'RankBrain': Artificial Intelligence & the Future of Search

Artificial IntelligenceRankBrain & Artificial Intelligence

Last week it emerged that Google has been using an artificial intelligence system, nicknamed ‘RankBrain’, to help refine and qualify its search engine results. Google deals with a slightly staggering, yet not that surprising, 3.5 billion searches per day (1.2 trillion searches a year) and 15% of these daily searches are new queries that haven’t been seen by the search engine before. RankBrain utilises mathematical ‘entities’ to understand written language and, if it doesn’t understand a particular word, it is able to make a guess about words and phrases that could have a similar meaning. This artificial intelligence system is particularly effective at working out the meaning behind more complex or ambiguous search queries and RankBrain has become Google’s third most important ranking signal (Google has refused to state the first and second). 

The rollout of RankBrain began earlier this year and reports state it has been live globally for several months. Once considered something out of science fiction, the use of artificial intelligence is actually a logical step for the search engine; Larry Page has often described his vision of the ultimate search engine as one that could ‘understand everything that you asked it and give you back the exact right thing instantly’. Incorporating and developing artificial intelligence is just one more step in Google’s overarching strategy for its search engine that they summarise in three words as: answer, converse, anticipate.

Answer, Converse, AnticipateArtificial Intelligence

It could be argued that Google has fulfilled the ‘answer’ part of its strategy for years; Google dominates the search engine market and the phrase ‘to Google’ has become part of our everyday vocabulary. The addition of answer boxes and the Knowledge Graph to the search results in recent years has demonstrated Google’s aims to improve how they answer our queries by providing the information faster and more directly. 

In terms of ‘converse’, Google’s ‘conversational search’ update allowed users to not only speak their search queries to their phone or computer, but continue their ‘search conversation’ with further questions. For example if you search for how old a particular celebrity or public figure is you can then ask a follow up question like ‘how tall is he’ and the search engine will be able to follow the ‘conversation’ and provide you with the answer. The Hummingbird algorithm update also improved Google’s understanding of conversational language, by putting more focus on the meaning and context of a search query rather than just looking at keywords.

With ‘anticipate’ Google wants to provide the answer to your next question before you have asked it. The Knowledge Graph was, in part, a way to do this. By not only answering your question but also bringing up an extra summary of relevant content around the topic, the idea is that any further questions you have on that topic are already answered. Google Now is also key to the ‘anticipate’ strategy and Google’s aim is for the personal assistant to give you the ‘right information at just the right time’. 

The three areas of ‘answer, converse, anticipate’ are all feeding into a larger strategy that is ‘understand’. This is semantic search; understanding the intent, context and relationship between keywords in a search query to provide the most accurate and relevant results. By fulfilling ‘answer, converse and anticipate’, Google is able to start moving towards ‘understand’ and the use of artificial intelligence – both of which are key in achieving the Google founders’ vision of the perfect search engine. 

Artificial IntelligenceThe Future of Search & SEO

What, then, does this mean for the future of search and search engine optimisation? If Google is working towards a search engine that understands everything it is asked, how do website owners ensure that their websites are found in the search results? SEO professionals claim that ‘content is king’ now, but the importance of having high quality content that addresses long tail search queries will only increase in the future. Website owners should be thinking like their users, what information do they want to know? What questions do they have that need answering? Does my website provide this information and answer these questions? 

Ensuring that Schema mark-up is implemented properly on your website will also be crucial; Google has often pointed out that Schema mark-up can improve the search engine’s understanding of the content on a webpage. It seems clear that if Google wants to produce a search engine that can ‘understand’, webmasters should be doing everything in their power to help this process if they want their website to be found in the search results. While we’re still a little way off the ‘Star Trek’ computer or the intelligent computer operating system in ‘Her’, RankBrain signals an intriguing step towards Google’s ‘ultimate search engine’.