Six attribution tools. None agree. You can't see causality, only correlation.

Three attribution traps. Same disagreement every time.

Growth-stage B2B attribution stacks converge on three failure modes. The team is not wrong. The models are not wrong. The stack was assembled without picking a primary, and now every conversation is a debate about whose dashboard tells the truth.

Trap one: last-touch in the CFO model

The CFO uses last-touch attribution or self-reported source because both are easy to defend in a board meeting. Last-touch overweights the channel closest to the form fill. Self-reported source overweights the most memorable channel. Neither model can see the first 95% of the buying journey that Gartner says happens before contact.

Trap two: multi-touch in the platform model

The marketing platform uses multi-touch attribution, which credits every channel that appears on the path. The number is always higher than the CFO number because more channels get credit. The platform model is internally consistent and operationally useful, but it cannot be reconciled with last-touch without choosing one as primary.

Trap three: no incrementality test anywhere

Both models are credit-assignment models. Neither answers whether the channel actually moved revenue. Causality requires holdout tests, geo experiments, or marketing mix modeling. Madison Logic's 2024 research finds most growth-stage B2B teams have zero incrementality testing in the stack, so every attribution conversation is a correlation argument.

The buyer is a committee. The model assumes a contact.

Most attribution tools were built on a model of a single buyer moving through a single funnel. The buyer in 2026 is not that. Forrester's 2024 research on buying-group engagement names demand units as the unit of analysis: clusters of contacts inside a committee, touching multiple channels over weeks or months. McKinsey's B2B Pulse research adds that the committee averages 10 people and 10 or more interactions before a decision.

A single-contact attribution model running on a committee buyer is structurally undercounting in some places and overcounting in others. The disagreement between models is the natural output of running mismatched models on mismatched data. Gartner's 2024 essay puts it plainly: most B2B firms have a pipeline-operations problem, not an attribution problem. Better tools do not fix it. Better operating choices do.

The disagreement is not a tooling failure. It is the predictable output of running attribution as if the buyer were a contact when the buyer is a committee. Pick a primary, demote the rest, run one causality test per quarter on the biggest channel. The argument stops in the same week.

Three exercises you can run this week before booking a call.

Exercise one: the source-of-truth audit

List every tool, dashboard, or spreadsheet that produces an attribution number in your company. Next to each, write the model it uses and who uses it. If the list is over four entries and there is no declared primary, the debate is structural. The fix is a declaration, not a tool.

Exercise two: the same-deal test

Pick three recently closed deals. Open each in every attribution tool you use. Write down which channel each tool credits as the source. If the answers do not match for more than one of the three deals, you have the disagreement at scale. Forrester and McKinsey 2024 research predicts this.

Exercise three: the incrementality test

Ask the team when the last holdout test or incrementality experiment was run on a meaningful channel. If the answer is never, every attribution number in the company is correlation, not causality. Madison Logic 2024 finds this is the norm in growth-stage B2B, not the exception.

The attribution disagreement is documented. The fix is operating, not tooling.

The pattern of multi-tool attribution disagreement shows up in every B2B revenue study from the last five years. The marketing review does not invent a model. It applies what is known to your specific stack.

What you walk away with after the 60 minutes.

The marketing review call is not a tool review. It is one operator sitting with your founder, head of marketing, and head of revenue ops if you have one, for an hour, naming the primary model and demoting the rest. Inside 48 hours you receive a one-page written summary with the following.

• The primary attribution model named. One model, declared for budget decisions, mapped to your revenue motion.
• The demotion list. Each remaining tool with a one-line rule for which question it answers and which question it does not.
• One incrementality test scoped. The biggest channel, a holdout or geo design, a budget number, and a decision deadline.
• A quarterly review template. One page, one model, the rule for when the model is overruled by causality evidence.
• Next step. If the gap turns out to be a pipeline operations problem, we name it and point you to Marketing Strategy Review ($5K) or Conversion Review ($3.5K). No upsell pressure on the call.

Related pain points and marketing offers.