What is RankBrain and how does it work

RankBrain is a machine-learning artificial intelligence system. Its primary purpose is providing more relevant search results for users with queries that are unfamiliar to a search engine.

The most important Hummingbird factors

RankBrain is the latest major addition to the Google search algorithm called “Hummingbird”. The algorithm consists of around 200 major factors and a couple of thousand minor ones. RankBrain became a part of the algorithm in late 2015 and became the third most important factor. The first two most important factors are still links and quality content.

RankBrain is specifically designed to alleviate the issues that can arise when searchers enter long-tail keywords that are somewhat confusing to the algorithm. Previously, the prime focus of the search engine was always keywords. Now, RankBrain’s role is to put more focus on the intent of the searcher.

Stop words and search results

Stop words are words that we commonly use in ordinary speech. Search engines usually omit them when looking for results.

Some examples of stop words would be “and, when, so, that, where, how, when” and so on. It’s estimated that stop words are 25-30% of all the written content online. So, where is the problem and how can RankBrain fix it?

One of the primary problems with previous search results were stop words. Before RankBrain, stop words were regularly excluded from long-tail keywords. Too often, this system caused the results to be very inaccurate.

Confusing query example

Gary Ilyes from Google presented the problem with the query “How to beat Super Mario Brothers without cheats”. If search engines omit the stop words – in this case, “How” “to” and “without” – so that the query actually becomes “beat Super Mario brothers cheats”, you will have the articles with cheat codes in the top results. All top results for this query contain cheat codes, which is the complete opposite of the searcher’s intent.

RankBrain is a learning system, designed to recognize and work with stop words. It is trying, to the best of its abilities, to provide educated guesses on the searchers intent. This machine-learning should ideally provide us with better and more accurate results, but it is still a long way from perfection.

How to optimize your website for RankBrain

RankBrain is primarily targeting non-understandable and imprecise queries. It is impossible to optimize content for such instances. Honoring the regular rules of SEO, the best you can do is to optimize your content for human readers, and not for search engines. RankBrain will reward content that is useful to actual people rather than content that is adapted to primarily draw search engines. The more relevant RankBrain thinks a certain web document is, the higher it will rank. Once again, as mentioned before, it is only possible to optimize for RankBrain in broad terms.

The future development speed of this machine-learning system is uncertain. In an ideal case, this part of the Google algorithm will become more and more adept at figuring out the searcher’s intent. At some point, it will have a massive impact on all the difficult to understand long-tail keyword queries. However, from an SEO perspective, there will be no specific way to ensure a practical optimization for search engines.

The best advice is to use common good practices and guidelines that most SEO’s are following today.