With flat budgets and AI in every workflow, the CMO must prove two things: that marketing creates value, and that AI in marketing creates value. Different evidence answers each.
Proving marketing's value
- Incrementality: experiments and holdouts on major channels.
- Marketing mix modelling, calibrated with experiments, to guide allocation.
- Brand health tracked over time (awareness, consideration, preference) and linked to pricing power and demand.
- Finance-agreed methods: the CFO accepts the method before seeing the result.
A reminder that brand and product drive demand
On its Q1 2022 earnings call, Airbnb's CEO Brian Chesky said "90% of our traffic or more is direct or unpaid", describing advertising as "a form of supplemental education" and not the core driver of growth. Airbnb is unusual, but the point generalises: measure what paid media adds, not what it touches.
Proving AI's value inside marketing
For every AI use case, track against a pre-AI baseline:
- Efficiency: cost per asset, cycle time, hours redeployed (and to what).
- Effectiveness: conversion, incrementality, quality scores.
- Risk: errors caught, complaints, compliance issues.
- Cost of AI: licences and usage, including surprise bills.
Gartner predicted in June 2025 that over 40% of agentic AI projects would be cancelled by the end of 2027, partly for unclear business value. Projects with baselines and owners survive reviews.
A quarterly ROI pack for the board
- Marketing contribution (MMM + tests), by channel.
- AI portfolio: each use case with baseline, result, cost and decision (scale, fix or stop).
- Reallocation: what moved from where to where, and why.
- Readiness: score and the gaps being funded.