RankBrain and Neural Matching: What SEO-conscious Law Firms Should Know About Google’s New AI

RankBrain and Neural Matching: What SEO-conscious Law Firms Should Know About Google’s New AI

Technology is evolving. Artificial intelligence systems that use machine learning, where a computer can interpret data and learn from it without further programming by humans, are gradually being implemented into our daily lives. And the search engines we use are no different. In the last few years, Google has begun to introduce AI into its algorithm to help bring its users more relevant search results than ever before.

The two AI systems Google uses are RankBrain and Neural Matching, both of which allow Google to better understand search queries by relating them to concepts, enabling them to therefore return the most relevant results.

The question is, how do these two systems impact upon the algorithm and what implications does this have for the SEO strategies of law firms? Furthermore, what should SEO agencies be doing to help optimise their clients’ sites with these AI systems in mind?

RankBrain and Neural Matching: What’s the Difference?

RankBrain and Neural Matching tend to go hand-in-hand and outwardly appear extremely similar, but there are key differences. In a recent tweet, Google explains the difference like this: RankBrain helps Google better relate pages to concepts, while Neural Matching helps Google better relate words to searches.

How Neural Matching Works

The language people use when they search will not always match the exact wording businesses use when writing information for their websites. Neural Matching is a way of working out what users actually want to search for – the user’s true intent – to enable the serving up of the most relevant content.

Described by Google Liaison Danny Sullivan as a “big change in search”, Neural Matching is basically, as Google says, a “super synonym system” (synonyms being words closely related in meaning to other words) that allows the search engine to better understand and interpret language.

The system matches words to concepts – or, in other words, it attempts to identify what the user is searching for even if they don’t use a specific keyword phrase. If the searcher doesn’t know a specific term and instead types another description of it, Neural Matching will be able to work out what they mean. For instance, someone may want information on a prenuptial agreement, but be unaware of the correct term and search instead for “signing an agreement before marriage”. Although they’ve not used the exact keyword phrase “prenuptial agreement”, Google will still understand what their intention is and return results relating to that topic.

In September 2018, Neural Matching affected 30% of all queries. However, with Google’s March 2019 core update, it is likely this percentage has increased.

How RankBrain Works

A significant proportion of the billions of daily searches occurring on Google have never been searched for by anyone else before (an article from 2013 had the percentage at 15%). The search engine needs to work hard to provide these searchers with the most relevant and informative pages, and part of RankBrain’s purpose is to make sure that happens.

Google confirmed its use of RankBrain in October 2015. In an interview from that year, Google said it was a factor in less than 15% of search queries (used just for the queries that had never been searched for before). However, in June 2016 it was confirmed that Google uses RankBrain for every single search. Of course, Google uses hundreds of ranking signals (characteristics of a website that the search engine algorithm considers when calculating webpage ranking) to help bring its users the most relevant results, which is why RankBrain attracted a lot of attention among SEO experts when the company acknowledged that RankBrain is the third most important signal used in search, behind high-quality content and the number of backlinks.

But how exactly does it work?

Here’s a good way to think about it: when humans communicate with each other, we can attempt to understand what someone means even if they’re not using precise language or the correct terminology. Up until the introduction of RankBrain and similar AI systems, it was not possible for a search engine to emulate how our brains work to understand meaning; a search algorithm could not interpret a vague or poorly worded search term (perhaps one using colloquial language), take a guess at what the user was trying to say, and then present them with the most useful pages of information. If the user tried to search the web using a phrase that the engine had not seen before, the results delivered were bound to be inaccurate and unlikely to provide the best answer to the query. However, RankBrain is able to interpret language and learn from previous searches and user behaviour on SERPs, understanding how a unique search is similar to one encountered before so that it can work out the intent behind it.

Much is still not understood about RankBrain, and Google is never overly willing to share too much information about its algorithm, but in a Reddit AMA in February, Google’s webmaster analyst Gary Illyes helped clear up some common misconceptions. Illyes described the AI as a “ranking component that uses historical search data to predict what a user would most likely click on for a previously unseen query”.

So, if someone searches for a complex query consisting of a multi-word, niche keyword phrase (known as a long-tail query) that no one else has ever searched for, RankBrain will detect how this is connected to other search queries for similar topics it has seen before and look at the results associated with those. Grouping together different queries in this way, even if the precise wording of them may seem unconnected, allows Google to match the unknown query to what it calculates as the most pertinent webpages. The machine learning system will continually collect data on user interactions with search results to help refine this process and improve the match between user intent to the results displayed.

This ability to better understand user intent will also help Google recognise the meaning of negative words such as “not” and “without” in searches. Previously, these words would be ignored, leading to search results providing the exact opposite information required. As the AI detects the context of a search, it should also help differentiate between homographs (words that have the same spelling and pronunciation but have two completely different meanings).

Optimising for RankBrain and Neural Matching

The key question remains, however: how do you optimise for RankBrain and Neural Matching? Many SEO professionals have attempted to speculate on the best way to optimise for the AI, but the short answer is that there is really no concrete way of optimising effectively for RankBrain, and it’s best to spend time instead making sure the content on your site stands above the competition.

It’s important to stress, despite Google calling it a ranking signal and prioritising RankBrain in its algorithm, there is no specific RankBrain score that you can increase by optimising your site, (it’s not similar to something like Moz’s Domain Authority). Neural Matching and RankBrain are there to increase the algorithm’s ability to understand the user’s search intent and to evaluate the content on webpages to help refine the content served up, but it will not change the rankings on SERPs.

From what Google has said, the bottom line appears to be this: RankBrain and Neural Matching have made it more necessary for websites to include high-quality, engaging content that flows naturally, and there should be extra emphasis on the word “naturally”. Speaking to the search engine news site The SEM Post, Gary Illyes said, “Optimising for RankBrain is actually super easy and it is something we’ve probably been saying for fifteen years now… to write in natural language. Try to write content that sounds human.” Illyes went on to say that writing “like a machine” is likely to confuse RankBrain and push you further down in the rankings.

An increased importance on natural language means it’s not necessary to try to squeeze keywords into your content for the sake of SEO. Google will be able to understand the intent of the content using the language that’s there. Besides, unnaturally fitting in a keyword phrase – such as “prenuptial agreement solicitors in London” – has long been considered a poor technique, because ‘keyword stuffing’ means that not only will your content quality likely suffer by having repetitive wording shoved into unnecessary places, but you’re also in danger of having Google class your content as spam, which could harm your rankings much more than the extra keywords will help.

A system that helps to better understand the searcher’s intent is good news for law firms, as it should help detect all the law-related searches where the user’s intention is to connect with a lawyer, even if the specific wording of the query is in some way ambiguous and they are not using a popular or industry-specific keyword phrase. The way that these AI ranking systems affect the algorithm will hopefully encourage webmasters to focus more on creating content with the user’s intent in mind, making it informative, engaging, and as useful to visitors as possible. At BORN, this has always been our primary focus when writing content.

BORN: Helping Law Firms Improve Their SEO

BORN has years of experience in SEO for law firms. Our teams of skilled content writers, SEO specialists, and website designers will work together to ensure the top search engines recognise your digital presence. Our services have helped many clients operating in the legal sector boost their business. Find out how we can help your law firm by contacting us today.

No Comments

Post A Comment