Results from marketing data company Sistrix suggest reasons why Google’s March 2019 Core Algorithm Update feels like a rollback. The data also provides an explanation as to why many publishers consider this a minor update despite reports that Google has stated it is one of the biggest updates in years.
March 2019 Core Algorithm Update Feels Like a Rollback
Brett Tabke, founder of WebmasterWorld and PubCon, was given advance notice of the update. He was informed that this update would be among the biggest in years.
When asked for his thoughts on the update so far, Tabke indicated that it seemed like a rollback of previous algorithms. This was based on numerous reports of previously penalized websites regaining traffic and search engine results page (SERP) positions, as if a previous update had been rolled back.
Brett Tabke’s observation:
"I think we may be seeing a rollback of a few of the last updates."
Why Google’s Update Feels Like a Rollback
The data Sistrix examined was based on UK winners and losers rather than USA data. Nevertheless, the information provides insight into why the update feels like a rollback.
A notable finding from Sistrix’s data shows that 75% of the winners were previous losers. This means that 75% of the websites that improved in rankings with this update had lost rankings in the 2018 updates.
Because many previous losers appear to be winning, it creates the impression of a rollback. However, I don’t believe Google rolls back updates. From conversations with search engineers, it’s understood that Google fine-tunes their algorithms.
I believe major updates are intended to improve how sites are ranked, with the side effect of positively affecting sites that previously lost rankings.
Although this may look like a rollback, it’s highly unlikely. Whatever changes were made appear like a rollback.
If 75% of the winners consist of losers from previous updates, then Brett Tabke’s observation holds some truth. The March 2019 Google update looks like a rollback, but it likely is not.
Anecdotal evidence and actual data from Sistrix suggest that up to 75% of the sites that improved in rankings were sites that lost rankings in previous updates. This gives the update the impression of a rollback.
Why the March 2019 Update Feels Minor
Sistrix noted that their data indicated sites that lost rankings tended to lose long-tail positions rather than major ones. This suggests that the amount of traffic associated with ranking loss was relatively softer compared to losing more critical rankings with higher traffic impact.
This observation aligns with anecdotal evidence that this doesn’t feel like a major update.
Importantly, 70% of the sites that were losers had been affected by previous algorithms. If this data is accurate and applies to other regions, most of the damage was to sites that had already lost ranking positions. This may contribute to the perception that this update isn’t substantial.
Sistrix’s data observation:
"The top losers in our data aren’t as strongly hit (in terms of percentage) as the winners. … among the losers there are many domains (70%) that were affected by previous core updates."
What is the March 2019 Update?
With numerous advances in information retrieval technology over the past year, pinpointing the exact nature of the update is challenging. It could involve one or more advancements.
For instance, Google recently published a research paper titled, "Non-delusional Q-learning and value-iteration."
Fixing Bias in Reinforcement Learning
This paper highlights potential bias in “reinforcement learning,” a core aspect of machine learning.
Excerpt from the Google research paper:
"We identify a fundamental source of error in Q-learning and other forms of dynamic programming with function approximation. Delusional bias arises when the approximation architecture limits the class of expressible greedy policies. …inconsistent or even conflicting Q-value estimates can result, leading to pathological behavior such as over/under-estimation, instability and even divergence."
I’m not suggesting Google has revealed a more accurate form of machine learning. This paper is simply one example of many Google research papers that might provide clues about the update.
A New Relevance Signal for Ranking
Another research paper introduces a new method for ranking web pages called "Learning Groupwise Scoring Functions Using Deep Neural Networks."
Proposal from the research paper:
Consider a search scenario where a user is searching for a name of a musical artist. If all the results for the query (e.g., Calvin Harris) are recent, the user may be interested in the latest news or tour information.
On the other hand, if most query results are older (e.g., Frank Sinatra), it’s more likely the user wants to learn about artist discography or biography. Thus, the relevance of each document depends on the distribution of the whole list. Also, user interaction with search results shows strong comparison patterns.
This means the age of relevant web pages to a search query might provide context regarding the user’s intent. Additionally, historical user preferences in SERPs can support the relevance of document age.
Ranking Algorithms Evolve
I’m not asserting that these research papers are responsible for the recent Google Update. Rather, they illustrate the current state of the art. Many people still adhere to the notion of 200 ranking factors and view updates as "targeting" low-quality sites. This understanding of Google updates is outdated.
As demonstrated by the above examples, Google’s algorithm is far more complex than just 200 ranking signals.
Google Explanation of the March 2019 Update
In the future, Google may choose to explain what was introduced. If the change resembles the algorithms discussed above, Google might choose to obscure specific details and focus on the algorithmic results.
Observation of affected sites can offer insights but drawing conclusions based on these observations is often misleading. The so-called Medic Update was named because it impacted many medical-related sites.
Yet, Google wasn’t targeting medical sites. Unfortunately, erroneous analysis led to confusion, with the false idea that Google targeted medical sites persisting. This highlights the dangers of drawing conclusions based on limited observations. The mislabeling of the Medic Update has been a significant source of confusion.
It’s prudent to await information from Google, find parallels with patents and research, and then start unpacking what happened. This is the most reliable approach to understanding Google updates.
Sistrix’s data provides a good explanation for why Google’s update feels like a rollback to many publishers. The data is not evidence of a rollback, but it clarifies why it feels that way. I do not believe this update is a rollback.
