Key Takeaways

●      Reddit’s AI visibility collapse (Oct) was the first public sign of Google tightening control over data access

●      Gemini and Google Search now operate on separate data systems

●      Traditional SEO metrics no longer predict AI visibility

How Google’s Indexing Change Triggered Reddit’s AI Visibility Collapse

The moment Reddit’s AI visibility collapsed, most people missed what was actually happening…

In mid-September 2025, Google deprecated the &num=100 URL parameter. (In short, this let users see 100 search result pages at once). This reduced the amount of content that could be surfaced, indexed, and accessed by external tools and AI scrapers. At the time, it looked like a small technical change.

In early to mid-October, the impact became visible. Reddit’s share of AI citations inside large language models dropped from roughly 29 percent to just over 5 percent in a matter of weeks. By October, the decline was clear across industry analysis and market response.

The num=100 removal, the collapse in AI citations, and the divergence between Search and Gemini pointed to the same conclusion.

Google was not just adjusting search. It was changing how information is gathered, indexed, and fed into AI systems.

This was the first phase of a larger shift that will determine who gets found, cited, and trusted in the next decade of search.

So we dug deeper into what Google was doing…

Gemini Is Not Built on Google Search

New research from Ahrefs revealed something unusual: Gemini only matches Google’s top 10 results 12% of the time, while ChatGPT matches at 62%. In other words, ChatGPT is essentially closer to Google Search than Google’s own AI.

This is the part that changed how we look at Google’s strategy. Here is what this means:

●      Gemini isn’t a Search upgrade

●      Gemini isn’t mirroring the open web

●      Gemini is drawing from a separate, proprietary dataset

That’s why AI strategist Kasim Aslam’s theory resonates: Google is intentionally severing Gemini from the open web, building a proprietary knowledge base that competitors can’t easily replicate. When you isolate the data, you control the output. And when you control the output, you control the market. This is also why we see sites dominating organic rankings but missing entirely from Gemini answers. Meanwhile, niche brand-owned pages appear quickly. It’s a different rulebook entirely.

What Changed for Brands? More Fragmentation, Not More Reach

For 20 years, ranking in Google was enough. If you won there, the rest followed, but that era is gone. Yext’s 2025 analysis showed that 52 percent of Gemini’s citations now come directly from a brand’s own website. That means Gemini rewards companies that explain their products clearly, structure their data, and update content with verifiable information.

ChatGPT, on the other hand, relies heavily on high-trust directories, vertical listings, Wikipedia, and well-maintained third-party profiles. In contrast, Perplexity draws deeply from expert sources and analyst-grade references.

The result:

● You can dominate Google and be invisible in Gemini.

● You can own Gemini and barely appear in ChatGPT.

● You can show up in ChatGPT and never surface in Perplexity.

In a nutshell, visibility is no longer linear. It’s fractured. These three engines now behave like separate universes. And the correlation between Google rankings and AI mentions is close to zero in some verticals. Being #1 on Google no longer tells you anything about how AI engines will surface you. This is the first major break between search visibility and AI visibility.

Google’s Strategic Data Lockdown 

Once we accepted that Gemini wasn’t tied to Search, everything else clicked. Google’s market share dipped below 90% for the first time in a decade, only to rebound after AI Overviews launched. Google knows that its long-term leverage is data, and right now, it is closing the doors on the only dataset large enough to train global-scale AI models: the Google index.

The num=100 removal was step one.

The Search/Gemini split was step two.

Next comes deeper dependence on proprietary signals only Google can serve. Competitors see this coming, and that’s why Google isn’t the only one protecting its data:

●  OpenAI launched a browser

● Perplexity launched Comet

● Others are building their own data collection pipelines

Everyone is racing to reduce dependence on Google’s crawling infrastructure. Because whichever company controls the data controls the next 20 years of search.

What B2B Leaders Need to Do Right Now

This new ecosystem isn’t won with “more SEO.” It’s won by proving authority everywhere AI engines look. They all work differently and we don’t know who will be the clear winner. You need to optimize for all. Here’s a short version of the updated playbook:

●  Build a consistent entity profile: Your identity must match across all profiles (IE Crunchbase, G2, Clutch, LinkedIn, and industry directories). AI engines penalize inconsistencies more than humans do.

●  Add structured data: Gemini rewards well-written, fact-based content housed on your own site. If your site doesn’t answer your industry’s core questions, someone else will, and Gemini will pick them. FAQ, HowTo, Product, and Article schemas turn your site into something AI can parse without losing context.

●  Build third-party authority: ChatGPT and Perplexity trust reputable validators. Secure citations, reviews, and mentions across the industry ecosystem.

●  Publish measurable insights: AI engines surface information. Publish data, benchmarks, comparisons, and original findings that AI can treat as authoritative.

Visibility Has Become a Data War: Will Google Win?

Google isn’t losing the AI race. It’s quietly consolidating the most valuable asset in the entire ecosystem: data. And while most brands focus on rankings, Google has already moved on. If you want to show up in AI, you need authority signals across every ecosystem, not just Google.

Being “findable” is now the minimum requirement. Brands must be machine-recognizable, machine-verifiable, and machine-citable across every AI system that buyers use.

If you bet on just one channel, you risk losing sight of the others without realizing it.

FAQ

Does ranking #1 on Google still help?

For humans, yes. For AI engines, not much.

How do we know where we show up in AI results?

Audits. Consistent, multi-engine, multi-query mapping.

If we can only do one thing first, what should it be?

Fix your owned content. Gemini depends on it, and ChatGPT still sees it as a validation layer.