I was in a meeting once. I will let the transcript speak for itself.

CMO: Great news. Our new attribution model proves marketing influenced 184% of pipeline this quarter.

CFO: I'm sorry, did you say 184%?

CMO: Yes. Multi-touch attribution.

CFO: So more pipeline than existed?

CMO: Correct. That's the power of orchestration.

CFO: Walk me through this slowly.

CMO: Absolutely. So the customer saw a LinkedIn ad, attended a webinar, downloaded an ebook, heard our CEO on a podcast, visited the website 11 times, clicked a retargeting ad, and then finally replied to a cold outbound email from sales.

CFO: Okay.

CMO: Therefore marketing sourced the deal.

CFO: Sales literally emailed them.

CMO: Yes, but with awareness.

CFO: Fine. How are you assigning credit?

CMO: We use a blended AI-weighted hybrid waterfall attribution framework.

CFO: Which means?

CMO: Every department gets credit simultaneously.

CFO: That sounds less like accounting and more like kindergarten soccer.

CMO: Look at this chart.

CFO: Why is every channel 40%?

CMO: Because modern buyer journeys are non-linear.

CFO: That adds up to 280%.

CMO: Exactly. It reflects complexity.

CFO: What would happen if someone accidentally visited the website once?

CMO: Then marketing influenced their decision.

CFO: What if they were already a customer?

CMO: Expansion influenced.

CFO: What if they were an employee?

CMO: Employer brand attribution.

CFO: This line item says "dark social acceleration."

CMO: Huge this quarter.

CFO: How do you measure it?

CMO: We can't directly measure it.

CFO: Then why is it $480,000?

CMO: Conservative estimate.

CFO: And what exactly does your AI attribution model do?

CMO: It identifies hidden intent pathways.

CFO: Using what methodology?

CMO: Probabilistic engagement resonance.

CFO: You made that up just now.

CMO: The AI made it up, technically.

CFO: Here's a radical proposal: what if we credited revenue to the thing that actually caused the purchase?

CMO: We tried that once.

CFO: And?

CMO: It heavily favored sales.

CFO: Interesting coincidence.

CMO: Look, attribution isn't about truth. It's about alignment.

CFO: Alignment with what?

CMO: Next year's budget request.

CFO: Final question. If marketing disappeared tomorrow, what would happen?

CMO: Pipeline would collapse within 3 to 18 months due to long-tail brand decay dynamics.

CFO: And if attribution models disappeared tomorrow?

CMO: We'd all have to start admitting uncertainty.

CFO: Ah. So that's the real product.

Here is the thing. That conversation is funny because it is a perfect distillation of a real problem.

Attribution is not a measurement challenge. It is a political challenge dressed up as a measurement challenge.

Every team wants credit. Every platform claims influence. Every model is built, consciously or not, to justify the budget that already exists rather than to reveal the truth about what is actually working.

The CFO in that room was not wrong. The CMO was not stupid. They were both operating rationally inside a broken system.

Here is how the system breaks and what to do instead.

Why multi-touch attribution lies.

Multi-touch attribution assigns credit to every interaction a buyer has before converting. The logic sounds reasonable. Buyers touch many things. Therefore credit many things.

The problem is that correlation is not causation and the model cannot tell the difference.

The customer who saw your LinkedIn ad and then bought your product might have bought anyway. The webinar they attended might have had zero influence on the decision. The podcast might have annoyed them. You will never know, and the model will give all three full credit.

The result is that every channel looks important, no channel is clearly accountable, and the budget conversation becomes a negotiation between departments rather than a rational allocation of resources.

Why last-click attribution also lies.

The opposite extreme is last-click, which gives all the credit to the final touchpoint before conversion. Usually a branded search or a direct visit or a sales email.

This makes sales look like heroes and marketing look like overhead. It also cannot account for the fact that someone who has never heard of your company is unlikely to search for it directly.

Both models are telling a story. Neither is telling the truth.

What actually works.

Three things that cut through the attribution noise:

Incrementality testing. Run your campaigns in some markets and not others. Compare results. The difference is your actual impact. It is not glamorous. It does not produce charts with 280% on them. It produces a defensible number.

Holdout groups. Stop showing ads to a random sample of your audience. Measure whether they convert at a lower rate. If they do not, your ads are not doing what you think they are doing.

Revenue cohort analysis. Look at which acquisition channels produced customers who are still customers twelve months later. CAC is only half the equation. LTV by channel source is where the real decisions live.

None of these fit neatly into an AI-weighted hybrid waterfall attribution framework. All of them will survive a CFO asking hard questions in a meeting.

The real product, as the CFO correctly identified, is certainty. The attribution model exists to give everyone in the room a number they can defend.

The better product is honesty. Admitting that marketing influence is partially measurable, partially probabilistic, and never 184%.

Start there and the CFO might actually trust you.

Rob

P.S. Dark social acceleration is real. It is also completely unmeasurable in any honest sense. Budget it like you would a hunch, not a line item.