Google Just Made AI Mode Universal. Your Website Accounts for 5% of the Answers.
Google completed its AI Mode rollout to all US users last week. McKinsey data shows brand-owned sites account for only 5–10% of what AI search actually references. Here's what that means for founders building visibility now.
On March 4, Google completed its rollout of AI Mode to every user in the US. According to Google's own announcement, AI Mode now includes Canvas — a workspace for drafting, researching, and building inside Search — and is available to all US users in English. What was an opt-in tab for early adopters is now the primary interface for everyone.
The announcement was framed as a feature expansion. It is also a permanent change to how the world's most-used search engine delivers answers. More of your buyers' first-contact research will now produce a synthesized AI response before any links appear. What the AI says about your brand in that response depends on inputs most founders have never looked at.
McKinsey surveyed Fortune 500 CMOs in September 2025 and published a finding that should reframe how every growth team thinks about visibility: brand-owned sites account for only 5 to 10 percent of the sources AI-powered search actually references when answering queries. The other 90 to 95 percent comes from third-party editorial — review sites, industry journalism, trade press, research publications. Sources the engine treats as credible because someone other than the brand said so.
Every blog post, every optimized landing page, every piece of owned content your team has ever published lives in the minority of that equation. The majority is controlled by sources you don't own and didn't write.
The executive who watched it happen
MIT Sloan published a piece this January documenting what this looks like in a sales conversation. A financial services executive watched a prospect pull up ChatGPT to research their category. The exec's company held the largest market share in the space. It had spent more on digital marketing and SEO than any competitor. When the AI answered, it named a much smaller player. The executive's firm wasn't in the response.
This wasn't a content quality problem. It wasn't a keywords problem. It was a mismatch between where the company had been building its visibility and where the AI was actually reading.
AI-synthesized answers don't work the way traditional search results did. The old model was a referendum on your website — its structure, its authority, its content. AI synthesis is a referendum on what the rest of the internet says about you. Specifically, what sources the AI treats as credible enough to pull from.
A study published on arXiv in September 2025 mapped citation patterns across 1,702 responses from Brave Search, Google AI Overviews, and Perplexity, covering 16 B2B SaaS verticals. The finding: AI answer engines "systematically favour earned media (third-party, authoritative domains) over brand-owned and social content, with social platforms almost absent from AI answers." High-quality pages on vendor-owned domains were getting bypassed. The deciding variable wasn't on-page quality — it was whether the content lived on a domain the AI already treated as authoritative from external signals.
Your website, regardless of how well it's optimized, does not carry that signal by itself. The signal AI engines respond to is third-party editorial placement.
The 16 percent gap
The same McKinsey CMO survey surfaced something that belongs in every revenue review: only 16 percent of Fortune 500 brands are systematically tracking their performance in AI-powered search. The other 84 percent are investing in visibility strategies while measuring none of the layer where AI now shapes first-contact research.
McKinsey projects that by 2028, $750 billion in US revenue will flow through AI-powered search. The brands building the right infrastructure today are doing it while most competitors are still optimizing for an interface that is being replaced.
The gap doesn't stay hidden. It surfaces as deals lost for reasons nobody can trace. As competitors showing up in buyer research before the first call. As a pipeline that softens in ways attribution models can't explain. And when AI-referred traffic does arrive — it converts at significantly higher rates than traffic from traditional search, because AI-sourced buyers arrive further along in their decision. Most brands are missing both the volume and the quality.
What this requires
Google completing its AI Mode rollout means this is no longer a future problem. Every B2B buyer in the US now has AI-synthesized answers as their default search experience. The research your prospects run before they contact your team is running through a machine that reads the editorial record about your company — not your content strategy.
The brands that hold position in that record are building editorial presence in publications AI engines treat as authoritative. Not more owned content. Earned placements — in credible outlets with editorial standards, in sources AI pulls from when answering category questions.
This is the mechanism Machine Relations is built around: earned media in trusted publications drives AI citation the same way it drove human brand perception. The publications AI engines trust are the same ones that built category authority with human readers for decades. The mechanism didn't change. The reader expanded to include machines.
If your visibility strategy lives entirely on your own domain, it's working on the wrong 5%. The other 95% is what the machine reads about you when someone asks who leads your category.
Find out where you stand. The visibility audit shows you what AI systems are currently saying about your brand and which sources they're actually pulling from.