The Technology Behind Google Panda

The Technology Behind Google Panda

Technology Behind Google Panda

Google’s recent Panda update has caught the ire of many because it has effectively removed both high-quality and low-quality, but mostly low-quality, websites from Google’s search engine results pages (SERPs). While Panda’s effects are noteworthy, it is the advances in technology associated with Panda that make it the most awe-inspiring update heretofore and likely give some clues about Google’s focus and future course of action.

Google’s Priorities

Many who are involved with web marketing are aware that Google has begun to place more emphasis on the user experiences that websites provide, e.g. website speed, when calculating the quality of websites. Panda is another, albeit very significant, step in that direction.

Contrary to popular belief, Google did not name the update Panda because of a profound love for those cute black and white bears; rather, the update is named after a Google engineer, Navneet Panda, who developed a much more efficient (read faster) method of performing machine learning.

In short, machine learning is a form of artificial intelligence that attempts to learn from the quality of known experiences to predict the quality of other unknown experiences. Although machine learning can be used in a multitude of different applications, according to Rand Fishkin of SEOmoz, Google uses machine learning in its algorithm to learn what humans like about certain websites and extrapolate from that learning to determine if people will like other websites.

Going Forward

What does this mean for those that operate websites? While it is unclear exactly how much prominence and weight Google allocates in its algorithm to the results of machine learning, judging from the fact that Panda effectively removed some websites from Google’s index, user experience will likely carry substantial weight going forward. At the very least, these modifications likely indicate that more thought and consideration should be placed upon a number of the aspects associated with user experience design and implementation, e.g. speed, content, & design.

While I don’t know, with any degree of certainty, whether the following suggestions will help a website in Google’s SERPs; and frankly, Google likely doesn’t know either because they have effectively ceded some measure of control to the software that performs the machine learning and subsequent website evaluations, the following suggestions should improve the user experience of a website’s human visitors.

1. Speed

Likely the easiest-to-quantify aspect of user experience is a website’s speed, i.e. does it load quickly and do the interactive elements respond quickly. While we know that Google already places some weight upon the speed of websites, the importance of speed might may be intrinsically magnified because of the machine learning included in Panda.

The nice thing about website speed is that many of the changes required to increase a website’s speed are behind the scenes and can often be implemented website-wide as opposed to one-at-a-time. Google provides a great resource to analyze some general and broad HTML-related speed factors called Page Speed Online.

One of Page Speed Online’s must under-utilized suggestions is to utilize a content delivery network (CDN), e.g. Google’s, Microsoft’s, or Amazon’s CDN for large files like rich media. A CDN leverages already existing global server networks to distributed content from the servers that are geographically closer to a website visitor in order to decrease latency and decrease the time required to download relatively large files. While CDNs can have a significant effect on the speed of a website, they may actually make a website slower, depending upon the existing server configuration.

Beyond Page Speed Online’s suggestions, using asynchronous JavaScript (AJAX), i.e. JavaScript that does not block page rendering, to load remote attributes and scripts, such as Facebook, Google+, LinkedIn, and Twitter buttons, often has a very substantial effect on website speed. In addition, AJAX can be used to delay loading large, non-essential graphical elements (as we do on my law firm’s website) and decrease the amount of time required before a user can easily start to utilize content on a page while still permitting a richer, but somewhat delayed, user experience.

2. Content

Although somewhat more complex than measuring a website’s speed, it is fairly simple for software to determine what types of rich media are present on a website. Because rich media is generally popular, it is possible that the machine learning implemented in Panda currently places, or will place, a higher value on thosse sites that contain rich media as opposed to those that do not.

3. Design

Perhaps the most difficult aspect of user experience evaluation, but likely the aspect at which machine learning can excel the most, is website design quality analysis. Computer software is by its very nature logical, but high-quality website design is often not logical per se. Machine learning can, in theory, enable Google to determine the quality of a website whose design is not logical, but indeed high-quality.

Although a discussion of good user experience design is well beyond the scope of this post, Smashing Magazine has a great primer on the elements of good user experience design.

Steve Cook is a business lawyer and software engineer who runs a law firm web marketing company in Phoenix called ESQ Creative.

Guest Blogger