The internet, and more importantly how we consume data from the web, has evolved at an incredible pace in recent years. One thing that has been steadily growing, and is only now really starting to make the headlines is Machine Learning.

Artificial intelligence would be the ultimate version of Google. The ultimate search engine that would understand everything on the Web. It would understand exactly what you wanted, and it would give you the right thing. We’re nowhere near doing that now. However, we can get incrementally closer to that, and that is basically what we work on.” – Larry Page, October 2000

Larry Page, co-founder of Google and now CEO of Alphabet Inc. gave the above quote in the year 2000.

Let that sink in for a minute – SIXTEEN years ago.

That’s seven years before the first iPhone was released, six years before Google was added to the Oxford English Dictionary as a verb, five years before YouTube even existed, four years before the launch of Facebook, three years before Skype, two years before the LinkedIn domain was even created and one year before we had the sheer wonder of Wikipedia.

Whilst a lot has come and gone (RIP Geocities), Google’s focus on “incrementally” moving toward an “ultimate version” of its core search product through artificial intelligence remains steadfast, if not stronger than ever.

In order to understand just how much of a role Artificial Intelligence plays at Google, we first need to take a step up the corporate ladder to Google’s parent company, Alphabet Inc.

Artificial Intelligence makes the mainstream

We humans have been more than capable of programming computers to follow the rules of chess for quite some time.

In 1997, IBM’s supercomputer Deep Blue famously beat chess Grandmaster Gary Kasparov, becoming the first computer to defeat a reigning world champion.

The ancient Chinese game of Go, is simultaneously “simpler” than Chess, using non-moving identical pieces, and it is also far more “complex”. In any game of Chess there are only 20 possible opening moves, compared to Go’s 361 possibilities.

The total number of game possibilities is just too large to compute; there are more Go positions than atoms in the universe, making Go’s need for more humanlike thinking far more interesting to those in the field of artificial intelligence.

Just last month we saw AlphaGo, a product of Alphabet Inc.’s DeepMind unit, beat the world’s top Go player, Lee Sedol, 4 games to 1. Before AlphaGo was revealed, experts in the field estimated that the type of ability we saw it demonstrate was still somewhere in the region of a decade away from reality.

AlphaGo’s ability comes in the form of a “neural network”, a program that somewhat mimics the way in which the human brain is structured.

There’s a hell of a lot going on behind the scenes with AlphaGo, far too much to cover in this article, but it starts to become very clear that Alphabet Inc. aren’t interested in developing these incredible supercomputers just to beat us at board games.

These supercomputers don’t just hold clever algorithms, designed with just one purpose in mind. Instead they have the ability to learn by themselves from experience, learning not just one or two, but a multitude of tasks, often those which a human would typically find difficult, at least to perform at scale.

So what does this mean for SEO?

As the world of Go has AlphaGo, the world of digital search has RankBrain. In an interview with Bloomberg, Greg Corrado, a Google senior researcher, referred to RankBrain as the “third most important signal contributing to the result of a search query”.

What is RankBrain?

RankBrain is Google’s codename for a facet of their search algorithm that uses artificial intelligence to extract meaning, context and semantics from search queries, ultimately marrying these queries up to search results it deems as highly relevant (but might not actually contain the originating search query).

Without the jargon, RankBrain thinks somewhat like a human and has been reported to outperform trained and qualified Google specialists in identifying the “quality” of a page.

Imagine a human brain, with a photographic memory of billions of websites and the ability to remember which websites were popular with users after they typed in a given search query; that’s RankBrain.

RankBrain looks at your search query, “understands” it, and then uses this understanding to find results that are likely to be useful to you.

What other ranking factors are important?

Last week, Search Quality Senior Strategist at Google, Andrey Lipattsev, was asked as part of a live Q&A, if RankBrain is the third most important ranking factor, what are the first two? His answer? Links and content, although he wouldn’t identify which of the two occupied the number one spot (If you’re keen to find out a little more about link building and content strategies, don’t hesitate to get in touch).

What does this mean in practice?

OK, so we’ve established that on page content, the links that point to it, and a super smart, humanlike algorithm should all play a part, in theory, in bringing users to content that perfectly matches their needs.

Does this change anything in the world of SEO?
Not in my opinion.

Yes, the Google search algorithm now has the power of Artificial Intelligence behind it (and has done for some time if truth be told), but that doesn’t mean it’s reason to go crazy.

