What Is a Broad Core Algorithm Update from Google?

on

After reading this article, you’ll have a greater understanding of what a core update is intended to do, as well as how to recover if your rankings have been harmed.

When Google announces a major algorithm update, many SEO consultants are left wondering what actually happened (besides their rankings).

Google’s acknowledgement of core upgrades is always hazy, offering nothing more than the fact that the update happened.

Core upgrades are often communicated to the SEO community via the same regular tweets from Google’s Search Liaison.

Google will make one statement when the upgrade goes live and another when it’s finished, with little information in between (if any).

This inevitably leaves SEO specialists and site owners with a slew of queries about how the core change affected their results.

Understanding what a wide core update is and how it varies from other sorts of algorithm updates will help you figure out why a site’s rankings went up, down, or stayed the same.

After reading this article, you’ll have a better understanding of what a core update is intended to do and how to recover if your rankings have been harmed.

What Is A Core Update, Exactly?
First and first, let me state unequivocally that Google makes hundreds of algorithm changes every year, frequently more than one per day.

Many of the well-known Google algorithm changes (Penguin, Panda, Pigeon, Fred, and so on) are designed to solve specific flaws or weaknesses in the company’s algorithms.

It was link spam in the case of Penguin, and local SEO spam in the case of Pigeon.

They were all there for a definite reason.

In many occasions, Google (sometimes unwillingly) told us what they were attempting to achieve or avoid with the algorithm upgrade, and we were able to fix our sites.

A core upgrade isn’t like that.

A core update, in my understanding, is a modification or alteration to the main search algorithm itself.

You know, the one with 200-500 ranking variables and signals (depending on whatever SEO blog you’re reading right now).

To me, a core update indicates that Google changed the priority, order, weights, or values of these signals somewhat.

As a result, they are unable to tell us what has changed without exposing the secret sauce.

Imagine 200 things listed in order of significance as the easiest way to see this.

Imagine Google rearranging 42 of the 200 variables in a different sequence.

Rankings might shift, but it would be due to a variety of factors rather than a single element or cause.

Obviously, it’s not so easy, but it’s an excellent method to consider a fundamental upgrade.

Here’s a made-up, somewhat more sophisticated example of what Google refused to tell us:

“In this core update, we increased the value of keywords in H1 tags by 2%, increased the value of HTTPS by 18%, decreased the value of keywords in title tags by 9%, changed the D value in our PageRank calculation from.85 to.70, and began using the TF-iDUF retrieval method for logged in users instead of the traditional TF-PDF method.”

(These are actual things, I swear.) I’m just not sure whether they’re genuine Google products.)

For starters, many SEO experts would be baffled by it.

Essentially, this indicates that Google may have altered the method they assess word relevance on a page, or the way they measure links in PageRank, or both, or a whole host of other criteria that they can’t discuss (without giving away the algorithm).

Simply said, Google modified the weight and significance of a number of ranking variables.

That’s all there is to it.

Google passed a fresh training set through their machine learning ranking model, and quality raters chose this new set of results as more relevant than the prior batch, and the engineers have no idea what weights changed or how they changed since that’s how machine learning works.

(As we all know, quality raters are used by Google to rank search results.) These ratings are used to determine one algorithm update to implement over another, not to rank your website. It’s anyone’s guess if this will be fed into machine learning. However, it is a possibility.)