Finance.

The framework

1. 01

   Unbundle the three jobs the budget was doing at once.

   Write them out separately and keep them apart. The target is what you are trying to achieve: ambitious, owned by a name, allowed to be a stretch. The forecast is what you honestly expect: unarmed, revised whenever the world revises itself, carrying no consequence for the person who wrote it. The resource decision is what money actually gets released, and against what condition. Separated, each one can be straight with you, because none of them is being asked to do another's work. Ask which number you wrote in the last forecast you submitted and which number you privately believed; if those differ, the fusion is already doing its work on you. Let a single number resume all three roles and you've rebuilt the budget under a newer name, with more meetings attached.

2. 02

   A horizon that never runs out.

   Choose a horizon long enough to be decision-useful and short enough to stay honest — four to six quarters suits most businesses — and re-forecast it on a fixed rhythm, adding a fresh period each time you close one. The horizon never shortens. You're therefore never steering by a plan with eleven weeks left in it, which is the state most companies are quietly in by the time spring arrives. The rhythm matters more than the length: a rough forecast refreshed every month will beat a meticulous one refreshed every year, in any market that moves at all.

3. 03

   Five numbers move the outcome; the rest are consequences.

   A rolling forecast is not a re-typed budget. Pick the five to eight quantities that genuinely move the outcome — orders, average order value, churn, price, headcount, lead time, cost of acquisition — forecast those, and let the financial lines fall out of them. This keeps each cycle to hours rather than weeks, which is the only reason anybody will actually do it monthly. It also makes disagreement productive: two people arguing about next quarter's win rate are arguing about something knowable, whereas two people arguing about a revenue total are just comparing temperaments.

4. 04

   Score your own forecasts, or the whole exercise becomes theatre.

   Each cycle, record what you forecast alongside what actually happened, and keep the log. Within a few rounds you'll know your own bias, and it will have a direction, because forecasting error is rarely random — the planning fallacy of Day 78 shows up in money as reliably as it shows up in timelines. That correction is worth more than any refinement to the model underneath it. It also settles the argument the rolling forecast exists to end: the number doesn't have to be defended, because it's now being measured for accuracy rather than for loyalty.