Scientific research proves Google Core Updates Algorithm is of Technical Nature

This article is a follow up on the article on how to fix the losses after the last Google Core Update (GCU) and the strong correlation between technical SEO and the lost rankings.

At the same time, we compiled a lot of data on the effects of the Helpful Content Update (HCU, which happened in September 2023) and both March and August 2023 Google Core Updates (GCU). So we decided to do methodological research on the reasons behind the changes to understand what’s going on with Google Search algorithms.

Just to be on the same page: to us, Technical SEO is more than just speed, headers, and meta-tags. For us, it includes coding, best practices, accessibility, technologies used, QA, privacy, security, etc. In other words: whether a website works as it should or just… works.

Research Methodology

Like I said, we used proven scientific methods and methodologies. In this particular case, we used a methodology named Comparative Analysis. For reference, the main characteristics are:

Observation and Analysis: It implies observing and analyzing different parameters. In this specific case, the code, SEO implementations, and strategies used by different websites.

Benchmarking: Benchmarking is basically a quantitative comparison method. It simply compares the performance of website A, website B, website C…. website N in order to extract quantitative data.

Pattern Recognition: As the name implies, it basically means recognizing and identifying patterns. In other words, we used comparative analysis to identify successful SEO implementations or uncover issues on those sites with problems.

Sample Selection Criteria

We analyzed websites for our clients as well as websites posted on different SEO groups where the owners complained about traffic losses. Another criterion was that the owner had to blame that loss on any cause but technical SEO. This was done in order to have a more similar sample universe since we didn’t want websites where the owner was already aware of our hypothesis.

Finally, losses had to be on websites already ranked and the loss had to be of at least 50% of the traffic. It sounds like a lot, but this is what is actually happening.

On top of the above, we contacted a few colleagues to provide us with more information under a strict NDA (Non-Disclosure Agreement).

This approach allowed us to reach a very important number of websites, 113 to be precise.

Google Core Update First Insights

As we already hypothesized in the article mentioned above, Technical SEO is the culprit for many of the losses in rank, traffic, and/or complete demotion. We were curious to prove this hypothesis and the extent of it.

But first, let’s see how webmasters and website owners identified the problems and what they thought the main reason was. Figures exceed the total of analyzed websites because in most cases there were multiple causes.

AI content: 42 websites

Almost one-third of the websites blamed the loss on a penalization due to the use of AI-generated content, even though Google publicly announced this wasn’t an issue. Nevertheless, with the emergence of AI generative technologies and its cultural importance, it’s easy to understand why this reasoning has some merit.

Backlinks: 37 websites

The second most mentioned reason was backlinks. Either wondering about the quality of backlinks or about the possibility of competitors getting better backlinks. This is an interesting take because neither of these reasons would explain complete demotions or websites being sent to the 100th page; in any case, websites would lose a few positions.

On-Page SEO: 36 websites

Here we’re talking about considering their on-page SEO as the problem. This overlaps with the first item (AI-generated) because the main issue was content quality, headers, intent, etc.

Ads: 29 websites

In this case, website owners hypothesized that a large number of ads was the main reason. One interesting fact is that 21 of them used interstitial ads or intrusive pop-ups, which is a known issue in user experience, so this one holds some merit and deserves further research.

User Experience: 17 websites

In this case, our obvious biases tempted us to make this play a big role (after all, everything falls under the user experience umbrella). However, we only considered those websites where the owners explicitly mentioned “poor user experience”, “poor UX” or “UI/UX issues” as a factor.

CMS: 8 websites

Finally, some webmasters expressed their concern about their choice of CMS or static code, and whether this could be a factor. The main concern was if using a CMS could have outranked them in favor of custom-coded websites.

Unknown causes: 83 websites

Of course, the main cause by many webmasters and site owners was “I don’t know,” which makes sense, nobody knows something without a previous diagnosis. Since this is unknown, I mention this in last place.

