Google Search’s machine learning system RankBrain is now used in the processing of every query received by the search engine, and affects search rankings “in a lot of queries”. Google Senior Fellow Jeff Dean revealed the news in an interview with Backchannel editor Stephen Levy published late last month.
The corporate rollout of RankBrain has taken place in the space of roughly one year. The system launched in April 2015, and as recently as October Google employees were publicly stating that RankBrain was only used with a minority of searches. Its wider adoption reflects a growing confidence in the machine learning system’s abilities – or perhaps more precisely, a confidence in the machine learning system’s growing abilities.
Here’s a quote from Stephen Levy’s article:
“Dean says that RankBrain is “involved in every query,” and affects the actual rankings “probably not in every query but in a lot of queries.” What’s more, it’s hugely effective. Of the hundreds of “signals” Google search uses when it calculates its rankings (a signal might be the user’s geographical location, or whether the headline on a page matches the text in the query), RankBrain is now rated as the third most useful.”
What is RankBrain?
RankBrain is an AI system that plays a role in Google’s search ranking algorithm, Hummingbird. The exact nature of that role is not publicly known, and is the subject of widespread debate in the search marketing community.
So far, so ordinary, but RankBrain is special for one very good reason: it independently refines its own processes as it goes along, resulting in better search results for the user. Google are keeping typically tight-lipped on the details of how RankBrain works, but we think it would be reasonable to venture that it will assess the efficacy of the search results it generates against metrics such as CTR and dwell time, before making adjustments to its processes accordingly.
You can read more about RankBrain and machine learning in this week’s spotlight feature.
Here’s a puzzler: how can marketers make contingencies for the effects of a top-secret system that can evolve of its own accord?
That’s a tricky question to answer, but as far as we’re concerned, RankBrain’s focus on quality for the end user points to an ever-clearer truth for search marketers: ranking well is becoming more greatly predicated upon your content’s quality and relevance to the user’s search. That means you’ll need to work harder than ever to ensure a tight relationship between your content and your meta data, and also to create a website your visitors will be pleased to have reached.