Your Attribution Platform Has No Idea the AI Funnel Exists

TL;DR: B2B buyers now do their early research in AI tools — ChatGPT, Perplexity, Claude — before they ever hit your website. Your attribution platform logs a direct-traffic visit with no context. The right response isn’t a new tool. It’s triangulation: mine your own search data, survey customers at conversion, and build a proxy model that correlates content visibility with downstream pipeline.

The Dark Funnel Got Darker

The original dark funnel — podcasts, LinkedIn scrolling, peer recommendations — was invisible because buyers weren’t clicking trackable links. Your attribution platform never knew the touchpoint happened, but the buyer still ended up on your website eventually. You could triangulate through self-reported surveys, branded search trends, and direct traffic anomalies.

The AI funnel is structurally different. According to 6sense research cited by MarketBetter, 73% of the B2B buying journey happens anonymously before a buyer ever contacts a vendor. A buyer who researches your category in ChatGPT or Perplexity may never visit your website until they’re ready to book a demo. By the time they show up, the decision is largely made. Your attribution platform logs a direct-traffic session with no referral source, your pipeline model assigns zero early-stage influence, and the analyst who built your attribution model has no variable to represent what actually happened.

This isn’t a future problem. It’s the current state of B2B buying in any category where buyers are technically literate and the purchase cycle involves meaningful research.

Start With What You Can Already See

The instinct when confronted with invisible buyer behavior is to buy a new tool that promises to surface it. That’s usually the wrong starting point. Most marketing teams are sitting on first-party signals they haven’t fully used.

Google Search Console is the most underused intelligence asset in B2B marketing. It tells you exactly what queries are driving impressions and clicks to your site — including the questions buyers are asking before they find you. Map your highest-impression queries against your content library. Where are you surfacing for informational queries but failing to convert? Where are buyers asking questions you haven’t answered at all? That gap is a proxy for where AI tools are filling the information vacuum instead of your content.

Layer Google Trends on top of this. If branded search volume for a category term is rising but your organic traffic isn’t growing proportionally, buyers are researching somewhere else — and that divergence is worth investigating. Tools like Answer the Public extend this analysis into the long-tail conversational queries that buyers type into AI tools because they want explanations and comparisons, not vendor websites.

These signals won’t tell you what any individual buyer did. They tell you what buyers in your category are collectively trying to understand — and whether your content exists in the places where those questions get answered.

The Customer Survey Is Not Optional

Directional signals from search data are useful, but they don’t close the loop. The only way to understand what actually happened before a buyer converted is to ask them.

Customer surveys designed specifically to capture pre-conversion research behavior are straightforward to run and systematically underused in B2B. The key is separating this from your NPS or onboarding survey — those have different objectives and the framing bleeds into the answers. A dedicated survey with a focused question set, deployed at or shortly after conversion, generates cleaner signal.

Ask which sources informed their research before they contacted you. Include AI tools explicitly as an option — most buyers won’t volunteer this without a prompt. Ask where they first heard about your brand, what questions they were trying to answer when they found you, and what alternatives they evaluated. At reasonable conversion volumes, even a 25% response rate produces statistically useful data. It won’t give you a tracking-pixel-level picture of individual journeys, but it will tell you which channels and sources are actually influencing buyers — which is more than your attribution platform can claim at this stage.

Pair survey data with internal usage and product data. If your product has a trial motion, usage patterns of trial users who convert versus those who churn often reveal what kind of prior research shaped their starting expectations. That’s intelligence you already own and rarely fully exploit.

Building the Business Case Without CRM Events

The business case for investing in AI search visibility — publishing content structured to appear in AI-generated answers, optimizing for how models reference and cite sources — faces the same objection as every other brand investment: there’s no direct line to a CRM event.

The right response is to build the case on correlations rather than attribution. Identify a cohort of closed-won deals and work backward through every available signal: branded search volume trends in the weeks before conversion, direct traffic spikes, survey-reported research sources, whether your content appears in AI-generated answers for high-intent category queries. The pattern across enough deals gives you a proxy model — not proof, but a defensible hypothesis that can be calibrated over time.

This is structurally similar to how brand investments have always been justified. The CPG industry established through Nielsen’s share-of-voice research that SOV above your market share predicts market share growth. The same principle applies here: if your content is consistently surfaced in AI-generated answers for research queries in your category, and competitors’ content isn’t, that’s a structural advantage that will eventually show up in pipeline — even if the mechanism is invisible to your attribution tool.

FAQ

Q: Should we be trying to “optimize for AI” the same way we optimized for Google?

Not in the same way. Traditional SEO targets ranking signals. AI citation is more about authoritative, well-structured content that answers specific questions clearly. The discipline is the same — write useful, specific content for real buyer questions — but the tactical levers are different. Focus on depth and clarity over keyword density.

Q: How do we prove to the CFO this is worth investing in?

Start with the survey data. If 30% of new customers say they first encountered your brand through content before ever clicking anything trackable, that’s your starting figure for the revenue share influenced by content that attribution doesn’t see. Document it quarterly. The gap between what attribution reports and what customers say is the business case.

Q: How often should we run the conversion survey?

Every month, consistently. A 12-month rolling dataset is far more useful than a one-off project. You’re looking for trends — which sources are gaining or losing share of first-awareness — not a point-in-time snapshot.

Additional Resources

From the Zaitz Marketing Knowledge Library:

• Why Your Attribution Model Is Lying to You — The structural limits of attribution and what to use instead
• What is Incrementality in Marketing? — How to measure causal impact rather than observed correlation
• What Consumer Goods Gets Right That SaaS Has Never Learned — The SOV-to-market-share model and other CPG disciplines that apply directly to B2B
• How to Measure Brand-Building Content Without Lying to Your CFO — The full proxy measurement framework for brand investment

External Reading:

• The B2B Dark Funnel: How to Capture the 73% of Buyers You Can’t See — MarketBetter, 2026
• What Is the AI Dark Funnel and How Does It Affect B2B Pipeline? — Pedowitz Group
• How B2B Buyers Research in AI Before They Visit Your Website — 5k (formerly Conklin Media)
• AI and Marketing Measurement: What’s Changed in 2026 — Funnel.io