RankBrain is one of Google’s top three ranking signals, along with content and links. This has been the case for over six months, so it is time we took a deeper look at what RankBrain is and how we can optimise our websites so that they perform better in Google’s natural search index. RankBrain is Google’s machine-learning system, which is used to improve search results.
The news of Google’s RankBrain update first hit the financial press in October 2015 with a report in Bloomberg. This in itself was a radical change to the we learned about Google updates – no longer were SEO community blogs the first to break the news, Google went straight to the financial press so that its shareholders and investors would be the first to learn about the positive new changes. Since then, RankBrain has been optimising Google’s index.
What Is RankBrain?
RankBrain is a machine learning system that uses artificial intelligence to better provide the best websites for specific search queries. The use of “artificial intelligence” in search has been on the cards for many years – many of Google’s most important search technicians have a background in AI. Larry Page, one of Google’s founders, was the son of Carl Page, a pioneer in computer science and artificial intelligence. Marissa Mayer, who was involved in Google search for over 10 years, holds a Master of Science in computer science, with a specialization in artificial intelligence. Today, many of Google’s research employees are involved in machine intelligence and machine perception. Although it is not all about search, the search index is always central to Google.
Prior to RankBrain, Google relied heavily on just content and links – now there is a whole new layer in the mix. RankBrain aims to understand what a user really wants, and provide the most suitable web pages in return. This is a core marketing principal, and from what we have learned so far, RankBrain is essentially carrying out constant market research to ensure that the system provides the best service. The big question for us is, how does it work and how can we optimise business websites to gain better rankings in Google?
To understand what RankBrain is we need to understand a little about the two core elements behind it: machine learning and artificial intelligence. Machine learning is when a computer can teach itself how to do something new by observing and testing new approaches – this differs from the traditional method or programming a computer to carry out set commands. Artificial intelligence is the process of a computer, or machine, acquiring knowledge and understanding of the world. Strictly speaking, artificial intelligence (AI) requires a machine to be sentient and to know that it exists, so RankBrain is not true AI. However, the term is commonly used to describe more advanced machine learning.
So, how is all this linked with Google search? Well, RankBrain forms a vital part of Google’s algorithm, the latest version being called Hummingbird. It is used to analyse billions of pages to find the ones that it computes to be most relevant and useful.
Is RankBrain a Ranking Signal?
No. Unlike content and links, RankBrain is not actually a ranking signal – in a way, it is how Google processes all the information from the 200 plus ranking signals that it uses. However, Google says that RankBrain is one of the top three ranking factors.
The key change is that, whereas in the past every Google search had a human input, with RankBrain, a computer is making the decisions on which pages to rank. The human input that we speak of is the process of creating stemming lists and synonym lists, as well as setting database connections to relate similar topics. Although Google’s algorithm has been using signals such as PageRank (essentially counting and qualifying links), and reading content in meta titles and headers, to understand what a page is about, it has relied on people creating these stemming and synonym lists.
The Long Tail Search Queries
For short search terms that are commonly used, the old system provided good results, but it was the long-tail searches. Whenever somebody typed longer questions into Google, it often failed to return the best result. Google would often take the keywords from the query and return only that, and the result was often a frustrated searcher hitting the browser back button.
In the Bloomberg article, Google explained how the search engine can now better understand the intent of a searcher based on the words being used within a specific context.
What this really means is that Google now interprets a search terms before returning the search results. For example, a long question typed into Google may include words that have multiple meanings, and if Google just took the main words as the key search terms, it would return irrelevant results. Now it can understand what people actually mean. How it does this is not clear, and while Google talks of machine learning, some SEOs, such as Rand Fishkin, believe that at times it may be using a much simpler method of analysing bounce rates and click trends in search.
It is thought that the main purpose of RankBrain is to allow pages that use more natural language to rank better than those which rely heavily on optimising keywords on the page and linking schemes. This is allowing pages that have good quality content to out-rank better optimised pages.
For example, in the past few months we have seen more pdf files appear in the search results. In general, pdf files do not attract so many links as other web pages, and they are not optimised with metatags. However, pdfs tend to have great content that can be more authoritative than HTML pages. It is therefore only natural that these should appear higher in the search results.
Of course, Google’s biggest bugbear is spam and it has invested heavily in systems to combat this. The Panda and Penguin algorithms have had a great impact on search results and managed to downgrade millions of pages that had either been over-optimised, or only ranked well because of aggressive link building campaigns.
One area that Google still has to manually tackle is that of complex link building strategies that are designed to hide their true intent from both users and Google. It can often be difficult for a SEO researcher to determine how a black hat SEO is gaming Google, simply due to the amount of data that needs to be gathered to formulate any concrete conclusions, but in theory at least, a computer should be able to identify patterns and behaviour to uncover link networks.
How To Outrank Competitors RankBrain
Because RankBrain is essentially analysing information to determine its relevance and importance, we believe that the best way to improve rankings is to improve content. Make your content more in-depth to provide Google with more information to work with. Past SEO research has already shown that pages with over 1200 words tend to rank better in Google, so make sure that you are always providing a more comprehensive description of products of services.
Do not be afraid to use technical jargon either – although it is always a good idea to write in plain English, it is advisable to also provide greater detail for those who want to know more. We have already seen businesses implement this approach by adding more detailed information into scroll boxes and pop-ups on a page – the information is read and indexed by Googlebot, but only seen when a reader chooses to activate the scroll box or click the “learn more” button,
Many businesses continue to operate from websites with limited written content, which is often duplicated from other sites. Carrying out a comprehensive site audit and implementing a site-wide content strategy is possibly the best way to take advantage of the Google’s newest ranking factor.