The dark funnel was already a problem for B2B marketers. Now it’s gone full invisible.
~30% of B2B vendor research now happens inside AI chat windows — ChatGPT, Claude, Copilot, Perplexity. These conversations leave no referral header, no form fill, no trackable touchpoint. A prospect asks an LLM for “top 5 sales engagement platforms” and gets a curated list with pricing comparisons. Your marketing analytics see nothing.
What Is the Dark Funnel?
The dark funnel is all the buyer research that happens before and outside your tracked channels. It includes:
- Dark social: Slack communities, Discord servers, private LinkedIn groups
- Peer-to-peer: Direct conversations with colleagues, mentors, and industry friends
- Analyst reports: Gartner, Forrester, G2 research consumed behind paywalls
- Offline events: Conferences, dinners, and in-person networking
These channels have always been untrackable. But LLMs add a new layer: private, conversational research at scale.
LLMs as Private Advisors
When a buyer asks an LLM for vendor recommendations, three things happen that traditional marketing can’t track or influence:
- The LLM draws from training data — articles, reviews, comparisons published before its cutoff
- The buyer follows up with clarifying questions specific to their use case
- The LLM provides a ranked shortlist with reasoning, all in a private thread
This is different from dark social. Dark social leaves some trace — a UTM from a shared link, a mention in community analytics. LLM conversations are completely invisible to your attribution tools.
Why This Is Different From Dark Social
Dark social is peer-to-peer. LLM research is buyer-to-AI. The dynamic is fundamentally different:
| Dark Social | LLM Research |
|---|---|
| Peer recommendations carry weight | AI synthesizes multiple data sources |
| Happens in communities with visibility | Completely private, no analytics |
| Influenceable through community presence | Influenceable only through training data |
The implication: your LLM visibility depends on content published long before a buyer ever researches. You can’t “campaign” your way into an LLM recommendation in real time.
How to Be Visible When Buyers Ask LLMs About You
You can’t track LLM mentions, but you can influence them. The LLMs are trained on public web data. To be present when buyers ask about your category:
1. Own the basics on your site: Clear, detailed product pages, comparison content, and pricing information. LLMs prioritize structured, authoritative content.
2. Build depth in analyst and review coverage: G2, Gartner, Forrester, and tech media coverage are heavily weighted in LLM training data.
3. Publish thought leadership at scale: Blog posts, whitepapers, and frameworks that define your category. LLMs cite sources that explain concepts comprehensively.
4. Optimize for question answering: Write content that directly answers the questions buyers ask LLMs: “What’s the difference between X and Y?” “What are the top 5 tools for Z?”
Attribution Implications
The old attribution model — first touch, last touch, multi-touch — assumes all touches leave some trace. LLM interactions don’t. This means:
- Your conversion data will increasingly show “unknown source”
- Win-loss analysis becomes more important than funnel metrics
- Brand and content investing precedes measurable ROI by months or years
The companies that win in this environment are the ones willing to invest in channels they can’t track. LLM research is only growing. The dark funnel isn’t getting any brighter.
