Let's dig into all the hullabaloo over Google's RankBrain over the past month.
I solemnly affirm not to use the words vector, entity, string, database, or semantic in any form or variation in this post. And oh, I will not advise you to implement schema on your website.
That promise out of the way, let’s dig into all the hullabaloo over Google’s RankBrain over the past month – it was revealed exactly a month back
– and get the gist filed away in our minds, once and for all.
If you don’t want to read all the 5000-word expert analyses, or are baffled by the mumbo jumbo, or simply don’t give two hoots about Bloomberg and co.’s business journalism, this post will tell you the following about RankBrain:
- What it is
- What it does
- What you need to know
- What it means for you
- What you need to do
By “you” here, I mean a marketer. If you’re a small business owner, you could be your own marketer. If you run a large business or are a developer (or a doctor who RankBrain failed and led here), just don’t bother.
If, however, you happen to be feeling geeky, here’s some preliminary reading:
Still with me? Read on for the real lowdown on RankBrain.
What it is and what it does
Leaving out the interpretations, these are the three exact sentences (emphasis mine) that Google’s Greg Corrado stated, as revealed by Bloomberg, which broke the news.
Read them very carefully, a few times.
For the past few months, a “very large fraction” of the millions of queries a second that people type into the company’s search engine have been interpreted by an artificial intelligence system, nicknamed RankBrain.
RankBrain is one of the “hundreds” of signals that go into an algorithm that determines what results appear on a Google search page and where they are ranked.
In the few months it has been deployed, RankBrain has become the third-most important signal contributing to the result of a search query.
First, it’s a system.
Then, it is a signal.
In effect, RankBrain is a set of principles and procedures that conveys instructions to the Google search algorithm. For what? More on that later.
Here’s the meat of the story.
RankBrain is NOT a ranking factor.
But it is a factor in the sense that it contributes to the result of the query, if not that of the algorithm.
This is how it works:
- The user types in a query.
- RankBrain, the “system,” analyzes the query, and if Google doesn’t understand what it means, tries to make sense of it.
- The algorithm picks “relevant” results based on RankBrain’s analysis. RankBrain doesn’t pick the results. RankBrain, the “signal,” helps the algorithm choose a result (that it otherwise may have ignored) to display, based on its relevance to the query.
- The algorithm orders the results by relevance. RankBrain doesn’t order or rank the results.
The SEO community isn’t very clear on this, although the experts agree that the lies lie in the semantics. (Yes, I said “semantics.” Promises, like links, can be broken).
What you need to know
Thinking deeply about something is a very hard job. And interpreting what others (especially Google) say is harder.
If you know enough to understand the terms, you’ll know enough to understand the implications.
To put things straight,
- A “link” is a signal that helps the “algorithm” order the results by relevance for a search query. A link contributes to “PageRank,” which is a ranking factor.
- “RankBrain” is a system that holistically acts as a signal to help the algorithm determine relevance. It is not a ranking factor.
- “Location” is another signal that helps the algorithm determine relevance. “Proximity” is the associated ranking factor.
That was about the jargon. Now, about the PR. Google is a company that exists to make profits. Hence its new-found love for business publications. WSJ and BBC covered Mobilegeddon (the SEO equivalent of Y2K, FUD gone dud). Consequently, someone at Google might have thought, “Why not hit up Bloomberg with a story that tells all and reveals nothing?”
And so, they shone the limelight on their work on AI technology, which everyone knew they were working on.
Which, as expected and known, isn’t as good as the human brain yet. But trust a tech company (and publication) to turn a negative into a positive.
So far, RankBrain is living up to its AI hype. Google search engineers, who spend their days crafting the algorithms that underpin the search software, were asked to eyeball some pages and guess which they thought Google’s search engine technology would rank on top. While the humans guessed correctly 70 percent of the time, RankBrain had an 80 percent success rate.
What? So that’s RankBrain’s fault, not the humans’! When you’re asking a person to “guess” what a machine would do, they expect the machine to be correct and efficient. And in estimating, they use their own brains. Clearly, Google’s existing search engine technology is 42.85% off the mark.
RankBrain, to its credit, is leaning more towards the humans than the current engine. And Google is now confident enough to show it off to the press (and public), given that it has become, over a span of a few months, the third-most important signal.
That’s not the whole deal. RankBrain is particularly good at dealing with the 15% of queries that Google has never seen before. However, what about the other 85%?
People also search for what they have seen before but can’t remember where. RankBrain has no idea of this.
Some “key articles” according to you, Glenn! See what I meant about what humans think is a better result?
For example, I’ve been routinely asked by clients (of that kind) – “Do I need to follow Google’s Quality Guidelines?” (Long-tail alert)
While the matter has been broached – and done to death – hundreds of times on Twitter, forums, emails, and even UX blogs, Google is sitting smug with these results:
However, it wouldn’t be wise to ask “If Google can’t guess properly, why guess at all?” That’s not how progress is made. Google has made great strides in eliminating spam and dodgy results from its pages.
One day, we’ll all get there.
But then, what matters to us will be someplace else.
People have “cracked” search results by manipulating links. You think they can’t manipulate words? You think a machine will have a better way with words than a person? Not happening anytime soon!
If you know enough to understand the past (history) and present (context), you’ll know enough to understand the future (implications).
What it means for you and what you need to do
Absolutely nothing, says Ian Lurie, who is mostly right about most things SEO.
Get back to work.
Read More On Digital Doughnut