Every budget model has a tab called Assumptions, or Drivers, or Inputs, depending on who built it. When did anyone last open it on purpose? At most companies the answer is last September, when it got rolled forward from the year before: a growth factor on revenue, a few points of inflation on costs, headcount off the org chart, and the revolver at whatever the bank charged in the spring. Then everyone spent October arguing about the headcount.
Budget assumptions built that way hold up as long as the outside world stays put, and in 2026 it hasn't. The Fed raised rates on September 16, its first increase since 2023. Canada started charging 50% tariffs on a long list of U.S. steel and aluminum products on September 8. And the ISM prices index for manufacturing came in at 71.1 for August. Anything over 50 means more purchasing managers are paying more than are paying less. 71 is a lot of them.
So before the 2027 budget goes to the board, give an afternoon to the handful of assumptions that depend on things nobody in your building controls. Start with these, in roughly the order they go wrong.
Which budget assumptions break first
Input costs. Ask where the number came from. If it's last year's average plus inflation and you buy metal, energy or freight, it was stale before anyone saved the file. A purchasing manager quoted in the August ISM report said energy, steel and labor costs were "increasing very quickly," and when we run briefings on manufacturers in the fall the same story keeps turning up: customer prices got set in the spring, and the costs underneath them climbed all summer.
Price versus cost. Easy to miss, because the two numbers live on different tabs. In the Richmond Fed's CFO Survey, fielded in late May and early June, the median CFO expected their own unit costs to rise 4.0% this year and their own prices to rise 3.0%. A full point of margin going the wrong way, and that's the middle of the pack. If your budget has price and cost moving together, you're more optimistic than the typical CFO in the country. Maybe you have a reason to be, and if so, fine. Just know that's the bet you're making.
Interest expense. On a floating revolver, every quarter point on $20 million drawn is $50,000 a year, before you get to the term loan. The Fed said the hike "will support a timelier return" to its 2% inflation goal, and that is not how a committee talks when it thinks it's finished. Budget the rate you could explain to the board if it went up once more. And if a maturity sits inside the next 18 months, the refinancing is mostly decided before you ever call a lender, so that clock is already running.
Volume. Run-rate volume is a bet that your end markets do next year what they did this year. Test it against your own customers rather than the national indexes, because the headline numbers this year are carried by a couple of sectors most mid market companies don't sell into.
Wages. The same Richmond Fed survey put median expected wage growth near 3.1%. Treat that as a floor. National figures smooth over the thing that hurts: a local market where a new plant opened ten miles away and started hiring welders at two dollars an hour more than you pay. HR usually knows about this before finance does. Go ask what the last five offers for your hardest roles came in at, and how many of those people took a counteroffer instead.
Trade. If you ship into Canada, some of your products got a lot more expensive for your customers on September 8. If you buy anything with steel or aluminum in it, the Section 232 duties are still sitting in your bill of materials. Neither one is in last year's actuals, so neither one is in a rolled-forward budget.
How to test budget assumptions before the board does
Add two columns to the Assumptions tab. Column one: what has to be true outside the company for this number to hold. Column two: the one public series you'd check to see whether it still does, which might be a steel price index, the Fed's next statement, construction spending in your region, or your largest customer's own guidance. If you can't name a series for a line, write that down too. It means the number is a guess, and you should at least know it's a guess.
For a metal fabricator, one row might read: steel costs up 5% in 2027. Has to be true: mill prices hold near summer levels and no new duties land on the grades we buy. Series to watch: the producer price index for steel mill products each month, with the ISM prices index as an earlier read. Ten minutes to write. Most models don't have it, because the tab was built to hold numbers and nobody ever added a column for why.
Then attach a trigger to each row. Something like "if the ISM prices index is still above 65 at the January close, we reopen the input cost line." The exact level matters less than setting it now, in a calm month, instead of in March with a variance on the page and the board waiting.
Eisenhower told a room of defense planners in 1957 that plans are worthless, but planning is everything. I think about that line most budget seasons. The budget is going to be wrong somewhere. Every budget is. What you get to decide is whether you're the first person in the room to know where, and a CFO who shows up with the assumptions, the series behind each one and the trigger levels is in a very different position from one defending a single number. The board can tell the difference even when they can't say what it is.
If you'd rather not build the watch list yourself, a free briefing on your company reads the outside conditions against your own budget lines and tells you which ones moved.
Why March is when this shows up
Budgets don't usually break in January. They break at the first real checkpoint, the Q1 close or the covenant certificate or the first reforecast, and that lands in March or April. It's also when fixing it costs the most, because now you're revising a number the board already approved and asking them to trust the replacement. Finding the problem in November is cheaper in every way, and the biggest saving is one nobody puts in a model: the board still believes the next number you bring them.
CFOmarketIQ watches these outside conditions against your own budget lines every week, and the first briefing is free.
Questions CFOs ask about budget assumptions
What are budget assumptions? Can you give examples?
They're the conditions your numbers depend on. Some are internal, like headcount and hiring dates. The risky ones are external: the price increase you expect customers to accept, input cost inflation, volume in your end markets, the interest rate on drawn debt, exchange rates if you export. A good one reads like "steel costs up 4%, and we recover 3% of that through price."
What inflation rate should I use in a 2027 budget?
Not CPI. Build it from your own basket, meaning the materials, freight, energy and labor you buy, weighted by how much you spend on each. Then check it against something outside, like the Richmond Fed's CFO Survey, which asks CFOs what they expect for their own costs and prices. Where the exposure is large, budget a range instead of one number.
How often should budget assumptions be updated?
Look at them monthly with the close and formally at each quarterly reforecast. Better still, make it trigger-based. Decide in advance which outside move reopens each assumption, so the update happens when something changes rather than when the calendar says so, since the two rarely line up.
A note on all of this: it is general information, not financial, legal or tax advice. Run anything you plan to act on past your own advisors first.