---
id: "concept-dark-funnel"
type: "concept"
source_timestamps: ["§ From Channel Management to Answer Engineering", "§ Where the Disruption Hits Hardest"]
tags: ["analytics", "buyer-journey", "visibility-gap"]
related: ["concept-generative-engine-optimization", "concept-prompt-authority", "claim-b2b-journey-compression"]
definition: "The invisible phase of the B2B buyer journey where prospects evaluate vendors using third-party AI tools, leaving no analytics trail for the seller."
external_validation: "conceptually-valid-emerging-term"
sources: ["geo"]
sourceVaultSlug: "hbr-seg-geo"
originDay: 3
articleStem: "hbr-tier1-01-gen-ai-b2b-buying"
sourceUrl: "https://hbr.org/2026/06/how-gen-ai-is-disrupting-b2b-buying-decisions"
sourceTitle: "How Gen AI is Disrupting B2B Buying Decisions"
---
# The Dark Funnel

The 'dark funnel' refers to the invisible phase of the modern B2B buying journey where prospects use generative AI tools to guide discovery, compare vendors, and evaluate fit. Because this research happens inside *third-party LLM interfaces* (ChatGPT, Gemini, Perplexity) rather than on a company's owned website or through a search engine that passes referral data, the traffic and intent signals remain largely invisible to the selling company. Sellers are left with limited insight into how, where, and to what extent their content is influencing buyer decisions.

This is the analytical blind spot that motivates the entire pivot to [[concept-generative-engine-optimization-d1]] and, specifically, to [[concept-generative-listening-systems]] (the 'Calibration' pillar of the [[framework-4c-generative-readiness]]) as a way to regain partial visibility. The demand-side symptom is journey compression — see [[claim-b2b-journey-compression]]. The concrete field example is HVAC installers moving from Google to ChatGPT/Gemini ([[quote-hvac-chatgpt-shift]], [[entity-imi]]).

**External validation (enrichment):** The term *predates* Gen AI — it originally described anonymous, untrackable pre-intent research (communities, review sites, word-of-mouth) and is used by firms like 6sense and metadata.io. GEO practitioners have extended it to AI assistants, noting LLM research produces *no referral data, no UTMs, minimal logs*. **Counter-perspective:** the funnel is 'darker than web analytics' but **not fully opaque** — vendors get partial visibility via logged interactions in their own chatbots/copilots and platforms that share referral data. Empirical measurement of AI's *share* of B2B research is still limited, so the metaphor may overstate current usage in some segments.


## Related across articles
- [[concept-conversion-pathway-compression]]
- [[concept-single-answer-insights]]
- [[claim-dialogue-replaces-search]]


## Related across segments
- [[concept-conversion-pathway-compression]]
- [[concept-zero-click-commerce]]
- [[quote-first-customer-algorithm]]