Google uses machine learning as part of its search algorithm
Daniel Foley October 07th 2016 in Marketing 0
Google RankBrain and Machine Learning
Googleâ€™s algorithms have been evolving at an ever increasing rate. With the introduction of RankBrain and access to a vast amount of analytical data, RankBrain is finally becoming more mainstream within Googleâ€™s core algorithm.
With the evolution of Panda, Penguin, HummingBird as well as specific algorithm updates (Payday / Farmer etc.) Google has been moving towards machine learning which is based more on user experience instead of dated methods such as link evaluation.
For many years, link weight has played a large part in Googleâ€™s evaluation of a website, its authority and value to the end user. Unfortunately, link evaluation has become a vast minefield of link networks, paid link building, blog networks and other schemes designed to influence organic performance. As such, Google has had to incrementally release updates to try and get 1 step ahead of SEOâ€™s, marketing companies, freelancers and link builders to focus more on content quality, website usability and other factors instead of pure link equity.
RankBrain is evolutionary as part of Googleâ€™s core algorithm, however, it is still combined with over 200 other metrics to weigh up the best matching search results to the query. Initially it was introduced to deal with longer-tail queries, questions and dynamic queries.
Whilst RankBrain uses machine learning, other Google algorithms such as Penguin and Panda are not. Core adaptations were introduced to de-value or re-evaluate websites based on a combination of different factors during a Google update / SERP refresh.
Moving forward, Google has been adapting machine learning technology as part of RankBrain, it may well lead to Google relying more on machine learning vs. pre-programmed evaluation. It would make sense given that there are now billions of websites for Google to evaluate and merit depending on content quality, website usability and user experience, link equity and more.
Google has access to growing data thank to the uptake of Chrome, Analytics products, Android and more. Using this data will take some complex machine learning, but inevitably, all of this data will have an impact on search results.
One key element to watch for will be the utilisation of analytics and chrome data to monitor user behaviour at query level. This will paint a picture of SERP and website performance at a granular level. Perhaps content and link evaluation alone are insufficient for true, real-time algorithm adjustments, whereas real-time user data and site performance will be a more reliable way of evaluating if a website is a best fit for a query.
There are 0 comments on this article.