Discovery moved to AI, verification stayed on Google, and most SEO dashboards are still measuring the wrong room.
Last quarter, an SEO lead at a Series C infrastructure company watched their informational traffic drop 30%. Their board deck had a red arrow next to organic. The CMO started asking hard questions about headcount.
Meanwhile, demo requests from organic landing pages were up 18%. Pipeline sourced from organic was flat. Conversion rates on product pages had quietly climbed for three consecutive months.
The dashboard said SEO was failing. The revenue said otherwise.
The Funnel Flipped and the Metrics Didn’t Follow
New data from Wynter puts numbers on something practitioners have felt for a while: 84% of B2B buyers now use AI tools during vendor discovery, and 68% start their research in ChatGPT or Perplexity before they ever open Google.
Read that again. The majority of your buyers are forming opinions about your category before they type a single search query.
What happens next is the part that breaks attribution. A buyer asks ChatGPT which platforms handle their use case. The LLM names three vendors. The buyer opens Google, searches one of those brand names, clicks through, and requests a demo. Your CRM logs it as branded search. Your attribution model gives credit to the brand campaign. The actual discovery happened inside a black box that your analytics stack can’t see.
Gaetano DiNardi calls this the “Dark SEO Funnel,” and the framing is useful because it maps directly to a pattern B2B marketers already understand. Dark social taught us that peer recommendations in Slack channels and DMs were driving purchases that showed up as “direct” traffic. Dark SEO is the same dynamic, except the peer is an LLM.
Two Kinds of Visibility That Don’t Show Up in GA4
In this model, there are two ways your brand surfaces inside an AI-generated answer, and they require completely different strategies.
The first is brand mentions. The LLM names your company when a buyer asks something like “best enterprise ABM platforms.” You don’t earn this through technical SEO. You earn it through entity strength: how consistently your brand appears alongside relevant topics across review sites, communities, podcasts, and third-party publications. It’s the compounding effect of showing up in the right neighborhoods over time.
The second is URL citations. The LLM links to your content as a source because you published something it couldn’t find elsewhere. Original benchmarks, proprietary survey data, contrarian analysis with receipts. This is information gain in its purest form. The AI doesn’t cite your “What is ABM?” explainer. It cites your report showing that median net revenue retention for Series B SaaS companies dropped to 109% last year.
One rewards breadth of presence. The other rewards depth of insight. Most teams are investing in neither because they’re still optimizing for keyword rankings.
The Metric That Actually Tells the Story
The strongest signal in this new environment is one that most SEO teams have historically dismissed as someone else’s job: branded search volume.
Here’s why. If discovery is happening inside AI tools, the moment a buyer moves from AI to Google, they’re searching your brand name. Rising branded search, especially alongside declining informational traffic, doesn’t mean your SEO is broken. It means the funnel shifted and your content is doing its job in a place your dashboard can’t measure.
The most honest performance narrative right now looks like this: informational traffic declining, conversion rates rising, pipeline stable or growing. That’s not failure. That’s a funnel shedding noise and concentrating signal.
What Changes Operationally
The practical shift is smaller than it sounds, but it requires discipline.
Stop reporting on thousands of keyword rankings and start tracking 30 to 50 high-intent prompts. The queries buyers actually type into ChatGPT before they ever reach Google. “Best tools for X” and “how does Y compare to Z” are the new money keywords, except they’re money prompts.
Invest in the ecosystem around your brand, not just your own domain. Reviews on G2 and Capterra, mentions in Reddit threads, appearances on industry podcasts. LLMs build their recommendations from the broader web. If the only place your brand shows up is your own blog, you’re invisible to the model.
And publish things that can’t be summarized away. If an LLM can answer the question without citing you, your content didn’t add anything to the knowledge base. Original data, proprietary frameworks, research that forces the model to link back. That’s the new bar.
The Dashboard Problem Is the Real Problem
The hardest part of this transition isn’t strategic. It’s organizational. Every team reporting SEO performance through traffic metrics is telling a story that’s becoming less true by the quarter. And the teams that update their measurement frameworks first will be the ones that keep their budgets while everyone else is explaining why organic “isn’t working.”
Google didn’t stop mattering. It just changed jobs. It used to be where buyers went to discover. Now it’s where they go to verify a decision they already made somewhere else.
The brands that thrive in this environment will be the ones that show up in the room where the decision happens, not just the room where it gets confirmed.