Google Rankings Still Control Your AI Visibility. The Data Is Uncomfortable.
A study of 11 major sites shows Google ranking drops cause AI citation losses of 22.5% on average. ChatGPT is the most dependent. Here's what to do.
Google Rankings Still Control Your AI Visibility. The Data Is Uncomfortable.
The pitch for AI search has always been independence. ChatGPT, Perplexity, Gemini: they’re supposed to be different from Google. New algorithms, new signals, new ways to discover brands. A fresh start for companies that never cracked page one.
That story just ran into a wall of data.
Lily Ray published a study analyzing 11 major websites that lost organic Google traffic during an unconfirmed algorithm update between January 20 and February 16, 2026. Every single one also lost AI search citations. Not some of them. All 11.
The average organic traffic drop was 26.7%. The average AI citation decline across all platforms was 22.5%. The correlation is almost 1:1.
And here’s the part that should worry anyone betting on AI search as an alternative channel: ChatGPT, a product built by OpenAI with no connection to Google’s ranking systems, showed the steepest citation losses. Not Google AI Mode. Not Gemini. ChatGPT.
The Platform Breakdown
The study tracked citation changes across four AI platforms during the same period. The results are striking.
| Platform | Avg. Citation Change | Sites With Declines | Worst Single Decline |
|---|---|---|---|
| ChatGPT | -27.8% | 11 of 11 | -42.3% |
| Google AI Mode | -23.8% | 11 of 11 | Not reported |
| Gemini | Moderate decline | 10 of 11 | Not reported |
| Perplexity | -2.9% | 4 of 11 | N/A (7 sites gained) |
ChatGPT’s numbers are the headline. Five of the 11 sites saw citation declines exceeding 34% on ChatGPT alone. One subfolder dropped 42.3%, which was worse than its organic traffic loss.
Google AI Mode’s -23.8% decline makes intuitive sense. Google’s AI products pull from Google’s index. If a site drops in Google’s rankings, of course it drops in Google’s AI answers too. Same data, same signals.
But ChatGPT’s -27.8% is harder to explain. OpenAI runs its own crawlers. It has its own retrieval pipeline. So why does it mirror Google’s ranking changes so closely?
Why ChatGPT Follows Google
The answer is architectural. ChatGPT’s web search function relies heavily on external search results to ground its answers. When a user asks a question that requires current information, ChatGPT doesn’t just search its training data. It runs web queries, retrieves results, and synthesizes them into a response.
Those web queries go through Bing. And Bing’s index, while independent from Google’s algorithm, often surfaces similar authoritative sources. When Google’s algorithm decides a site’s content folder is less authoritative (as happened in the January 2026 update), the signals that led to that conclusion, things like content quality, freshness, and topical authority, are frequently the same signals Bing uses.
The result is a cascade. Google drops your rankings. Bing’s signals often agree. ChatGPT’s retrieval system pulls from Bing’s results. Your AI citations fall.
It’s not that ChatGPT is literally checking Google rankings. It’s that the web’s quality signals are more correlated than anyone wants to admit. When multiple ranking systems independently evaluate the same content and reach similar conclusions, losing ground in one means losing ground everywhere.
Perplexity: The Exception That Proves the Rule
Perplexity’s numbers are the outlier. Only 4 of the 11 sites saw citation declines. Seven actually gained citations despite losing Google organic traffic.
This isn’t a mystery. Perplexity built its own search index. It uses Brave’s search infrastructure and its own crawling infrastructure rather than depending on Google or Bing for retrieval. When Google re-evaluates a site’s authority, Perplexity’s index doesn’t automatically follow.
That independence has practical implications. If you’re tracking your AI visibility across platforms and you see your ChatGPT and Google AI Mode citations drop in tandem but your Perplexity numbers hold steady, the problem isn’t your content. It’s your Google rankings.
And if all four platforms drop simultaneously, the problem is almost certainly your content itself.
The Volatility Problem Makes It Worse
The Google dependency story gets more concerning when you layer on citation volatility data. BrightEdge tracked weekly citation changes across major AI platforms and found that AI search is consolidating, not expanding.
