Google’s neural matching: Why it matters and how to optimise for it

Blog banner

Last September, Google’s Danny Sullivan told us the search giant had started using a new AI technology called neural matching to connect search terms with concepts. At the time, Sullivan said neural matching was impacting 30% of all search queries – so who knows how much this has increased since then.

Neural matching is clearly playing a crucial role in where Google is heading and this means marketers need to get familiar with the technology and its implementation.

What is Google’s neural matching algorithm?

In many cases when people turn to Google Search, the query they type in doesn’t match the keywords of the content they’re looking for. This is particularly common when there’s a technical issue with devices and software or when people are trying to diagnose a medical condition.

People don’t automatically know the name of every condition; instead, they type in the symptoms. Or when something seems weird about their TV, but they’re not sure exactly what, Google has to try and match queries like “why does my tv look strange” to an issue known as the “soap opera effect”.

screen shot of someone searching for 'why does my tv look strange'


Matching such a vague search query to a very specific technical issue is a real breakthrough for Google and shows how effective its AI technology is becoming.

How is Google doing this?

Our guess is that Google is looking at engagement metrics for vague, ambiguous and keyword-less search queries on a mass scale. When millions of people search “why does my tv look strange”, Google can pinpoint the right topic by ruling out content that consistently gets fast bounces for that search term, suggesting it doesn’t cover the right topic.

Pages that do cover the right topic will keep people engaged with the content and on the pages for longer.

There’s probably a lot more to it than that but the key thing to understand about neural matching is that Google is using it to match topics/concepts to search terms.

How is Google’s neural matching different from RankBrain?

Unfortunately, a lot of people writing about neural matching right now aren’t explaining what it does very concisely and this is leading to some confusion. We’re seeing a lot of questions asking how neural matching is different from RankBrain – another AI technology Google rolled out in 2016.

Well, Google has a pretty good explanation of this:

How to optimise for neural matching

Most of the content we’ve seen covering neural matching over the past six months or so says there’s no way to optimise for it. This is something Google is doing to improve the way it handles search queries and all you can do is create the best possible content – a sentiment echoed by Google itself.

But whilst there’s not much you can do to optimise for neural matching on a technical level, there are plenty of things you can do on a strategic level.

1. Look beyond keywords

Neural matching exists because a lot of searches (at least 30%) don’t match the keywords used in the content people are looking for. Most people don’t know what “the soap opera effect” is and optimising for this keyword is going to have limited results. However, optimising for this topic is going to get results with neural matching, regardless of which keywords you optimise for.

The challenge for marketers is to pinpoint the topics their target audiences are searching for when the obvious keywords aren’t there. If you’re only using keyword tools to pick your content marketing topics, you’re probably missing out on a lot of potential traffic.

So how do you pinpoint those topics?

2. Focus on the problems your target audiences face

Based on the information Google has given us about neural matching over the past nine months, it appears to be most active when users have a problem they don’t know how to describe. They’re typing in medical symptoms instead of specific conditions. They’re describing the sound their car is making rather than typing in the name of specific problems. Or they’re using incredibly vague search terms like “why does my tv look strange”.

To capture these opportunities, you need to know what problems your target audiences are facing. You also need to pinpoint what information is going to solve their problem, help them accomplish tasks and make decisions.

3. Create the best content for each topic

Increasingly, Google Search is becoming an environment where one piece of content wins the race. This is a symptom of the mobile age and new search technologies such as featured snippets. When Google says you should focus on creating great content, it really means you need to be creating the best content for any given keyword/topic these days.

search in google for 'my car is making a funny sound'

With neural matching, the focus is on topics rather than keywords but the same rules apply when it comes to making your content visible on Google Search. Featured snippets are particularly common for symptom-orientated searches and this means there’s often only one spot up for grabs – especially on mobile.

If neural matching was affecting 30% of all searches in September last year, there’s every chance this number has increased. In 2015, Google said RankBrain was used for less than 15% of all queries, but a year later Danny Sullivan confirmed it was now being used for every query.

This is a healthy progression though. Google is becoming less dependent on keywords and getting better at matching content to search queries. This opens up a lot of new opportunities for marketers and actually makes the optimisation process slightly easier – as long as you’re able to pinpoint those opportunities, create the right kind of content and deliver it in the right place.

Want to chat?

If you’d like to speak to one of our search marketing experts, don’t hesitate to call us on 02392 830281 or submit your details here and we’ll call you.

Lee Wilson profile picture
Lee Wilson

Lee has been working in the online arena, leading digital departments since the early 2000s, and oversees all our delivery services at Vertical Leap, having joined back in 2010. Lee joined our company Operations Team in May 2019. Before working at Vertical Leap, Lee completed a degree in Business Management & Communications at Winchester University, headed up the online development and direct marketing department for an international financial services company for ~7 years, and set up/run a limited company providing website design, development and digital marketing solutions. Lee had his first solely authored industry book (Tactical SEO) published in 2016, with 2 further industry books being published in 2019, and can be seen regularly expert contributing to industry websites including State of Digital, Search Engine Journal, The Drum, plus many others. Lee has a passion for management in the digital industry and loves to see the progression of others through personal learning, training and development. Outside the office he looks to help others while challenging himself, having skydived, bungie jumped and abseiled (despite a fear of heights) with many more fundraising and voluntary events completed and on the horizon. As a husband and dad, Lee loves to spend time with his family and friends. His hobbies include exercising, trying new experiences, eating out, playing countless team sports, as well as watching films (Gangster movies in particular – “forget about it”).

More articles by Lee
Related articles
data science doodle

How to apply data science to your SEO data

By George Karapalidis
blue background with question mark

SEO FAQs – 7 questions our customers often ask

By Sally Newman
hedge maze

Getting your structured data right is key to successful SEO

By George Stone