You can check the chart below for a quick visual reference on how webmasters perceived Google Core Update’s impact on their sites.

The graph above shows the possible reasons attributed by webmasters to explain their traffic loss.

Website build-up

Another interesting factor for us was where websites were built, which technologies were used, and whether these websites were built by the site owners, an established development agency, a freelancer, or some other option. Keep in mind that due to the nature of this research, we couldn’t find the full info on this since we didn’t have access to many of the webmasters.

An interesting fact is that websites with issues were custom coded in an unusual proportion. I mention this because WordPress powers 40%+ of existing websites, and probably around 70% if we consider all CMSs. Therefore, we should be seeing figures of around 80 websites using CMS and 33 using no CMS or using custom-coded CMSs.

However, despite what one may expect (not us, because we already had some data and expected what finally happened), the reality is quite different: only 12 sites used WordPress, 5 Wix, 3 Shopify, 2 Squarespace, 2 Bubble.io, Custom Coded 39, undisclosed 43. Regarding the last ones, we can only say they weren’t built in WordPress, Wix, Shopify, or Bubble.io since they have easily identifiable “digital marks” (structure, comments, name conventions, etc). While we are sure they’re mostly JS or HTML-based static websites, it is possible they’re using some kind of uncommon CMSs.

One important discovery: of those 39 sites using static code, custom CMSs, or JS libraries, 19 of them used ReactJS. That means almost 50% of this group of sites and almost 60% more than WordPress. Considering that the real figures based on their market share (40% and 4.2%) should be around 45 for WordPress and a bit less than 5 for React, this means around 320% overrepresentation for ReactJS. This has a logical explanation, but it is still very important information once we get down to conclusions.

GCU implications on traffic loss

As we already mentioned, none of the site owners blamed technical SEO. This doesn’t mean nobody did; a few webmasters actually did, but none of this group, for the simple reason that they wouldn’t be part of this research. We didn’t vet a website because of content issues since we didn’t know what we’d find. The important part of this research was the owners and webmasters’ perception of the problem.

Since we’re focusing on GCU (Google CORE update), most of the research comes from these updates, but it’s worth noting that around 60% of the webmasters blamed the 2023 Helpful Content Update (HCU) as the main culprit or in part. Basically, the descriptions were different variations of “it all started with the HCU last year, then it continued on March’s Core Update, and now my traffic vanished with the August Core Update.”

Now, we need to mention something: an HCU is completely different from a GCU; one is for content, the other is for everything (it may include content as well). So that is very important information since it already gives us some initial analysis: if a site was hit by the HCU and then by the GCUs, it clearly means the main issue was content and the issue was never resolved. While the GCU may have found more issues, the root of the problem was in the HCU.

Again, we didn’t have all the REAL data (namely, Google Search Console and Google Analytics) for all cases we studied. Fortunately, we had access to real data for a majority of cases (whether as guests or screen captures of the data we needed). In other cases, we had to rely on the webmasters’ words, and in a few cases, we couldn’t get proper information on when the “hits” took place. See graph below.

The graph above shows when were the websites hit by HCU, GCU, any of both or none.

Please note: for ease of visualization, events were added when the traffic loss was evident. So if only HCU, then it’s labeled as HCU, GCU1 means traffic loss only in March 2024 GCU, GCU2 means August 2024 GCU, HCU + GCU means both HCU and a GCU hit (it doesn’t matter which because in all these cases the search traffic losses took place in March and in August), Undisclosed represents the ones we couldn’t get data. Finally, the biggest and more important one for our study: traffic losses didn’t happen on an update; they either happened before or after the Google core updates.

Examples of websites hit by HCU, GCU, both or none

In order to provide a data driven visualization, we’ll present some examples of the different cases we studied. Of course we have a lot more than these, but in order not to make the article too long, we think one example for each one will suffice.

