Machine learning and artificial intelligence are terms that any carbon-based humanoid operating in 2017 has probably heard or discussed in one way or another. What was the stuff of science fiction two decades ago is now science, and it is present everywhere.
Machine Learning – type of artificial intelligence enabling computers to learn without the explicit programming to do so.
It is essentially the evolution of programming, whereby millions of formulae and functions are converged to work in synchronicity and analyse masses of data. Almost like adding a trend line to a data set, on top of a data set, on top of another data set, ad infinitum.
I say evolution, as when we look at the exponential increases we’ve had in computing power and the information generation, the processes we use to crunch the numbers had to keep up and could no longer depend on teams of (now obsolete) human brains.
Without getting too bogged down with the 'what ifs?' of an omnipresent cyber intelligence in our lives, there has been some discussion about how machine learning (ML) and artificial intelligence (AI) can affect digital marketing, especially when it comes to organic search.
Skynet Engine Optimization?
If Penguin, Panda, and Possum are more than just charismatic creatures to you, then you’ll know Google has been experimenting with machine learning for many years already. To put it simply, they silo sections of the data collected by their web spiders, analyse it for good and bad practices, and update their search algorithm to favour the good and punish the bad.
So what can SEOs & webmasters do to futureproof themselves against this digital Judge Dredd of organic search?
The short answer is... nothing you weren’t doing already.
That is, focusing on creating a quality website which serves a clear purpose to your intended audience, in terms of information, experience and association.
Here’s how Google sees them:
Information: Google looks for quality content, so the amount and variety per page, whether it is duplicate or not, each page serves a specific purpose, and the user stays and digests it. It uses analytics such as: sessions; pages per session; average session duration and bounce rate to gauge how valuable that result is for a particular search term. If users are going to another site to find the information, they will naturally start ranking higher for said search term. Given the emerging popularity of voice (natural) search, look to optimize content for these types of queries.
Experience: Sites that are built with good user experience (UX) in mind would get favoured. Meaning less invasive advertising, quicker loading, clear call to action and logical navigation would all send positive signals to Google, and keep users coming back. My Gravytrain colleague Jo offered his advice on the interplay between UX and SEO in this recent column for The Drum.
Association: Google also looks at where your site sits within the infinite landscape of cyberspace. If it is snug among its industry chums (read – linked to from similar sites) and maybe has the odd pat on the back from an online authority, chances are Google will see it as relevant and increase the ranking of your keywords. On the other hand, if all your sites’ associations (read – links) come from unrelated, unknown, spammy or paid link sites, Google will not think that highly of you, hence ignore you when it comes to getting a spot on The SERP.
Personalised search results
All search engines, albeit Google, Amazon, Bing, Facebook or Spotify are using machine learning one way or another, and all have a similar objective – to provide you with the most relevant results for you.
To do so, they gather information about you based on previous searches, activities, location, interactions and conversations, then analyse this data for trends, which act as a kind of filter on future searches in the hope that the results they deliver will be more satisfying, as they are more catered to your preferences. I am sure many of you will have noticed how the ads you see all over your internet are scarily similar to your recent searches. That’s machine learning getting fat on all the internet cookies you provide it.
Enter arguments about invasions of privacy and personal information. But are personalized search results based on your preferences really a bad thing?
I, for one, welcome our AI overlords.
Guy Swanborough is an SEO account manager at Gravytrain