What Does Google’s Machine Learning Algorithm Mean For Marketers

In their efforts to improve search, Google has developed a new piece of the Hummingbird algorithm which uses artificial intelligence to understand human speech patterns.

It’s called RankBrain.

According to Google, the search engine giant receives over 470 million unique and ambiguous searches on a daily basis. This accounts for 15 per cent of their search traffic. The Hummingbird algorithm was struggling to answer ambiguous searches, typically longtail keywords that use more than four words. RankBrain resolves this problem by recognising associated words.

Although the search engine giant officially announced the algorithm update a couple of weeks ago, RankBrain has been working in the background for several months now. According to reports, the new cog outperformed engineers in user-testing exercises, accurately determining 80% of user intent compared with developers who were only correct 70% of the time.

How can artificial intelligence improve search?

Artificial intelligent (AI) computer systems are designed to mimic the cognitive function of the human brain. Programmed to understand natural speech patterns, search engines have the capacity to be more intuitive.

Mainstream tech companies have been working on developing AI for some time now. Voice recognition technology has improved vastly since developers were able to get the software to recognise accents and dialects. And it is this type of machine learning RankBrain brings to Hummingbird. As a result voice search on mobile handsets will become much more powerful over the next few years.

For example, if you wanted to know if Leonardo DiCaprio is making a new movie, you could speak into your handset and ask the question. Google will not only tell you his new movie, but all his other films. This is standard. But with RankBrain, you can ask “what is it about?” and Hummingbird would intuitively know by using ‘it’ you mean the movie. This intuitive function could change the way consumers search for products, content, entertainment and information dramatically.

So what can content marketers do to ensure their business has visibility in search results?

How does Hummingbird work?

Before you can create engaging content that will rank in search engines, it is important to understand how Google’s Hummingbird algorithm works – otherwise you will be stuck in the past using keywords and keyword density.

To give you the brief version, Hummingbird can understand the context of page content and intuitively matches search terms with content based on how keywords are used with other words. For example, when you search for a word that has more than one meaning, Google will give you options to choose from. So if you type in “Who is Isis” Google will bring up information about the terrorist organisation. However, if you type in “Who was Isis”, the results will show websites featuring information about the mother goddess of ancient Egypt.

In the not-too-distant future, we should expect to find search results options become more refined based on the user-behaviour of search engine users. Therefore, based on the first search above “Who is Isis”, if you have been researching ancient Egypt or mythology, you are more likely to get results on the goddess Isis as opposed to ISIS in current affairs.

Another example is the differentiation of words from different countries. For example, a thong in the UK is a sexy undergarment whilst in Australia it is a type of flip-flop. Therefore if you are searching for thongs in the UK you will get different search results than if you search for thongs in Australia.

How will RankBrain affect content marketing?

If you are writing great content and incorporating longtail search terms already, RankBrain shouldn’t have too much of an impact on your content marketing campaign. The principle purpose of RankBrain is purely to resolve the problem Google was having with ambiguous searches. As far as the Hummingbird algorithm is concerned, it is producing better quality content in search results. Therefore publishers can expect to find their content performing well if it is engaging enough across multiple platforms. And this is where the game-changer comes in.

Marketers need to think about how they create content, the platform the content will be hosted on, and the type of device that is most likely to be used to access the content. With this in mind you should be thinking about producing different types of content and publishing them at strategic times of the day. For example, during commuting hours in the morning and evening, most people will be using mobile handsets, possibly lunchtimes as well. During work hours, mid-evening and late at night, more people will be using laptops and PC’s – thus larger screens. For mobile hours, create content that is easy to digest – short articles, useful tips, news pieces and video ads. However, the crux here is that persistent short pieces of content will not rank well in search engines. There is an unwritten rule, but anything under 250 words is considered “thin content” which search engines consider light on information. If you continuously publish content that has a short word count on your blog, search engines will not rate your site highly.

Varied content

Search engines want content curators to publish detailed content that is unique, engaging and offer value to readers. But the only way they can quantify these metrics is by the word count and user-engagement. Studies have shown that search engines favour long form content, typically 2000-10,000 words. Therefore, marketers need to think about creating in-depth content . Yet this goes against the trend of an increasing number of end-users browsing the web with their mobile phones. We are in limbo.

Remember earlier when we mentioned voice recognition? For anybody that has seen the movie Her with Joachim Phoenix, this is the direction search is going. We are not at the stage of interacting with our devices and operating systems just yet of course, but we are already starting to see how voice search is impacting the way people conduct search. And because semantic search algorithms recognise natural speech patterns, search terms have a broader meaning thus search engines need more information to decipher user intent. And this is where long form content excels. There is more on-page information for search engines to work with to match content with intent, especially when the search term is unique – which natural speech search will be more often.

The future of search

RankBrain is already having a slight impact on search results, and is ranked by Google as the third most important ranking factor in their search algorithm. And its potential for learning will mean it has a greater impact on search in the future. Search technology examines user-behaviour and reader engagement is used as a premise for ranking a webpage. So, if an end-user clicks on a link in search results, but leaves almost immediately and tries another link from the same results, Google knows the first link that was tried does not provide sufficient information against the search terms. The web page will be downgraded. So it is becoming more important to write information-filled articles that link back to products and services.

Blog posts should be single-point focused so the content strictly relates to the article title without deviating into other areas that are loosely associated. If you have a product you can discuss at length, semantic search capabilities will benefit long-tail search terms. So if you are writing a blog post to promote a revamped iPhone, discuss what has been revamped and how this is an improvement on the original. Don’t have a filler section promoting iPhone accessories at the bottom to fill out the word count. The reason being is that anybody searching for iPhone accessories wants information quickly. They do not want to scroll through five paragraphs of text before they can read about accessories – thus will click straight out.

You should also check your bounceback rates in Google Analytics and address any content that has a high percentage of people leaving immediately. The update to the Hummingbird algorithm emphasises the need for quality content. And RankBrain is the next cog in the wheel towards the evolution of voice search and the need for varied content that is accessible from multiple devices.

Author: Jonathan Gillespie