Website Hit by HCU

This is a website that was hit by the Helpful Content Update (HCU). The website was purely AI-based, so the owner thought the reason for the drop was the AI-generated content. He tried to recover by removing AI content, manually writing posts, and purchasing backlinks. Unfortunately, he was never able to regain the traffic and eventually considered the website lost.

An archetypal example of a website hit by HCU that never recovered

This was probably the most difficult case because HCU is for content. However, to our surprise, a simple analysis of the console showed us this:

Console’s errors (

For non-technical people, this website had many errors, including attempts to inject Web3 (crypto). The red messages indicate these errors. The yellow messages are warnings, but if you look closely, you’ll see they are actually errors as well: essentially, half of the website wasn’t loading. The curious part is that an analysis using CWV showed extremely good results, almost 100% in every category, yet there were still many errors.

It wasn’t hard to figure out the reason, but we asked the webmaster, and it turns out a web development agency rebuilt the website for ultra speed. They achieved this by loading an empty website first and then loading the resources, resulting in the issues mentioned above. Since the website owner saw the near-perfect speed results, he never suspected that this would be the cause of the problems, and this led to even more issues.

In short, the timing was right as it happened after an HCU, but even then, we must attribute the traffic loss to technical reasons. A change in content did nothing, and the website was never able to recover its traffic and ranking again.

Real hits by HCU and GCU

This website shows evidence of real hits by both HCU and GCU

This website is an example of one that was significantly impacted by the Helpful Content Update (HCU). The webmaster tried to implement fixes, but these only worked until each of the core updates in 2024. We can see how the traffic was higher before both the HCU and the March and August Google Core Updates (GCUs), but then it was affected by the HCU and the GCUs.

The reason for this problem was the use of ReactJS in CSR mode and (possibly) a database issue, although this still needs further testing.

Note: While the percentages appear positive, this is because they are compared to the previous period (13 to 24 months ago), during which the website only existed for 5 months. Therefore, it had significantly less traffic, which is represented by the dotted line.

False HCU and GCU identification

In this case, which we can categorize as “none,” we can clearly see that this website had its best traffic during the HCU and GCUs but lost it one or two months later. Since we know that traffic and rank changes occur during update periods or in the month immediately after, this gap indicates the traffic loss is due to another reason.

As for the increases in traffic, we can only speculate, as nobody knows the Google algorithm. However, our theory is that the website had “something,” and the update improved its ranks to test user reactions. Since user behavior didn’t show any improvement, the website was subsequently demoted.

A website incorrectly labeled as hit by HCU and GCU

Further investigation supports this theory. The content is good, and while it has many errors, it’s more or less average. However, the website had very serious UX issues—so serious that no user was able to complete the user flow, resulting in zero conversions.

In our opinion, which is shared by a few colleagues we’ve discussed this with, this is the real reason for the demotion. The webmaster is already aware of this and fully agreed once we showed him the proof, so he is now in the process of getting his website redesigned.

The same issue occurred with the graph above, as well as with many of the websites for which we had access to the stats: traffic loss was incorrectly attributed to the updates. In this case, the reasons were purely technical. If you check the image below, you’ll notice most of the items are in red, including an 11-second LCP.

Is Google Core Update Technical?

After all the studies and research we did, we proved our initial hypothesis, already mentioned in a previous post: the statistical significance of the obtained data proves without a shadow of a doubt that the 2024 Google Core Updates were eminently focused on Technical SEO.

How did we come to this conclusion? Of course, using data and UX research, as usual. But let’s be more specific.

When studying all the sites, the first thing we did was to check when they happened. Much to our surprise, the biggest number of affected sites (44) didn’t happen on any update. This proves that the updates had no direct relation whatsoever. Of course, it doesn’t mean that events happening AFTER the updates weren’t influenced by previous updates.

