary and related terms.
Here’s some insight into trends and temporal issues;
“A longer time period may be used to establish consistently popular units, such as over one year or over five years. Alternatively, a relatively short time period such as a week, a day, or an hour may establish which units are currently popular.
A shorter time period may reflect current trends in the popularity of a unit. If the search engine optimizer 104 is repeatedly updating a page, then the short time period for popularity may be more effective to match the trends in popularity. Conversely, when the search engine optimizer 104 updates a page infrequently, such as once a week or once a month, then a longer time period for popularity may be more effective in determining those units that are consistently popular.”
I never thought we’d see a search patent with kw injection in it (hee hee);
“…the optimization may be based on injecting popular related units into the page or emphasizing existing popular units.
The page editor may edit the page such that the selected popular units from the selector, may be injected into the page, and the more popular existing units may be emphasized on the page.”