Remember, Google are especially keen on driving users to the most relevant content for their search query. Take Silverbean for example. There’s us, the world famous digital marketing agency, and Silverbean Coffee, manufacturers of what seems to be a really swanky industrial coffee machine.

Pre-RankBrain, the guys over at Silverbean Coffee probably received a little bit of web traffic from people using search queries, that might overlap between our two industries, when they were really looking for us (and vice versa). Post-RankBrain (or more precisely, post Hummingbird) Google search would be far more likely to return the right Silverbean for the right search query.

I say forget about the how of RankBrain – What should it matter to us in the world of digital marketing how Google interprets search results?

To me, whether they use artificial intelligence or a WiFi enabled Sorting Hat, it should be of no concern to us.

I mean, of course, the technology is fascinating, and as good as magic anyway, but I think focusing on the how over the why is folly. Whilst the how might be fantastically complex, the why is phenomenally simple, match users and their search queries to the content most likely to satisfy their needs.

What does this mean for my current content?

Start out with a little exercise. Sign into your Google Search Console account (haven’t got an account? Shame on you!) and take a look at the queries that are bringing you impressions and clicks.

See any irrelevant queries in there?

For instance, you could be selling widgets in Newcastle Upon Tyne and seeing search queries that look like they might be from people looking for widgets in Newcastle, Australia. If so, you’re going to lose that traffic shortly, and rightfully so. Your content will never be relevant for those queries.

Pop into your Google Analytics account and look at those pages which are attracting users but have very low engagement (low time on page, high bounce rate). Do you think those pages are going to be displayed in search results over competitors with more engaging pages?

No such luck.

At Silverbean, we’re huge proponents of avoiding the trap of Pointless Marketing. You need to recognise that every time someone enters a search query and visits your site, they’re exhibiting countless signals as to whether you should be relevant for that search term again in the future.

You may very well have built up years of authority through your impressive backlink profile and domain strength, but don’t expect that to continue if you’re missing the “secret sauce”- giving users what they want.

I can’t wrap up how to do this in a nutshell, as of course it depends entirely on your industry, your product range, your purpose, your company’s’ tone of voice, and your customer base. Don’t take this lightly. If your online presence doesn’t marry up with the means by which you should be satisfying customers, get back to the drawing board.

There’s no quick fix for this!

What about the future of machine learning?

I’d like to make a little prediction about machine learning.

Google Search as we know it will continue to evolve at an alarming pace. Give it a few years and the Search Results page as we currently know it will be long gone.

If you’ve read this far then you’ve already exhibited some signals to Google that this is an engaging piece of content (hopefully!). Did you skim a section? Scroll past it at a much faster pace than the rest? Then you’ve possibly demonstrated that particular section didn’t represent what you were expecting to get out of this article.

I’d like to propose a future vision of Google search whereby your results are perhaps a single page made up of aspects of content that have been proven to be engaging.

Imagine a version of Wikipedia that, instead of being written by a community of human “experts”, was “written” by machines that scoured the “web” for the most accurate, relevant, and authoritative voices on any given subject, compiling them into one easily digestible form.

Of course, this future content isn’t static. It evolves and grows and becomes expertly tailored to you, the individual, as those “trust signals” you exhibit become ever greater and from a wider range of sources.

Much like how social sharing now demonstrates the popularity of content, it’s not impossible to imagine a future where your mobile phone’s microphone listens to you talking about this content the next day at work, further demonstrating the authority that a given piece of content deserves.

I firmly believe that human authors, writing engaging content for other humans, must always form the bedrock on which the technology, and the content presentation, rests. No greater proof exists, at least in 2016, than Tay, Microsoft’s Artificially Inept twitter bot (but I’ll leave it to the BBC to explain this full fiasco!).

UPDATE: On the 5th of April 2016, Persado announced that they would generated $30m in venture capital funding to support their new platform. They are claiming that their machine learning based system generates the “precise combination of words, phrases, and images that can help to motivate any audience, at scale and in real time”.

We’ll reserve judgement as to whether this really lives up to it’s claims, or whether it’s just a smarter, more dynamic A/B test.

Let me know how you think machine learning will affect the future of SEO in the comments box below, or over on twitter @Silverbean or @stephen1986. I’d love to hear your thoughts and opinion!