SEO Watch: Google and the rise of AI
Nick Garner from 90 Digital on how artificial intelligence is revolutionising search marketing
Last issue, I talked about SEO in the context of product ï¬rst, in particular, prioritising user experience. In this article, I want to touch on something higher level but strategically just as important: the rise of Google and artiï¬cial intelligence and what it means to all of us.
In SEO, what gets most attention are algorithm updates where something happens with Google, rankings jiggle about, people scream because theyâve lost rankings, other people feel lucky because their rankings go up and generally there is a lot of noise. But underneath this, there is a tide in Google and that’s its move towards artiï¬cial intelligence.
101 guide to artificial intelligence
The principles of artiï¬cial intelligence have been around since Greek times, with mythical machines that could think. By the 19th and 20th centuries, artiï¬cial beings were often used as characters in novels such as Frankenstein, for example. But the jump from a scary monster like Frankenstein to what could be an even scarier monster began in 1956 at a conference at Dartmouth College in Hanover, New Hampshire.
In its most simple terms, artiï¬cial intelligence emulates what we can do; taking information from several sources and making a balanced decision. In another sense, artiï¬cial intelligence allows computers to think for themselves and to exponentially grow in knowledge and capability over time. Their only constraint would be the hardware they sit on and the quality of information they can obtain. I liken it to an infant learning and ultimately becoming self-sustaining.
Google has a problem. It prevails across a vast data ecosystem, with some of the largest data sets ever seen and with this data it has to make decisions on your behalf in order to give you search results that are most relevant to you. If this doesnât happen, Google dies. Its success to date has been largely based on algorithms built by engineers. In very simple terms an algorithm is a set of rules ie, if âthisâ, then âthatâ. A good example of an algorithmic rule might be;
⢠if this keyword
⢠in that page
⢠then make sure itâs visible
⢠for this keyword
⢠in our search index
Imagine you have hundreds and hundreds of rules like this all working together, thatâs a big algorithm. Strangely enough, when you break algorithms down into their constituent parts theyâre pretty straightforward. I know this, because Iâve been building quite a few of them recently.
Compared to artiï¬cial intelligence, man-made algorithms are not that powerful. This is why Google has put so much eï¬ort into developing artiï¬cial intelligence across its businesses. Whether one needs a driverless car, to understand the content of an image, do better voice-to-text recognition or decide which search result goes ï¬rst, it all needs something to make the right decisions.
RankBrain â AI now
Out of nowhere on 26 October 2015, Google told us it had an algorithm that is now the third most important ranking factor there is, behind links and words on a page. RankBrain is part of a bigger search algorithm called Hummingbird, which also contains components such as Panda (quality of content), Penguin (quality of links) and the Payday loan update (identiï¬cation of aggressive SEO).
If I said to you, âParis⦠Hiltonâ, what would you think? A person or a hotel? This kind of ambiguity is a huge problem for Google and having âif this, then thatâ rules just donât deal with ambiguous questions like âParis Hilton?â Thatâs where artiï¬cial intelligence comes in. Google trains its AI algorithms and the programmed artiï¬cial intelligence takes over and progressively learns more and more about the context of words, so it can clearly deï¬ne the meaning of ambiguous terms.
RankBrain is just one part of a universe of artiï¬cial intelligence algorithms Google is working with. Another one appears to be around site quality. How do we know this? Because on 19 November 2015 Google released Quality Rater Guidelines 2015, which is a training manual for quality raters who are training Googleâs artiï¬cial intelligence algorithms. These are people who look at websites and search results, make a judgement on these sites and search results, and input evaluations into a system.
Artiï¬cial intelligence, like any life form, needs training. And as we know, if you train something well it will go on to evolve around its original trajectory. Quality rater guidelines are in many ways the most authentic and comprehensive description of exactly what Google wants in a good website. There is no PR spin, or dumbing down to accommodate beginners. The content is just straightforward, to the point and absolutely fascinating.
Itâs like an instruction manual for a software program or a car, but these instructions are the blueprint of how to build a site users really want. If Google trains its algorithms to look for sites users do not want, the consequences would be disastrous. Thatâs why this document for quality raters is the blueprint for site quality.
Taking action
We all move away from pain, right? What does pain mean to you in the context of SEO? Does it mean rankings are falling and therefore your return on investment is dwindling? Maybe your conversion rates arenât great, so youâre having to spend more and more money on acquiring new customers? For anyone running an online business this is truly painful.
Google feels pain too. Its pain point is making sure the best sites rank. You might think: âWhy isnât my site ranking better?â Google will be thinking: âWhy do we get so much spam in our search results?â Whatâs the answer to less pain for everybody? Simple. Build websites people want. If you do that, Google will want you to rank and do whatever it can to rank you.
Iâve seen so many cases where websites are running as an application. In other words itâs a one-page website which has virtually zero SEO optimisation, and they rank in competitive keyword neighbourhoods within egaming. That doesnât make sense, until you remember that Google wants sites to rank that users want. Ultimately, therefore the answer is to build a website people want.
Where would you start? How do you know what people want? Well, fortunately, Google’s Quality Rater Guidelines 2015 describes every general characteristic of what a good site should be.
The principal idea in the document is pretty simple: (I repeat myself) âgive users what they want.â But what do users actually want? They want something that satisï¬es them. Google uses the term âsatisï¬esâ a lot. And itâs a great phrase because something thatâs satisfying leaves you content. What would satisfy a user going to a casino site? Would it be site speed? Mobile usability? Perhaps product recommendations that were good? If itâs a sports betting site, maybe its navigation? Or helpful information in a contextualised form?
Customer satisfaction
Finding out what âsatisï¬esâ a user comes from objectively looking at what you do. You know what you want from a sports betting site, a casino, poker room and if you stop to think about what you like and dislike, youâll get closer to understanding what satisï¬es you when you visit an egaming website. Think about âobjective observationâ. If you look at things exactly as they are, without personal bias getting in the way of the truth, you will begin building a site that users really want.
A strange thing has emerged over the past two years: sites people want in relation to particular search queries are ranking. Ironically by forgetting about SEO, focusing on what users really want and getting your site out there, with a strong product, generates a huge number of engagement signals that Google will pick up on. If you socialise your website, you will get really good links and if you do sensible SEO housekeeping, all these elements come together and you will rank.
Itâs not about âshortcuts, tools, tips and tricksâ, but knowing where Google is going with artiï¬cial intelligence. It will hopefully help you come to terms with the new reality in SEO â Google cannot be gamed like it used to be. The irony is, if you concentrate on product and user experience, ROI goes up because your retention rates increase and advocacy for your brand increases. And you get a double bonus: Google will give you tons of free traffic because it needs you in its search results.