When researching websites hit by 2023 HCU, there were two options to study: those websites that were fixed and those that weren’t fixed. Of the 11 sites we got as hit only by the HCU, all of them had some kind of fixing, and only 3 of them were hit by any of the GCUs. From the 9 hit by both HCU and GCU, only 4 of them tried a fix. All 4 focused on content, all 4 failed, which may explain the GCU hit.

Now, about the GCUs: there were 42 affected websites if we consider GCU1, GCU2, and HCU+GCU. 35 of them tried some kind of fix or recovery, mostly improving content. Of those 35, 18 succeeded in having short improvements, then they were hit again by the next update. 9 had a clear improvement, getting to the same or even better traffic that they had before the traffic loss, and 8 were affected by the August GCU, so it’s too soon to tell.

Main Reason for traffic losses

As mentioned above, there was a big hint about the reasons for the so-called Google Core Updates losses: many of them didn’t happen within a core update period. Interesting, isn’t it?

So, we analyzed all the websites. Yes, the 113 websites, one by one.

Since general reasons (content, backlinks, authority, etc) are arguable, take time to prove, and are too broad, we only focused on finding elements that proved or disproved our hypothesis: whether the problem was technical SEO or not. And as said before, the evidence is quite conclusive: 53 websites had technical issues (from technical SEO to coding issues), 31 that are likely (but evidence is not conclusive: they have technical issues, but either they are not important or there are other more important issues), and then 29 that had no evident technical issues or were unlikely.

Finally, we offered the solution to all of those websites with technical issues, whether they were our clients, asked in public groups, or were provided by colleagues. This had three reasons: first, as a show of appreciation for those that provided the data for this report; second, to test whether those fixes would improve the traffic or not; and third, to show this kind of fixes are relatively quick to implement and their effect is almost immediate. No more “we have to give it 6 months” (or a year or whatever): technical fixes are boolean; they either work or not.

And you can see the result in the graph below. Of those 84 with technical issues, 37 are already fixed, and 28 we don’t know (either we have no info, or it’s too soon to tell). Only 19 were not fixed, which includes those where the fix wasn’t applied for different reasons. And out of the 113 websites we analyzed, only 29 hadn’t technical issues (actually 7, and 22 that we deemed as unlikely, yet not definitive).

Green lines show the number of websites affected. It’s easy to see Technical SEO is the main cause.

Conclusions and Insights

Based on the data and the quite important number of websites we analyzed, it’s safe to say that the March and August Google Core Updates’ algorithms are meant to act on technical SEO and User Experience rather than content, backlinks, or authority.Another important discovery was finding that WordPress websites weren’t specifically targeted, even though it’s the CMS with the largest market share. At the same level of importance, we found that ReactJS tends to have been targeted.

A likely explanation is its architecture: ReactJS loads in blocks when used with Client-Side Rendering (CSR). This issue has been widely discussed by many technical SEOs and developers. CSR offers the ability for perceived fast load times. However, it only loads a minimal part, so Google crawls that minimal part, not recognizing all the content, or perhaps penalizing for large layout shifts. We strongly recommend using a Server-Side Rendering (SSR) approach (for example, using Gatsby).

Gatsby may have issues and an awful performance, but it’s a much better choice than pure React.

Finally, regarding content, we couldn’t find any data demonstrating any influence of AI, neither good nor bad. Furthermore, we couldn’t confirm that content improvements had a real influence, not even for sites hit by the HCU. However, this could just mean that Google found other issues besides content.

Last Thoughts

The most important takeaway from this research is that SEO experts and webmasters will need to tighten up their websites from a technical point of view. If I had to guess the rationale behind this approach, I think that with so many pages having the exact same information, especially because of AI and automation, content has lost most of its meaning.

It’s just speculation, but I wrote an article about a certain lack of relevance in Google Search Results some years ago, and the HCU seemed to confirm what I wrote, with the GCUs as the last nails in the coffin.

On the bright side, if technical SEO is the main reason for Google targeting websites, then there’s no more ambiguity: technical SEO issues are relatively fast to fix, and the results are almost immediate.