If your company runs on a calendar year, the 2027 budget is probably in its last lap. The department plans are rolled up, finance has reconciled the totals, and a deck is being prepared for the board. Most of the numbers in that deck rest on a short list of assumptions about rates, demand, input costs and hiring. Most of those assumptions were written in September.
That gap is the whole problem. A budget isn’t wrong because the people who built it were careless. It goes wrong because the world it describes keeps moving after the spreadsheet is saved, and nobody is assigned to notice. Stress-testing a budget before approval isn’t about building a better forecast. It’s about finding out, cheaply and now, which load-bearing assumptions have already been overtaken.
Here is a method you can run in one afternoon with the people who built the plan.
1. Write down the five assumptions the budget can’t survive without
Not fifty. Five. Ask one question of every major line: what would have to be true for this number to hold? You will hear dozens of answers, and almost all of them collapse into a few families: the cost of money, the level of demand, the price of key inputs, the availability and cost of people, and one assumption specific to your industry. Write each one as a sentence with a number in it. “Our blended borrowing cost stays near 6.5%.” “Freight runs at the September rate.” “We can hire 40 engineers at the planned salary band.”
If an assumption can’t be written with a number, it can’t be tested. That’s worth knowing on its own.
2. Attach a signal and a date to each one
For every assumption, name the public signal that would tell you it broke, and note its value on the day the assumption was written. Borrowing costs map to a corporate bond yield or your lender’s reference rate. Freight maps to diesel prices. Hiring maps to jobless claims or postings in your category. Consumer demand maps to sentiment and retail data.
Then pull today’s value. This is the step most planning processes skip, and it’s the one that matters. We track the common ones on a live page measuring what has moved since September 1, the week most planning cycles opened. In this cycle, the 30-year mortgage rate, diesel and corporate borrowing costs all moved while budgets were being built. None of the moves was dramatic. Each one changes a line somebody already filled in.
3. Sort the gaps into three piles
Compare each assumption to its signal and put it in one of three piles. Holding:the signal is where you assumed, or close enough that the line doesn’t change. Drifting: the signal has moved in a direction that matters but not yet by enough to rework the budget. Broken: the line no longer describes the world, and approving it as-is means approving a number you already know is wrong.
Be honest about the drifting pile. That’s where annual plans actually die. A broken assumption gets fixed because someone notices. A drifting one gets approved, then shows up as an unexplained variance in the April review, by which point the decision it should have changed is three months old.
4. Re-size the bets that are hard to reverse
You don’t need to rebuild the budget around every drifting assumption. You need to look at the commitments that lean on it and ask how reversible they are. A marketing program you can pause in a week can ride on a drifting assumption. A facility lease, a multi-year contract or a hiring wave can’t. This is the logic of the Stake Sizing Matrix: size a bet to how hard it is to undo, not to how confident the spreadsheet looks.
In practice this usually means staging a few big commitments instead of approving them whole. Approve the first tranche now and tie the rest to the signal you named in step 2. That’s a small change to the deck and a large change to your exposure.
5. Put an expiry date on every assumption, and book the meeting
The last step is the one that turns a stress test into a habit. Give each of the five assumptions a date by which it has to be argued again, and an owner who brings the signal to that conversation. Then put a standing monthly hour on the calendar, before the budget is approved, where those owners check whether separate signals are starting to point at the same future. We call it a Convergence Review. The name matters less than the fact that it exists before anyone needs it.
An assumption with no owner and no expiry date is the first to rot. The budget doesn’t fail on the day it’s approved. It fails quietly over the next six months, one unexamined line at a time.
What this replaces
None of this requires cancelling the planning cycle you’re in. It requires treating its output as a starting position instead of a finished document. That is the core argument of predictive planning: the annual plan isn’t wrong, it’s late, and the fix is a standing loop that keeps checking it against the world. Scan for the signals, turn them into a small set of stories about how the year could go, stake commitments sized to their reversibility, and steer when the evidence moves. We call it Woodring’s Loop, and it runs every week, not every September.
If you want to know how far your current process is from that, the Maturity Read takes about six minutes. If you want the whole method, it’s in the book.