Google recently updated the algorithm it uses to rank search results, and websites have since reported major disruption to website traffic. In many cases, websites are reporting drops of 30 to 40 percent in website traffic since last weekend, which can have monumental impacts on sales. In other words, people aren’t happy.
The new algorithm is based on something Google calls ‘bidirectional encoder representations from transformers’ or BERT – a tool that optimises natural language processing using AI, to better understand what people are searching for when they try to search for something in Google. In short, the new algorithm tries to mimic the human brain so that it can more intelligently decipher what users want to know. What is super cool about the new algorithm is that it is actually capable of learning, too. Over time, it develops associations with particular search queries and the pages users end up spending the most time on after typing in a specific search query. It then uses natural language processing to improve search results. The development is incredibly exciting, made even more so by the fact that Google has now open sourced the technology, stating when it did so that “anyone in the world can train their own state-of-the-art question answering system (or a variety of other models) in about 30 minutes on a single Cloud TPU, or in a few hours using a single GPU.”
Apparently, Google routinely updates its algorithm, but this particular update significantly changes the way search queries are interpreted by the machine. In other words, pre-algorithm update, Google may have plucked the words ‘where’, ‘buy’ and ‘dog’ from a search query saying ‘where can I buy a dog in India’, but the new algorithm would give equal weighting to the ‘India’ part. As you can imagine, this significantly changes the particular search results delivered by Google. BERT is what Google describes as a “neural network-based technique for natural language processing (NLP)”. It’s a new approach to SEO that more accurately interprets user queries, and technically that should mean happier customers all round as a result. Unfortunately, it hasn’t worked out this way for some. And it forces businesses who depend entirely on their Google ranking to consider whether limiting their business success to this one platform is the best idea. Companies that depend entirely on Google stand to lose up to 30 percent of their business following a ‘routine update’ such as this, so it may not be the best business approach.
Google’s new ‘humanistic’ approach to SEO is the first time the tech giant has dipped into the revolutionary concept that amalgamates deep learning and AI, but it certainly won’t be the end of it. The digital landscape is set to be completed uprooted by the deep learning revolution, and we have been seeing small hints of what may come since 2012, when Google created its largest neural network yet using 16,000 computer processors and more than one billion connections. Over the course of three days, the algorithm it created successfully began to identify cat images across the internet with accuracy. It was a test to see how accurately they could do so in the hope that Google could do the same with its search query function. Obviously, it worked.
But this wasn’t the first time Google’s search function was used. The basic online search function ‘Google search’ was developed almost 15 years prior to this, in 1997, when Larry Page, Sergey Brin, and Scott Hassan developed a function that would look for text in publicly accessible documents offered by web servers, as opposed to other data, such as images. Since, Google has come to dominate the search listings field, overtaking its competitor Yahoo – which, as opposed to Google’s approach of using an algorithm to direct people seeking answers to the correct pages, simply orders search results according to who pays them the most. Google’s entire business model focuses on its algorithm – everything else is secondary. So it’s no small thing that Google have gone and changed their algorithm, with an end goal of enhancing all round user experience.
For the average, ‘untechnological’ person this is an extremely hard concept to grasp, that of how Google went from having an idea that people should be able to use a keypad to search for questions, to then have a machine sift through billions of webpages to find the most relevant content. But it is just the beginning of brilliant things machines are capable of being designed to do.
The future of SEO is exciting. Sooner than we can imagine, it is expected that we will be using voice search more regularly than typing our search query, black hat tactics will become obsolete, and voice will become integrated with visual data to create a hybrid search process. Will brands begin investing in understanding Google’s algorithms rather than developing their brand? Quite possibly. The future of digital is in understanding SEO, and brands would be well advised to focus their efforts on doing just this.