One SEO professional was wondering about how the Google algorithm is trained. More specifically, they were wondering about how Google’s algorithm could be trained to understand whether there were real changes, or if things were being tested.
Is machine learning used? Or are human reviewers used?
John answered that for Search, at least for these types of general ranking updates, it’s not something where they have human reviews in the loop.
There is so much content out there, they are unable to review them all. Furthermore, many times, they also don’t have a clear yes or no understanding where somebody can go through a website and say “Oh, this page is a good one. Or this page is a bad one.” Something like this is just not scalable at all.
So these are all, essentially, algorithmic approaches that Google takes. And they do use a lot of machine learning.
It’s difficult to say, though, exactly which technology is used for which updates.
The SEO professional proceeded to ask if Google intends to integrate this kind of understanding into the core Google algorithm?
John explained that it’s always tricky to define exactly what is part of the core update, or the core algorithm.
He doesn’t really have an answer for that, because what usually happens is—when Google sees that an algorithm is working really well, they will then keep using that algorithm for a longer period of time.
At some point, you can call that as being “part of the core algorithm,” even if it’s not in a file called “core algorithm.”
This happens at approximately the 47:17 mark in the video.
John Mueller Hangout Transcript
SEO Professional 8 47:17
Okay, because those were very nice upgrades. And we’re looking forward to having them in Europe, I think. I’m curious to just learn a little more,in general, how could Google train like an algorithm to understand whether there was real testing or not? Is it machine learning? Or is it also even human reviewers?
John 47:38
For Search, for these kinds of general ranking updates, it’s not something where we would have human reviewers in the loop. Because like, there’s so much content out there, we can’t review at all. And a lot of times, we also don’t have kind of like a clear yes or no understanding where someone can go through a website and say, Oh, this page is good, this page is bad. It’s just not scalable. So these are all essentially algorithmic approaches that we take. And we do use a lot of machine learning, whether we use that for this particular update, I don’t know. My guess is probably also, at least to some extent. But it’s hard to say which exact technology was used for which update.
SEO Professional 8 48:22
Okay. And do you tend to integrate this kind of understanding of a web page into the core Google algorithm?
John 48:33
I mean, it’s always tricky to kind of define what is a part of the core update—the core algorithm. So I don’t think I have an answer for that—because usually what happens is, when we see that some algorithm is working really well, then we’ll kind of like, just keep using that for a longer period of time. And at some point, you could say, well, it’s part of the core algorithm, even if it’s not in a file that is called “core algorithm.”
SEO Professional 8 49:05
So I have a small case study where basically like for example, a website upgraded like 20 comparison articles, two comparison articles with testing. And like you could clearly see like an article that are on the -website bots, the stuff tested for free time. And so it was like one year ago, and unfortunately like it’s been one year and there hasn’t been any noticeable positive evolution for these specific articles. So it’s just that, because of that’s…not been integrated into the algorithm yet, I’m guessing.
John 49:44
I don’t know. Is it the case that they have some articles that are good and some articles that are kind of like old style or…?
SEO Professional 8 49:53
Yeah, yeah.
John 49:56
I could imagine that kind of scenario throwing things off because for some kinds of understanding of a website, we do look at the bigger picture. And sometimes that bigger picture goes so far as kind of like the website overall. And if we have this mixed set of like, these are really good. And these are kind of okay or not that good, then kind of what averages out, it’s really hard to estimate.
SEO Professional 8 50:21
Okay, I understand. Okay. Okay. And so long term quality should win, of course. But do you have any tips on what can websites do like short term to, to do because you know, long term, you can work on upgrading new stuff in it, but short term…?
John 50:37
Short term, I don’t know, it’s, it’s always tricky, because there’s like some things that move very quickly in Search and some things that take a longer period of time. What I would always recommend for sites in general is to make sure that they’re not purely kind of like, only focused on Search, to make sure that they have a diverse set of traffic to the website. And usually, what you’ll see is also that if you significantly improve the quality of a website overall, you’ll also see better traffic from social media and kind of like the different other channels that you use to drive traffic. And that usually kind of makes different channels a lot more stable on the one hand, but also means that if you make bigger changes, you’ll see some of these channels reacting faster, some of them taking a little bit longer to kind of like update on that. So that’s kind of like, almost like saying, Well, you shouldn’t use search for these kinds of fast changes. But I think that’s kind of the reality as a business, you, you try to cater to a wide variety of options.
SEO Professional 8 51:45
That’s all for me. Thanks.
