I recently wrote about an unconfirmed Google algorithm update that rolled out in mid-January 2026, which negatively impacted the organic search visibility of dozens of major brands. For most of the impacted sites I analyzed, the impact was disproportionately targeted to the company’s blog, or another folder containing informational articles and resources.
That same organic trajectory has continued into mid-February for all of the subfolders I analyzed, using the Sistrix U.S. Visibility Index:
Zooming out, this is what the drops look like when you look at the visibility trends across the whole domains, not just the blogs:
Here is another example of the visibility impact on the company blog for the biggest company in the list (in terms of both ARR and organic visibility):
And this is what the impact looks like when you look at the company’s full domain’s visibility in organic search:
Needless to say, these recent organic visibility drops were extreme, relative to the sites’ overall SEO trajectories over the past few years. Drilling down into 11 of the sites that saw extreme declines over the last month, I wanted to see if this new data could help answer another question:
Do drops in Google organic search visibility coincide with similar drops in AI search citations?
My working hypothesis is that these drops are no longer just isolated to traditional search. Instead, I suspect we will find that, for most LLMs, AI search citation trends mirror what happens in Google’s organic search results, for two reasons:
1. The Direct Pipeline: Google’s AI Ecosystem
For Google’s own AI products – AI Mode and Gemini – the correlation should be strongest. Presumably, Google is using its own index and top-ranking search results to formulate AI search responses; therefore, dropping in organic rankings should logically cause those pages to be cited and referenced less frequently in generative answers.
2. The Downstream Effects: Third-Party LLMs (ChatGPT & Perplexity)
The link between Google organic rankings and third-party LLMs like ChatGPT and Perplexity is more nuanced, as we don’t know exactly which search engines these LLMs are surfacing for web search.
While there is a growing body of evidence (and industry reporting) suggesting that ChatGPT likely scrapes Google during live web searches, we still technically lack official confirmation from the source. Perplexity, on the other hand, is currently believed to utilize the Brave Search API as a core part of its retrieval process, alongside its own specialized “PerplexityBot” crawler.
To test this out, I wanted to drill down into the subfolders that saw substantial visibility drops on Google in recent weeks, to see whether the trend line for AI search citations followed suit.
To start, I honed in on a list of 11 sites whose subfolders saw substantial organic traffic drops between January 20, 2026 and February 16, 2026.
I used the Ahrefs MCP server with Claude Cowork to pull in estimated global monthly organic traffic numbers for each path (subfolder) in the list. Because most of the traffic declines started around January 21, 2026, I pulled the projected monthly organic numbers for January 20, 2026 and the most recent date, February 16, 2026.
I also redacted the site names, leaving the name of the subfolder and a brief, anonymized summary about the company type and the subfolder’s purpose:
These subfolders experienced anywhere from a -5.7% to -53.1% drop in estimated monthly organic search traffic since January 20th, 2026.
Using Ahrefs Brand Radar, you can drill down to see the number of AI search citations that a given subfolder has received across various LLMs over time. For example, here is the ChatGPT citation trend line for the first subfolder listed in the above table (U.S. data):
This is the corresponding chart showing the organic traffic trend for this same subfolder, which began dropping around January 21, 2026:
I used the Ahrefs MCP server with Claude Cowork to pull global traffic and citation data, and to analyze this same pattern for 11 of the subfolders that saw big drops.
Note on methodology: While 11 subfolders are a small sample size, I was specifically looking for a “clean” data set – subfolders experiencing a similar algorithmic demotion on Google during the unconfirmed January 2026 update. By narrowing the scope, I could better isolate whether a loss of traditional search visibility translates directly into a citation drop in AI search.
Below are the high-level summaries of how organic traffic and citation counts changed across Google and various LLMs, including AI Mode, ChatGPT, Perplexity, and Gemini:
Findings:
- The data shows a broad decline in both SEO traffic & AI search citations: Every subfolder in the study (11 of 11) experienced a drop in both Google organic traffic and total AI search citations, with a significant average citation decline of -22.5%.
- Google’s AI Mode (-23.8%) and ChatGPT (-27.8%) showed the most severe declines, closely mirroring the -26.7% average drop in organic traffic.
