Incrementality Proof Ladder

The Incrementality Proof Ladder ranks attribution evidence by how much causal weight it can carry in a client conversation. Last-click sits at the bottom: cheap to produce, easy to dispute, and increasingly blind to AI-assisted buying journeys that never register a referral click. Multi-touch models and media mix modeling sit in the middle, useful for directional budget shifts but still correlational. Holdout and geo tests sit at the top, because they answer the only question a CFO asks: what would have happened without this spend? Agencies that climb the ladder deliberately, pairing a daily-updated MMM from a platform like Prescient AI with periodic holdout tests from Measured, can defend a retainer with causal language instead of dashboard screenshots. The trap is skipping rungs: pitching incrementality results from a model that was never calibrated against a real experiment invites the exact scrutiny that ends contracts.