Online marketing information can change quickly This article is 9 years and 358 days old, and the facts and opinions contained in it may be out of date.
I would say at least 3 out of 10 valid SEO theories evolve from sheer stark raving lunacy, so I’m gonna take a bit of a stab at a decent discussion on one. Perhaps this is already common knowledge, but I haven’t heard it discussed a whole lot as of yet, and I really just felt like going off on a rant this afternoon. It should be disclosed that I am by no means qualified to discuss specifics of algorithms, and you are at risk of developing your own wild-eyed ideas from reading the information contained below.
As I was scoring the various elements of Rand’s incredible search engine rankings factors document, I noticed that many of the variables could be judged more or less important depending on what I would be using if I were doing a test query. I think most SEO’s are aware of this fact, but I really haven’t seen it discussed all that often.
During several sessions given by search engine representatives, I’ve heard them mention that they most certainly distinguish intent in different types of queries. Yahoo mindset is one of the more interesting recent SE developments, and illustrates in very simple form the deciphering of intent between commercial and informational queries.
From a taxonomy of web search by Andrei Broder:
A taxonomy of web searches
In the web context the “need behind the query” is often not informational in nature. We classify web
queries according to their intent into 3 classes:
- Navigational. The immediate intent is to reach a particular site.
- Informational. The intent is to acquire some information assumed to be present on one or more web pages.
- Transactional. The intent is to perform some web-mediated activity.
Before we discuss these types in detail, we need to clarify that there is no assumption here that this intent can be inferred with any certitude from the query. The examples below might have alternative explanations.
These seem like they are probably just top level taxonomy now. However, imagine if there were query specific algorithms for transactional vs. informational + the taxonomy of the DMOZ categories. So for transactional + business there was one algorithm versus a fairly different one for informational + computers. Then imagine going a heirarchy structure deeper (something like informational + business + jobs). Thinking about each level and the variables involved in a ranking algorithm, it is fairly easy to say that different weights and dampening factors could be used for different semantic taxonomies.
While I would imagine intent is quite ambigious and difficult to guage, Yahoo Mindset demonstrated that they have the top level pretty well clocked. How deep does the structure go now, and what are the criteria for determining intent? How big of a role is query intent or query specific variables currently playing in determining search engine rankings? I suppose at some level this is probably a fairly remedial way for my less than genius mind to comprehend in some way how semantics are being applied in the search engines in regards to rankings.
I remember Tim Mayer mentioning in one of his sessions that some of the main measurements for spam were “intent and extent”, which seemed very logical to me. It is also the reason why their is a double standard for large corporations using “IP-delivery” vs. small businesses “cloaking”. The intent is different, and the extent of the impact is much higher relevancy for a large company with a lot of information showing up in the results often vs. a small company with only a bit of relevant information showing up often. This could be seen as a double standard by some, but it is a valid one nontheless.
With search intent being so important to search engineers, it seemed odd to me that I couldn’t find more discussions from SEO’s on the subject query specific ranking variables. Of course, as with most search concepts I’m sure the engineers have much different terminology for the concepts I am trying to verbalize.
Examples of query specific variables
These are the handful that really stuck out in my mind as “query specific variables”. To some extent, near any variable in the search algorithms could likely be query specific. These are the ones where it was apparent enough to me to raise the question of just how much of an impact the other variables are reliant on the intent of the query to derive their individual weight. When ranking these in Rand’s study, it was fairly difficult because what could be highly important in one type of query may not be important at all in another type of query. Thus started my fascination with query specific intent and optimization.
Variables that potentially have large varying degrees of importance based on query intent:
- Rate of content additions
- Rate of link acquisition
- Rate of link rot
- Keyword frequency in document
- Related word use in document
- Related word use in anchor text
Consider for each of these how different you would expect the weighting of the variables to be if you were looking for a news article versus a how-to document. Extreme examples of how query intent SHOULD influence rankings can quickly lead to the more ambigious areas of deciphering where the impact level should be with the artificial intelligence for ranking documents for any query.
Query Specific Optimization
It is already fairly tough already to do engine specific optimization. Personalization and vertical search are quietly rolling into the search results. Will we soon to have to worry about query specific optimization? As vertical search gets further embedded into the search results we will see a very visible example of just how important query specific optimization will have become. Strategies for acheiving granularity as well as “short-tail” phrases will be increasingly important. Maybe then marketing agencies will start to wise up that SEO is not just about exploiting short term algorithmic holes, but rather about long-term online marketing strategies that place odds on the proven best practices and ways to make them better.
Using all the tools at your disposal will be extremely important with query specific optimization. Using a press release often to get in the “one-box” for a new query. Using a froogle feed with individual price per product breakdowns for top listings on a product search. These are just examples of query specific optimization for the one box. What happends when the one-box is no longer identified and vertical search is just a portion of the “natural search results”?
Search Ranking Variable Questions for Speculation
- How large of an impact does query type have on how rankings are determined?
- How specific are engines able to currently determine search intent beyond the examples given by them?
- How granular are the search algorithms based on query intent?
Consider one last example of different algorithms being applied to:
- 1. Informational + arts
- 2. Transactional + business
- 3. Informational + business + real estate
- 4. Transactional + computers + softwareDMOZ structure for reference
For highest relevance, I would certainly want different algorithms applied to each of these queries.
Of course, most SEO’s have been on to this for years. I’ll gladly take 10 visitors arriving for “buy plasma tvs online” than 100 visitors for “plasma TV comparisons”. I’m just guessing the SE engineers have intent much more refined than this, and curious as to how large of a role it is already playing in the rankings for individual query specific results.
Resources on Query Specific Intent and Query Specific Ranking Variables
- Taxonomy of Web Search by Andreai Broder – PDF form
- Web search intent induction via automatic query reformulation *PDF file
- What if search engines could read your mind?
- A database of intentions – John Battelle
- Search engine optimists by Bryan Eisenberg