You better have your intelligent head on, because you’ll need it to synergize with this post and stay on the sweet spot of the SEO curve.
Notice I say the “sweet spot” because you don’t want to be too far behind, e.g. spamming keywords (which used to work), nor too far ahead because you don’t know which direction search will take in the future. Get it wrong and you risk dropping off any and all indexes.
I don’t mean you will have to use all your 180 IQ points while you read, I’m going to lay this out in a way that the average, switched-on human can fully understand.
Today we’re going to focus in on artificial intelligence (AI from now on) and how recent advancements in implementation have radically altered the way SEO works on the most cutting edge search engines.
The use of AI in search engine technology is not particularly new. The very simplest idea of counting keywords in a document to score its relevance is actually a form of AI.
Now we have come much further and the game has changed. Let me bring you up to speed.
Google’s RankBrain is one of the newer AI algorithm’s currently deployed by our old friend the big G.
It is far more advanced than a simple keyword counter and backlink checker. This system actually uses a similar method to “think” about a website as does a human brain.
RankBrain relies heavily on neural networks.
AI has two categories: strong and weak. Strong AI is something that you might see in a science fiction movie such as Ex Machina, 2001 An Odyssey, or JARVIS in Iron Man. These have intelligence capabilities that approach and even surpass human abilities.
Weak AI is pretty much what we have in real life. Everything from IBM’s Watson to contextual advertisements on web pages.
No surprise then that RankBrain falls into the camp of weak AI, but it is several steps closer to the “strong” category than say, a very good chess playing program.
Since Google’s public announcement early 2014 that they were going to buy the highly specialist AI company company DeepMind, we can deduce the types of technology that they are integrating into their platform.
It’s about 95% certain that backpropagating neural networks are being integrated into Google’s search process, as that is what DeepMind were particularly good at, producing almost scarily impressive machine learning results from their labs and think tanks.
If you are not trained in the black arts of AI, you may well be wondering what in the sweet blue and purple is a “backpropagating neural network”?
A simplified explanation is as follows:
A neural network takes at least one input and either gives an output, or not, based on whether the total value of the inputs reaches a threshold value (a predetermined value of some amount, whether energetic or numeric).
Bear with me, we’ll get to the relevance to SEO very shortly.
With backpropagation, an expected output can be compared with the actual output. If the actual output is not what is expected, the values between the input and threshold are adjusted, from output back to input, for the next pass.
Read that last paragraph again carefully. When the concept clicks in your mind, you will see how the machine “learns” new information. Your own brain will have also gone through a very similar process.
What it means for SEO
It means the way that current SEO experts have been evaluating Google’s algorithm changes has become almost redundant.
So now every time Google looks at a site, it has learned something new about it. Either how good the information is within the content, or how it relates to changes to all the other content in the same field.
RankBrain also constantly learns about particular search results.
The kind of information taken into account by the intelligent search engines is not a blanket view as may have been the case in the past. By that I mean that not every search will be judged by the same criteria.
For example, traditional SEO advice has been to optimize keywords, have great content and get as many high quality backlinks as possible, no matter what your site is about.
Sound advice indeed, but now the search engine may not give a damn about your backlinks, depending on the specific search carried out by the user, and where your content sits within their ecosystem.
On another type of site and/or search, backlinks may be more important than ever, but keyword density not so much.
Again this depends on what the back propagating algorithm has learned about that specific search, and possibly even who is doing the searching.
Google never got rid of their pre-update algorithms
They have certainly maintained and updated those earlier, core algorithms that count keyword density, inbound links, read META titles, tags and descriptions, etc, etc.
RankBrain uses all those “old” algorithms, plus the newer developed ones, on a case-by-case, search-by-search basis.
To stay ahead of the game, you will need to know how Google ranks your site, according to a particular search term, so that you know where to best invest resources.
This will mean a lot of testing and observing results, as well as studying where high ranking competing web sites and pages are optimizing, in order to glean some insight where possible.
If you are doing SEO, you will need to literally dissect these sites at every level to try and figure out what Google finds important for your target keyword.
A short term optimization that makes sense within this relatively new search paradigm, is to dig in deeper within your niche, and do not diverge from it.
RankBrain will be learning constantly about your niche and already has a profile of what it wants to put at the top of Google’s results pages for your target keyword.
Study what’s already there, emulate the consistencies that you find and make improvements wherever possible.
Competition for search engine rank has become orders of magnitude more focused.
This specialization extends to your backlink profile. RankBrain is certainly checking out what type of backlinks are attached to top quality websites and you will need to emulate that profile to remain competitive.
The general advice here is to keep your backlinks topic related as well, so as not to be seen stepping out of line according to RankBrain’s expectations.
Where do we go from here?
RankBrain and similar machine learning algorithms represent a quantum leap in search engine evolution. Will we reach the point of hyper-intelligent machines functioning alongside us with god-like computing power? At this point, almost anything seems possible.
SEO is becoming more and more technical, specific and specialist. Vagueries and generalities in SEO are mere shots in the dark nowadays. SEO practitioners are going to have to swot up on the technical nature of artificial intelligence in order to stay relevant and competitive.
Search terms and keywords must now be analyzed in isolation of one another.
Narrow field, niche sites are the logical choice to stay above water for the time being.
Those who can navigate this new environment will be in great demand. The computer scientists win again.