Finance.

The framework

1. 01

   Not everything goes to zero, and pretending otherwise kills the method.

   Zero-basing the whole business at once is how this dies: an eight-week exercise administered by finance that ends in a flat percentage taken off every department, which is the incremental budget with worse manners. Scope it instead. Contractual obligations, statutory costs and the direct costs of delivering what you sell aren't up for re-argument this cycle. What is up is the discretionary base — software licences, agencies and retainers, travel, subscriptions, memberships, the tools and services still renewing on a card somebody set up years ago. Take one slice, do it properly, and rotate the slice each year.

2. 02

   Make every activity apply for its own money.

   A decision package is short, and it's written by the person who runs the activity rather than by whoever owns the spreadsheet. Five things: what this is for in one sentence, what measurably happens to whom if it stops, how you'd know it's working, what else the same money would buy, and the cost at three levels — below today's spend, at today's spend, and above it. The bottom level does the real work, because it forces an owner to describe the activity on less money, and a striking number of them describe something that would be fine.

3. 03

   One list, every department, argued in the same room.

   This is the step that separates the method from a cost-cutting drive, and the step almost everyone skips. Every package from every department goes into a single ranked order, so the marketing retainer sits directly above or below the extra support hire and the two get argued by the same people at the same table. Ranking is uncomfortable precisely because it makes colleagues compare things they'd much rather keep in separate budgets. A department that ranks only its own packages has learned nothing; it has sorted the things it was always going to keep.

4. 04

   Draw the line where the money stops, then hold it.

   Run the available money down the ranking until it's gone, and leave it gone. What follows is the pressure test: the owner of an unfunded package returns with the same activity at a slightly smaller number, or reframed as a risk, or attached to a customer's name. Answer with the ranking rather than with a verdict on the person. And write the reasoning beside each funded line while you still remember it, because eleven months from now that note is the only thing standing between this exercise and the anchor from Day 74 — last year's number, quietly doing this year's thinking.