A consumer brand's data lead turned a list of AI ideas into a budget the CFO could sign

The data lead had a weekly business review, a pipeline in production and a spreadsheet of agentic ideas that leadership kept parking as research. The budget deadline was weeks away. We ran the million-dollar roadmap with him: each leader names what they would fund from a hypothetical $1M, every candidate gets an estimated return and effort with the arithmetic shown, and the first ask is the three items that share a foundation.

What was on the list

An operations control tower across procurement, warehouse and logistics. Executive explainability, the 'why did sales move' layer over the existing review. A customer-service assistant that captures case notes while the agent stays on the phone. Demand planning that respects warehouse capacity. Customer 360 across retail and direct.

Each one had a champion. None had a number next to it, so every budget conversation became a conversation about whether AI was worth it at all.

What changed

The roadmap put a return estimate and an effort score next to each item, with the inputs visible. The first ask was sized at about 15% of the whole: the three items that share one semantic layer, so the second quarter starts from a working base. The pitch to the CFO became three projects with acceptance tests, not an open-ended AI budget.

The data lead took the AI vertical under his own function, with resources attached, rather than more scope at the same title. Governance landed where it usually does: under data.