- While Gemini also saw broad declines (10 of 11 sites), Perplexity proved to be the most resilient, with only 4 of the 11 sites seeing a drop and a much milder average change of -2.9%.
- This data supports the theory that Perplexity is primarily using non-Google search surfaces to generate its responses.
Looking at the changes in estimated organic traffic for each subfolder compared to total AI search citations between January 20 and February 16, 2026, the correlation is clear: Significant losses in organic search visibility are almost universally mirrored by a corresponding decline in AI search citations.
Drilling down into specific LLMs, including Google’s AI Mode, ChatGPT, Perplexity, and Gemini, shows how the decline was nearly universal for most platforms, whereas Perplexity frequently displayed a significant divergence, showing positive citation growth for the majority of the subfolders despite their organic traffic losses.
ChatGPT (green) consistently shows the deepest declines across almost every subfolder – often exceeding AI Mode and Gemini. This is intriguing because ChatGPT isn’t a Google product, yet it appears more sensitive to these organic ranking shifts than Google’s own Gemini.
This appears to be another clue that ChatGPT is reliant on Google’s search index during retrieval.
AI Mode and Gemini tend to move in the same direction but not the same magnitude. Despite both being Google products, AI Mode declines are generally steeper than Gemini’s. This could suggest they weight or source from Google’s organic index differently – perhaps AI Mode is more tightly coupled to live SERP rankings while Gemini draws from a broader or cached knowledge base.
The few sites where Perplexity did decline (e.g., Site J, Site K) are also the ones showing relatively smaller organic drops. So even in the cases where Perplexity tracked downward, it doesn’t appear to be correlated with the severity of the Google organic loss – further evidence that Perplexity is likely pulling from a different retrieval pipeline.
The below table shows all the organic search vs. AI search citation data in one place:
The table reveals a clear pattern: every subfolder that lost organic visibility on Google also saw a decline in total AI search citations, with an average drop of -22.5% across all LLMs.
ChatGPT was the most severely impacted platform, with citation declines reaching as high as -42.3% (Site E) and exceeding -34% for five of the eleven subfolders – often surpassing even the organic traffic loss itself.
Google’s AI Mode followed a similar trajectory, while Gemini showed more moderate declines across the board.
The most notable outlier is Perplexity, which actually showed citation growth for 7 of the 11 subfolders, reinforcing the theory that it retrieves from a non-Google search index.
Perhaps the most interesting finding is that ChatGPT – a non-Google product – appears more tightly coupled to Google’s organic rankings than Google’s own Gemini, suggesting that ChatGPT’s web retrieval pipeline is heavily dependent on Google’s search results.
One recommendation I’ve been making since AI search entered the SEO conversation is that you shouldn’t invest in AEO/GEO tactics that could be detrimental to SEO performance. For example, using hidden prompt injections, cloaking, or self-promotional listicles (tactics that some have advocated for to boost AI search visibility) might be temporarily beneficial for AI search, but could cause massive headaches with Google and Bing’s organic search ranking algorithms down the line.
Now, we have even more evidence that AI search is fundamentally connected to SEO performance: If you drop in organic search, you can likely expect a corresponding drop in citations not only from Google’s own AI search products, but from other LLMs like ChatGPT, which appear to also be heavily reliant on Google’s search results.
The one notable exception is Perplexity, which showed citation growth for the majority of subfolders hit by the recent algorithm update. That said, it’s important to weigh this against the scale of traffic and LLM usage at stake. According to a recent article by Similarweb, ChatGPT received 5.8 billion web visits in August 2025, compared to 148.2 million for Perplexity.
To add to this, when you factor in Google’s organic search traffic – which still dwarfs all the AI search platforms combined – the vast majority of your search-driven visibility across both search engines and AI chatbots is still flowing through a pipeline where Google’s rankings dictate the outcome.
For the past year, the SEO industry has been asking how closely traditional SEO and AEO/GEO are really tied together. I think this data helps answer that question: Not only is a strong SEO foundation critical for AI search visibility, but tactics that hurt your organic rankings can have a cascading negative impact on your AI search citations as well. In other words, the fastest way to lose visibility in AI search might be to lose it in Google first.
More Resources:
This post was originally published on Lily Ray NYC Substack.
Featured Image: PeopleImages/Shutterstock
