18th Aug 2016

Google’s Newest Employee: RankBrain

In October 2015 Google unveiled its first attempt at using artificial intelligence (AI) to enhance user searches in an attempt to help them produce more accurate and relevant results. This AI was called RankBrain and was bought in as part of the Hummingbird algorithm that predominantly controls the search criteria. Google, in its usual shroud of secrecy, informed the digital marketing world that RankBrain was the third most important element of HummingBird but refrained from telling us the two most important criteria until earlier this year. In order they seem to be links (like votes), words (what is typed in and what it has in its databanks) then AI. That’s more information than previously released, but not an awful lot more.


Why is RankBrain so Important to SEO?

Bloomberg was one of the first to report on the deployment of AI as part of the search algorithm, but informed the eagerly engaged public that it would only be dealing with 15% of total enquiries processed by Google. That’s still a huge number, at a whopping 3.5 billion enquiries daily it amounts to approx. 525 million RankBrain searches every 24 hours, so it’s no small task to calculate. To try and work out how important RankBrain is to SEO, we really need to understand what searches it deals with mainly because the chances are you’re search is not going through AI – so it makes no difference. But if you are one of that prized 15% – it makes a huge difference.

How does RankBrain work?

That’s the current billion dollar question, and as always, Google is keeping the answer close to its chest. What we do know is that RankBrain is adept at selecting relevant pages to the more ambiguous searches by its ability to compute vast amounts of language that has been translated into mathematical vectors. When all is said and done, RankBrain is still a machine – albeit a ‘learning’ machine, or more correctly, one that has a better ability to cross reference the language entered by the user.

In effect, Google has created a ‘guessing machine’. It takes random enquiries that are vague or unrecognized and looks for reference points through its vast language vector database and throws up the closest results it can find based on the original search criteria. The next question is:

Is RankBrain any good?

The answer at the moment seems to be… yes, it is.

One of the tests it went through was humans vs AI – and not just any humans.

Google search engineers were given a random set of pages and asked to sort them into the order they thought the engine would sort them by relevance. The same pages were run through RankBrain. The humans were right 70% of the time but AI got it right 80% of the time, but how relevant those results were to the enquiry was not part of the test.

One mantra that courses through Googles blood is that everything can be ‘fixed by technology’. Technology will always be forefront of the search engine, and any other endeavor they undertake. In this case, they have used technology where the humans fell short. RankBrain can cross reference massive amounts of information, but also look through preferences and choose results that the user might like. AI can see patterns that human operators can’t – and they can see them quickly too. This increases the probability of them being right.

Bloomberg used the example of a search query that was pretty vague. They typed in, “What’s the title of the consumer at the highest level of a food chain?” RankBrain kicked in and found them the right answer.

The Content Marketing Institute went one further. They asked a multitude of random, but realistic questions. One was, “shape of conversion optimization in the future of digital marketing and beyond”, and then performed a voice activated enquiry of, “the name of the fat guy with the beard in the movie with the four guys in Vegas?” Both were random to the point of almost being obscure. How did RankBrain do? Really well. In fact, spot on.

Even though RankBrain cannot truly ‘think’, it can pick up the context of the question quite well from a long tail enquiry and produce the answer you’re looking for. It’s not infallible, but it’s still quite impressive.

How can RankBrain help my marketing?

That’s a loaded question. RankBrain can help make your marketing easier – well, your SEO marketing easier. You no longer have to be a slave to keywords. The new algorithm gives you the freedom to write your digital content as you want it to be written, and if it’s done well, it will rank highly in a Google search. Don’t get too excited yet. It is not a guarantee that you will turn up at the top in a search result. RankBrain is only one of hundreds of filters for content so you still have many others that are not run by AI that it has to pass through.

Just being able to pass the ‘context’ test does not mean it will pass that same test every time something even vaguely similar is typed in. At the end of the day, it’s still a machine – and one small part of a very large machine at that. You still need to apply the rules of ‘smart marketing’ to your campaigns if you want to consistently rank anywhere near the top.

And don’t forget – it only deals with 15% of all user searches. This 15% was the amount of ‘new’ or ‘unrecognised’ searches that were performed each year that were left unanswered or drew up random results that had no correlation with the enquiry. If you’re relying on RankBrain as a ‘hail Mary’ on your search rankings, it stands to reason that it’ll only save you 15% of the time. Making sure all your content passes this contextual test is a smart thing to do for your SEO, but you’ll still need to make sure you have keywords mixed in with this natural language to help with the context.

What else does RankBrain look for?

This is another guarded secret from Google, but we do know that the AI is programmed to look for the most relevant, quality content, so increasing your natural style of content and increasing your authority in the industry will help get you noticed. White papers, case studies and authoritative links can all help raise the profile of your site which will help your SEO.

Make sure your content is well researched, contains authoritative links to credible sources that support your statements and raise your credibility in the industry. It’s worth investing the time to create the best content you can, mostly because it’s not just RankBrain that values it, the other bots do too.

The employment of RankBrain is the future of Google. So far it has been proved to be such a success that it makes no sense to abandon it. Chances are that Google will look toward AI as the solution to more of its problems, increasing its importance in digital marketing. As users get more of the information they are looking for, Google will follow the pattern. After all, what the people want is what Google wants.

To keep up with the development, you need to take care when crafting any content, have a definite buying persona in mind, writing in an informative, conversational tone. Think about long form key words like a Google bot would, but don’t let them dominate your content. In other words – write with passion! Then back it up with creditable references.

Get used to creating content with AI in mind as it’s not going to go away any time soon. By keeping one eye on the bots and the other firmly on your audience, Google’s RankBrain will help you get the best ranking for your company.