Thanks to Google itself and a lot of researchers, we know a lot about Google’s search algorithm. However, Google reveals what it wants search marketers to know and keeps the rest confidential. In reality, everything that happens between submitting a query and seeing the search results is a black box.
What We Know Comes From Three Sources
1. What Google Tells Us
Google wants to be genuinely helpful. It provides lots of honest and valuable information critical to effective SEO. Google’s vision aligns well with good marketing practices, so it’s worth paying attention to its advice. Another reason to listen is that, where once Google preached best practices only to get trounced by manipulative optimization techniques, the search engine has become adept at identifying and punishing websites that spam, manipulate ranking authority, publish low-quality content or otherwise breach Google’s terms of service.
On the flip side — and at the risk of sounding a bit tin hat — assume that anytime a Google spokesperson reveals something about their algorithm or suggests a best practice, it’s done with an agenda in mind.
Google wants to promote the company’s vision of providing certain types of search results and preventing others. While creating an algorithm to yield the best possible results is crucial, Google uses its own form of propaganda to convince companies to create the types of content it wants to feature (in terms of topics, writing quality and crawlability).
Much of what we know about improving search engine rankings comes from experience and observation. This is why I refer to SEO as a craft: part science and part art. As SEOs, we depend on each other’s positive and negative experiences to discover trends and understand their significance. As an example, understanding what types of content earn the most links or social mentions is less about rigorous testing and more about learning from personal and shared experience.
In the past, much observation was targeted toward beating Google’s organic search algorithm. It resulted in techniques like link wheels, content networks and content spinning. Going forward, observation will be less about trickery and more about best practices. People are looking for answers to questions such as what types of topics work best for B2B or B2C sites, how to make product pages “rank-worthy,” or how to track visitors without hurting SEO.
Correlation tests are fascinating. Researchers scour Google’s statements about search and observations by SEOs for whatever they can quantify. They’ll go through HTML, social media sites, content patterns and more looking for anything they can isolate and measure.
Organize all the possible signals, gather measurements on hundreds or thousands of data points, compare them to ranking results or site traffic and then publish the results. Even with the obvious caveat, correlation does not guarantee causation, it’s easy to understand why correlation study insights are popular and valuable.
Let’s look at Hummingbird, Google’s new algorithm. While it affected 90% of search results, Google implemented Hummingbird — an entire replacement of the algorithm — a month before notifying us. No one noticed.
One reason SEOs may not have noticed is because the biggest changes happened to long-tail queries. The long-tail makes up the bulk of search queries; yet, each query is low frequency. Due to low search volume, plus Google’s (not provided) referral string, it’s easy for minor results, on a site-by-site basis, to fly under the radar.
This meshes with what Google says about its new emphasis on matching the meanings of phrases with concepts rather than just matching the individual words in a query to documents.
The real reason people didn’t notice Hummingbird coming online is that while it is a totally new algorithm, it mostly matches the old one. When referring to the Google search algorithm, in reality, we mean a collection of formulas. Every year, Google makes hundreds of changes to the algorithm. Imagine these piling up year after year. Some rewrite old sections. Some are entirely new additions. As changes build-up, the entire collection becomes more complex and less optimized.
It makes sense that Google would reorganize all the pieces and combine many formulas to optimize, organize and clear the path for future improvements. Let’s call it housekeeping or spring cleaning. That’s conjecture on my part, but I don’t think it’s a big stretch, especially considering Google makes no algorithm updates without rigorous testing.
Another feature of Hummingbird is to make it easier to add and test future updates to the algorithm. In any environment, one can only test a limited number of things until it becomes too messy to evaluate the results. I’m sure Google has a long list of improvements they want to make, and Hummingbird is designed with that list in mind. Don’t be surprised if Hummingbird helps Google to make future changes more quickly and in greater numbers.
So while Hummingbird adds some new updates — apparently, how well it identifies and handles context — I expect it’s mostly reorganization, optimization and laying the groundwork for future improvements.
At Psynapsis.net our team of experts will provide you with all the tools and information that you company needs to improve it’s rankings and drive more audience to your website.