Google's Panda Update: Is it all Behavioral?
On February 24th Google issued their Pandate: "Thou shall produce quality content or thou shall pay the price with lost traffic and revenue!" What followed was vague statements from Google regarding what quality is. In my mind, this update by Google is an admission that they cannot determine the quality of an individual page. Why do I think that?
As I analyzed the impact of the Panda update on various sites, I noticed (although not every page).
- All page types were hit
- Pages with great content were hit
- Pages with thin content were hit
- Pages with mediocre content were hit.
Google did not take a scalpel and slice up poor content. They took a bazooka to the entire site.
That immediately raised the question in my mind:
Why would Google release an update that indiscriminately hits all pages on your site??
The answer is simple, they cannot evaluate the quality of single piece of content. If they could evaluate a piece of content, they would. If I had a blog with 100 great posts with in-depth analysis and 100 thin posts which regurgitate other chit-chat around the web, wouldn't it be in Google's best interest in boost my great posts and demote my poor ones? Of course it would!
Sure, they are good at detecting duplicate content. That problem can be solved with raw computing cycles. They have learned how to detect machine generated fake prose. I am positive they have invested millions of Google dollars on the problem of content quality and the Panda update tells me they can't tell by the piece of content itself (at least not with out an unacceptable level of errors).
How come? Well the aspects that make up quality for a page differ from one page to the next. Sometimes complex prose is required (for example in a medical site). Sometimes short and sweet is best (a FAQ). Sometimes an image or a map does the job.
What is a quality page?
If you look at statements coming from Google, and if you think about it logically from Google's perspective, they want to deliver pages that result's in a quality search experience.
The quality of a page must be considered in the context of the search result that drives a user to a page.
The page must answer the question the user had in mind when they entered their query into Google. It is very difficult to look at a page and determine whether it will result in a quality search experience simply from the content on the page.
Consider the question: What was the 2011 deadline to file my taxes without an extension??
The Answer, 'April 18,2011' is a completely satisfactory answer from a content perspective. How can Google possibly judge that small piece of text and determine it answered the searcher's question?
If that is the entire page, it is a quality page search result page because it meets the need of the user who typed the question into Google. In fact, additional superfluous content on the page, not directly related to the question (for instance related tax questions) may hurt the quality of the page because they will draw in traffic not specific to the main point of the page. So length and depth of content does not give a clear indication of the quality of the page.
In other cases, a wonderful page may fail because it still does not satisfy a basic need of the user.
Consider another page about a hotel. The users searches for the "Hilton, New York" because he wants to book a room. That is all he wants, been there, done that, just wants to book. He lands on a fantastically written page, great pictures, user reviews, great content. But the "book button" is obscured behind a tab. The user can't find it and can't book. Great content, but a poor page; the user will go back to the search engine to find another site.
So Google says, "I give up, I can't figure out the quality of a page by the algorithmic analysis of a page content."
User Behavior is the Likely Answer
There is one true indication on the quality of a site, and that how the user responds to the page based on his actions. In the past I was doubtful that Google would use behavioral data to rank pages. Very few individual pages have enough pageviews to reach statistical significance. But Google has gotten around that by looking at the whole site. By aggregating sites performance data as a whole, Google is able to reach conclusions about the site, even if they can't reach a conclusion about an individual page. It is the only thing that makes sense to me based on the analysis of the various sites I have done. Plus listen to the recommendations coming from Google in response to Panda, it is all about the user experience.
If it is behavioral, then what metrics are they using?
While I am quite sure they are not using Google Analytics for data (for many reasons - not the least of which is it is an incomplete data set), they still have plenty of information to go on:
- Did the user return and select another result from the SERP for the same search?
- How quickly did the user return? If it was very quick, then likely the page did not meet the need?
- Does the user return to the site?
- Did the user request a site be banned from the results?
- Did the user click on the more results from this site button?
If this is indeed behavioral as I believe, then the face and nature of the professional SEO has changed. If your SEO isn't focusing on user experience, he is not doing his job.
There is a lot to think about as a result of the Panda update. Even if you haven't been affected, you could be in one of subsequent updates. Feel free to join in with comments. I am particularly interested in ideas related to what user data Google could be collecting to reach their conclusions.
Coming up in my next 2 posts:
- Is Google looking at bounce rates?
- Beware of the back button