Explanation of the Helpful Content System’s quality rating system is provided by SearchLiaison.
A misleading section in Google SearchLiaison’s Helpful Content System guidelines that might have unintentionally caused problems for publishers who were not at fault was clarified.
Beneficial Content Platform
The foundation of Google’s Helpful Content System is a machine learning model that use classifiers to produce a signal that Google’s ranking algorithm utilizes to filter out inferior content.
An algorithm in a machine learning model that gives an input a label is called a classifier. The machine learning model in the context of the Helpful material System is labeling website material, which in turn provides a signal, such as a thumbs-down.
Additionally, that signal is weighted, so a website with little to no harmful content has a lower thumbs-down rating than a website with a lot of harmful information, which would receive a higher thumbs-down rating.
One of hundreds or thousands of signals (such as links, relevancy, etc.) that are utilized to rank a website is produced by the Helpful Content System.
Google Advice Inadvertently Made Opaque
To help provide clarity to this signal and help publishers and SEOs understand why sites lost rankings, Google changed its Helpful Content System guidelines during the most recent Helpful Content System.
The adjective “opaque” describes anything as lacking in transparency or clarity. Sadly, there was one section of that advice that appeared to be inadvertently unclear and perplexing as a result.
This is the relevant passage:
“Are you rescheduling pages to appear more recent even though the content hasn’t changed significantly?”
The section targets certain individuals who try to manipulate Google’s freshness algorithm by altering the publishing date after making a very little modification to the content, giving the impression that the previously published page has been updated.
However, the issue is that a lot of users revisit websites and make little edits to the text to:
To make it more grammatically correct or understandable, add or remove a word.
Swap out terms to improve clarity of the message
A lot of people make numerous minor, valid updates to articles.
A little enhancement now had the potential to lead to a negative assessment by the Helpful Content System because of the instructions that seemed to forbid making minor adjustments that result in date changes.
This is exactly the issue flagged on X (formerly Twitter).
Luke Jordan (@lr_jordan) expressed this legitimate worry:
“Google doesn’t have enough context knowledge to establish general guidelines.”
Websites that use a “last updated” date for “small” updates are penalized.
However, a patch or update in a game could be as easy as changing the price of an upgrade from 5 points to 6 points.
Furthermore, the slight increase might significantly alter how useful it is.
Users will consult the date and patch number because they want to be sure the content is current and pertinent.
It might be necessary to change the patch number from 9.0.1 to 9.0.2 and the number 6 to 5, depending on how valuable the upgrade is.
It makes no sense if the date indicates that the guide was last updated six months ago.
Additionally, since the (very out-of-current) date appears in Google results, less people would click on it. CTR is another ranking consideration.
Naturally, they might just act as though they comprehend everything, and helping a lot will always be the best course of action.
Google SearchLiaison made a reply.
“No, we don’t do this if the upgrades are intended to benefit users.
Not anything we express.
Not as per our policies.
Although SearchLiaison is accurate, it does seem to say what Luke Jordan says it implies due to the opacity of that one passage.
Luke continued, saying:
“So, just to be sure, do you know if an article that changes just one character is meant to be helpful for people?”
One more post from Luke, along with a screenshot of the relevant section in the guidelines:
“because it is expressly stated in your guidelines that you should not alter the date of pages if the content has not changed significantly.”
In reply, SearchLiaison said:
“Those questions are asking if you work for Google in any capacity.
If you’re simply updating the date because you believe “that will make Google think this is fresh,” you’re probably following other practices that generally match the signals we use to determine if a piece of content is helpful.
There is more than one issue. It’s not explicit.
Furthermore, it doesn’t matter if your main goal isn’t Google.
Complementing Other Conduct
According to SearchLiaison, the date change strategy is merely one of several strategies the machine learning model employs to determine the statistical likelihood that the webpage is using SEO strategies for Google rather than making an effort to provide informative and valuable content.
There is a statistical property that causes the statistical model to make poor choices when one metric is used in isolation.
For this reason, it is well known in statistical models pertaining to search that combining many signals to compute the statistical probability yields a more accurate result than utilizing a single signal (metric).
If you’re unfamiliar with this, view this PDF of a statistical spam identification system that classifies webpages as spam or not based on a variety of factors, including user engagement metrics, off-page content, and on-page content.
Without putting words in SearchLiaison’s statement, it appears they are suggesting that when there are no other negative signals present, doing one action that might be a sign of unhelpfulness is insufficient to label the webpage as such.
As stated by SearchLiaison:
If you’re only altering the date because you believe “that’ll make Google think this is fresh,” you’re probably following additional actions that, taken together, match the signals we use to determine whether a piece of content is helpful.
It’s fortunate that SearchLiaison made this clarification because, in my opinion, the section was unduly general and may generate false positives, which occur when a legitimate website is mistakenly flagged as spam.