Week over week, 96.8% of cited domains saw zero change. That sounds stable. But among the roughly 3% of domains that did change, 87% were declines. Only 0.4% of all tracked domains gained new citations in any given week.
The top 1% of domains now account for 64% of all AI citations. That’s extreme concentration. The rich are getting richer, and the rest are fighting over scraps that keep shrinking.
A separate BrightEdge analysis on citation volatility quantified the stability gap by authority level:
| Domain Market Share | Weekly Volatility (Google AIO) | Weekly Volatility (ChatGPT) |
|---|---|---|
| Dominant (>5%) | 2.8% | 1.4% |
| Major (1-5%) | 6.9% | 4.9% |
| Medium (0.5-1%) | 12.2% | — |
| Small (0.1-0.5%) | 19.9% | — |
| Tiny (<0.1%) | 42.9% | 50.7% |
Dominant domains experience 1-3% weekly volatility. Tiny domains experience 40-50%. That’s a 70x stability gap. If you’re not already a major cited source, your AI visibility swings wildly from week to week, and 87% of those swings are downward.
The 50-citation threshold appears to be the inflection point. Domains cited fewer than 50 times per week experience roughly 50% volatility. Cross that threshold and volatility drops to around 8%. Get above it and you’re relatively safe. Stay below it and your AI visibility is a coin flip.
What This Actually Means for Your Strategy
Here’s the uncomfortable truth that nobody in the GEO space wants to say out loud: for most brands, Google SEO is still the most important lever for AI visibility.
Not because AI platforms use Google’s rankings directly (Perplexity proves they don’t have to). But because the same content quality signals that Google rewards are the ones that Bing, ChatGPT’s retrieval system, and Google’s own AI products all converge on.
The narrative that GEO is separate from SEO, that you need a brand-new playbook for AI search, is only partially true. You do need to understand how brand mentions matter differently from backlinks, how to structure content for citation, and how off-site presence feeds into AI discovery. Those are real, important tactical adjustments.
But if your Google organic visibility is declining, your AI visibility is almost certainly declining with it. Fixing the foundation matters more than optimizing the surface.
Three things to do right now
1. Audit your Google organic trends before blaming AI platforms
If your AI citations dropped recently, check your Google Search Console first. Look at the specific content folders that lost visibility. Ray’s study found the January 2026 update hit company blogs and informational resource pages hardest. If those are the same sections losing AI citations, the root cause is Google, not some AI-specific signal you missed.
2. Treat Perplexity as your independence benchmark
Because Perplexity uses an independent index, it’s the best platform for isolating content quality issues from Google ranking issues. If your Perplexity citations are stable while ChatGPT and Google AI Mode citations decline, your content is fine; your Google rankings are the problem. If Perplexity drops too, the issue is the content itself.
3. Target the 50-citation stability threshold
BrightEdge’s data shows that domains cited fewer than 50 times per week face extreme volatility (50%+ weekly swings). Getting above that threshold is the difference between stable AI visibility and random noise. This means concentrating your efforts on the queries and topics where you have the best chance of becoming a consistently cited source, rather than spreading thin across hundreds of queries where you’ll always be marginal.
The Bigger Picture
There’s a version of this story that’s optimistic. The fact that Perplexity operates independently proves that AI platforms don’t have to mirror Google’s judgments. As Perplexity grows (it had 148 million web visits in August 2025, compared to ChatGPT’s 5.8 billion), and as other platforms build out their own retrieval infrastructure, the Google dependency could weaken over time.
But right now? ChatGPT processes over 100 million weekly queries. Google AI Overviews appear on over 40% of searches. These are the platforms where AI visibility matters most, and they’re both tightly coupled to Google’s view of the web.
The teams that will win in AI search over the next 12 months are the ones that stop treating GEO and SEO as separate disciplines. They’re connected at the root. Ignoring that connection means building your AI visibility strategy on a foundation you’re not monitoring.
RivalHound tracks your brand’s visibility across ChatGPT, Google AI, Perplexity, and more. Start monitoring to see where you stand.