Have you heard about Google’s RankBrain? Here’s why you need to take it into account for your search engine optimisation.
What is RankBrain?
Many of the details (like much of Google’s operations) are shrouded in secrecy. But since it was revealed in 2015, more information has come to light.
Basically, it’s an artificial intelligence machine learning program. It teaches itself and applies its learnings to search results.
It first came about (as far as we know) in 2013 when Google were looking for solutions to answer the 15% of queries they’d never had before. In order to interpret what people were looking for, it starting using machine learning to understand links between words.
What impact is it having on search results?
Although we don’t know exactly, Google have said it’s their third most important ranking factor (behind content and links). From 2016, they have been employing it on all searches. It’s likely that it’s doing two main things:
1. Interpreting the intent of the query, rather than matching the exact words
Whereas previously exact match keywords would show top, now matching the exact keyword becomes less important as RankBrain learns what we really mean.
For example, when searching for the name of a film, you’ll be shown nearby screening times above information about the film, as Google infers (perhaps from previous search behaviour) that you’re interested in watching the film, rather than finding out about it.
Another example is the search term “How to run a marathon”. None of the top results contain an exact match to the term “How to run a marathon”. Instead, all the results show information on what they’ve interpreted to be your intent – how to train for a marathon.
2. Learning which results people are finding most useful and applying this back to search results
It’s also likely that RankBrain is learning which results are successful for a search term, and which are duds. This means if a result has a high bounce rate and low time spent on site, then it may be demoted down the rankings.
What should we do?
Keep creating valuable content that’s written for humans.
As Google is learning to interpret the overall sentiment of the query, there’s less need to create dozens of pages to target different variations on keywords. Instead, you can create one really good page that targets all searches around the same intent.
The content needs to be excellent, so that people don’t leave straight away, sending you plummeting down the rankings.
Continue keyword research
Some SEO analysts have suggested that RankBrain is also learning what is important for different fields, and so may also be applying different ranking factors dependent on the topic. So it’s even more important to do keyword research to understand what Google might be prioritising for your target topic. For example, some results might have a greater weight on freshness (such as information on natural disasters), while for other topics (perhaps reference posts) inbound links would still be an important metric.