You type the word “text” into Google; what sites do you think will come up? Well, thanks for Google’s push to create semantic search, you are likely to, initially at least, see links for both the communication method and the definition of it as a book or a piece of writing. As you select and search further, Google will get a better idea of what you are looking for and provide you with more relevant returns. The algorithm that is driving this whole process is called Rankbrain, and it makes Google just a little more, albeit eerily, human.
What is Rankbrain?
Rankbrain is Google’s new brain child. It is an artificial intelligence system that incorporates machine learning in order to improve the process that Google uses to identify, sort, and interpret search engine results. It essentially educates itself as opposed to being programmed or taught by humans. This is the core of machine learning. Rankbrain is not Google’s algorithm for search, rather a component of it, a cog in the wheel. To be more precise, Rankbrain reviews, sifts, and sorts billions of pages that it has been made aware of through its machine learning process and identifies those that it deems to be the most relevant for a specific search term, keyword string, or query.
Rankbrain and Google’s Search Algorithm
In typical Google fashion, not much information has been released about Rankbrain. The company tends to be pretty tightlipped regarding its ranking factors, but it is releasing some information, just a little, to provide some usable information. This strategy is reminiscent of Google’s push for mobile optimization, encouraging businesses to create mobile friendly websites and improve their SMS marketing practices to reach the 87 percent, or 6.8 billion, people worldwide who use mobile phones.
Hummingbird, Google’s latest search algorithm, incorporates Rankbrain into its inner workings. It is the overall search algorithm that has been around for years, but was not given a formal name until 2013, after one of Google’s famous, usually covert overhauls. It contains a variety of parts including Pigeon (local results); Penguin, Panda, and Payday (spam fighters); Mobile Friendly (boosts mobile friendly pages to the top of the mobile rankings); Top Heavy (lowers rank of ad heavy pages); and Pirate (fights and prevents copyright infringement).
Some users confuse PageRank with Rankbrain, but they are very different. PageRank addresses how pages are ranked by identifying other pages that have links pointing at the pages and giving them credit accordingly. It is the first part of the ranking algorithm that Google ever named.
How does Rankbrain Learn?
All of Rankbrain’s learning is done offline. According to Google, it is “fed” batches of previous searches. Each batch of historical searches is a potential Rankbrain update. As it reviews them, it actually learns from them, how to make predictions, thus strengthening Google’s semantic search. Before those predictions go live though, and are allowed into the official Rankbrain protocol so an updated version can be released, experts at Google test each prediction to ensure it directs as it should. If it does and gets the all clear, then Google goes live with the latest Rankbrain version and the offline learning begins all over again with a new batch of historical searches,
What is Rankbrain’s Connection to SEO?
Websites can utilize Rankbrain to increase their SEO by doing away with traditional SEO that consisted of one or two relevant keywords and short content. Instead, they need to opt for several related long tail keywords or phrases and longer, complex content. The keywords should be terms that a user would select to find the information. In other words, a website about mobile marketing would use terms that may include text marketing, SMS, and digital marketing.
Rankbrain is a powerful tool that savvy search engine optimizers can use to not only improve their own performance in the search engines, but also to better serve their customers by making their site easier to find.