RankBrain is the name given to an artificial intelligence (AI) program that is being used to improve the search results. RankBrain is a machine-learning AI that is constantly improving the search results. The objective is to create a machine that can think more like a human, and therefore, improve the results that are returned. It does this by analysing search terms and then returning more relevant results.
Improving Search
We do not know any of the inner workings, but we can make a few educated guesses about how it is functioning. Google knows that the first results in search do not always provide the best information for the searcher. When a poor result is viewed, the user will return to the search results and seek another website in the listings – this causes a website to have a high bounce rate, which is reported in Google Analytics. Over time, pages with high bounce rates will be demoted and those with a lower bounce rate will rise up in the rankings.
This is just one simple method that Google may use to improve rankings. The purpose of an artificial intelligence is that it must try to think like a person, and anticipate what factors on a website will provide the best results. These factors may not be related to quality of the content, site architecture or external links, which are the main ranking factors at the moment. There are many excellent websites that currently rank badly in search, but provide a superior service. RankBrain may be great news for such sites and businesses.
Greg Corrado, Google’s senior research scientist working on RankBrain, spoke with Bloomberg about the development – the fact that Google has done an interview with a financial magazine suggests that this could be a major shift for the company, and something that might be of interest to shareholders. He explained that RankBrain is essentially another filter that is applied to the search results. According to Corrado, RankBrain has outperformed humans at selecting the best pages to match a search – similar test results were reported during the testing of Penguin and Panda, both of which have radically changed SEO practices.
A “Large Fraction”
Google has described RankBrain as filtering a “very large fraction” of search queries, which Bloomberg quoted as being 15 percent of daily queries.
The key difference between RankBrain and previous search features is that RankBrain has been built to learn. By monitoring searches and how people interact with the pages they view, Google’s RankBrain can learn which results are providing the best experience. At the end of the day, Google wants to deliver people to the web pages that provide them with the best experience possible, and as good as the algorithm has been, the common frustration among many webmasters is that it often overlooks some excellent content.
Harder to Game
There is another possible reason for handing the search engine over to AI – in theory, it should be harder to manipulate the search results with SEO. If factors such as page content, title tags, headers and links are demoted as the main ranking signals, SEOs will have to focus only on creating the most useful resources and services. Could this mark the “end of SEO” again?