There was a time when AI was only seen as science fiction material, but advancements in technology have made it quite real.
It's quickly becoming an essential part of daily life and businesses are not lagging in adapting it to their marketing strategies. The last week of October marks the breakthrough news of Google's RankBrain, the exact point where AI and SEO meet.
What is RankBrain?
It's a machine learning system that works alongside current algorithm factors to help to determine the best results for a particular search query.
Machine learning is where a computer holds the ability to teach itself how to do something, instead of following detailed programming or instructions laid out and taught by humans.
In line with Google's aim to serve better search results to user search queries, which there are currently an average of 3.5 billion per day, is RankBrain.
Google regard RankBrain as one of the most important among hundreds of ranking signals that go into their algorithm. Signals are things Google uses for determining how to rank various web pages.
The company says that RankBrain is the third most important factor that help denote the ranking, though the first and second most important signals remain untold.
They have consistently spoken of about more than 200 major ranking signals, which can be attributed to more than hundreds since they might have up to thousands of variations or sub signals.
Google's RankBrain seemingly correlates to query processing and refinement, whereby it utilises pattern recognition to accommodate complex and ambiguous search query sets.
This capability allows RankBrain to display search results based on what a person who has executed the search is actually looking for, rather than exhibit results based solely on keywords – exactly what SEO revolves around.
The keywords themselves are primarily assessed to put reasoning behind the search, considering other searches performed to yield targeted results. Hence, users will be connected to specific topics where they are bound to find exactly what they need and want.
Google is indeed pushing forward in the changes they are implementing, but RankBrain manages to make a big difference and stands out among previous changes as it works with significant enhancements on entity search.
This is not the sole major change to search, which has went through algorithm updates to the page layout itself. The Panda update was a major evolution, while the Penguin update established link-building further, and the Hummingbird that altogether shook the world of SEO.
Let's take a look at the quick timeline of the few important changes Google has made on how a search works in the past years.
The Google Panda traces its roots in 2011, and the latest update primarily filters out sites with bad content. It identifies low quality content with precision, which favourably rewards websites containing high quality content by enhancing their rankings.
The Penguin launched in April 2012 and revolved in the idea of putting the search engine focus on quality links vs quantity. Websites were then focused on getting incoming links predominantly from sites which have a similar theme or subject as their own site.
The Pigeon was an algorithm launched in July 2014 to provide more useful, relevant, and accurate local search results. They appeared to be closely tied to traditional web search ranking signals which improves Google's ranking parameters on distance and location.
These changes did not stop at search, but its structure was likewise altered with the Alphabet umbrella. So many changes occurred in a very short time.
From Strings To Things
Entity Search has evolved from finding "strings" pertaining to the strings of letters within a search query, to finding "things" pertaining to entities. The Knowledge Graph which launched in the year 2012 paved the way in which Google learned how to search for things and not strings.
Hummingbird, the search platform Google launched in September 2013, changed the future of SEO. It was derived from a name that meant precise and fast.
In addition to Knowledge Graph-based facts, it was designed to focus more on the meaning behind the words.
Each word is paid attention to, where the whole sentence and entire query is taken into account, rather than focus only on specific words. As part of the update, RankBrain was introduced, again revolutionising Search Engine Optimisation.
The Hummingbird algorithm was part of Google's effort to incorporate semantic search into its search engine. It sought to improve accuracy of search through understanding a searcher's intent and the contextual meaning of terms.
Aside from that, it considered a myriad of points such as location, variation of words, synonyms, concept matching, took both generalised and specialised queries, and worked on understanding and processing natural language (NLP).
Still, after two years, many users deem that Google seems to fall short of expectations. It remained a navigational search rather than semantic in nature.
This is where machine learning comes in or the New SEO arises. Google's RankBrain is designed to refine the query results of the entity search with a stellar AI. In turn, SEO is no longer what it used to be.
It’s now about designing a website that caters to both user experience and SEO at the same time. You get filtered relevant results while irrelevant items are filtered out. No doubt, this will give you a much more favourable experience as a user, but what about businesses?
How Does RankBrain Affect Search Traffic?
RankBrain makes SEO more crucial, but entails an effective SEO strategy incorporated into a website's content.
Google has long been moulding their algorithm to focus on serving results made up of quality content that's readable for humans, and not made for search engines.
By quality content, it means those that are built around what consumers are really looking for and what RankBrain is definitely looking for: informative, useful, helpful, related content people search for every day.
Companies, brands, and business owners would not find this any cause of concern for these should be the type of content they would already have.
How Does RankBrain Work?
To be clear, RankBrain is part of Google’s overall search algorithm and not in any way the new way it ranks search results. The ultimate goal is to present users with the right content, regardless if they entered the most appropriate keyword or not during their search.
RankBrain utilises artificial intelligence to embed vast amounts of written language into vectors or mathematical entities that the computer should be able to understand.
Entities, for us, refer to nouns in general. It may be persons, places, things, or ideas which have their meanings defined in Google's databases which serve as references. Entity search does not rely on understanding the nouns, but try to analyse how they are related.
A fresh example Google provided was by entering the question "How many tablespoons in a cup?"
This gave different results depending on the country it was searched in, where it is given that they have different corresponding measurements. This was attained despite their similar names and exactly the same words entered during the search.
Another example of a search is "What’s the title of the consumer at the highest level of a food chain" where consumer can mean different things.
A consumer may be someone who buys products, while it can also be a scientific term for something that consumes food. Thereby, the name for that would be a predator, which is what Google has provided good answers for.
Google transforms words that appear on a page into entities which mean something and possess related attributes. This is what the human brain would naturally do, and it's what we call Artificial Intelligence on computers.
When RankBrain encounters a word or phrase that it is not familiar with, it would then make a guess on which words probably have a similar meaning to filter the results.
This makes it more effective in handling unknown queries. It does all its learning offline, sourced from historical searches, and can surprisingly learn from its mistakes. It can then come up with predictions based on these learning information to provide more accurate results in future searches.
A Glimpse of The Future
Since RankBrain can translate an ambiguous search into more specific terms, it can then deliver results that are most likely precise.
Even when entering different entities in queries that a computer probably wouldn't understand, Google can leverage all it knows to gather and provide an accurate answer.
During tests, Google engineers who design the algorithms were able to correctly rank 70% of sites coming from a range of search terms. RankBrain outperforms them and actually achieved a score of 80%.
Though there is no specific figure, Google said that a very large fraction of queries are processed by RankBrain. What's certain is that it holds a component where it somehow directly contributes to whether a page ranks.
Also, they were initially unable to produce a semantic engine so they built one based on facts, which is not bad at all. In its few months of deployment, it has showed a promising future.
Over time, it will undoubtedly get better upon all the learning of new entities and discovering the relationships between them. Sure enough, it will soon present better results than what it can give you today.