Every November, the same ritual plays out. Someone opens last year's spreadsheet, bumps most of the numbers up by three or four percent, adjusts a couple of lines they remember arguing about, and calls it next year's budget. In my audit work, we had a name for that: relying on a control nobody ever tested.
Here is what audit work taught me: a budget line is not a number. It is an assumption wearing a number. Auditors don't ask whether a figure is big or small. They ask what it is based on, and whether that is still true. Budget season is your chance to ask before the year tests them for you.
Read every line as a claim
Pick any line in your draft budget and finish this sentence: "We believe this will cost what it costs because..." If the honest answer is "because that's what we put last year," you have found an assumption, not a plan.
Three claims show up in almost every budget I ever reviewed. The first is the inertia renewal: subscriptions and tools that renewed because nobody owns the cancel button. Seats were bought for forty people; twenty-two log in. A tool was bought for a project that ended in the spring and has been renewing ever since. None of this is fraud. It is just money leaving on a schedule nobody re-approved.
The second is the untested growth assumption. Revenue will grow twenty percent. A product line will expand. Maybe it is all true — but in most budgets I have seen, the growth figure was decided in a meeting two or three years ago and copied forward ever since. Ask who tested it, and against what, and the room gets quiet.
The third is headcount math built on last year's org chart. Teams merge, roles change, positions sit open for nine months and the work still gets done. Yet the budget funds the old shape of the team, because the old shape is what the spreadsheet knows.
Pick three categories and start them at zero
There is an old method called zero-based budgeting, and all it means is this: instead of starting from last year's number and adjusting, you start from zero and make every dollar argue its way back in. Done across a whole company, it is slow and painful, and I am not recommending that. Done across three categories, it is the most useful hour in budget season.
Choose categories where assumptions pile up. Software and subscriptions are the obvious first pick. Contractors and outside services are usually second, because engagements outlive the problem they were hired to solve. Travel, training, and marketing spend reward the same treatment. For each one, write down what you actually need next year, priced as if you were buying it for the first time — because in a real sense, you are.
Put the cushion where everyone can see it
Most budgets contain contingency. It is just hidden. Managers pad every line by ten percent because experience has taught them the budget will be cut and something always goes wrong. I understand the instinct. I have done it myself.
The problem is that padding spread across forty lines makes all forty numbers unreliable. Nobody — including whoever built the budget — can later tell the real estimates from the cushions. So move the cushion into the open: one visible contingency line, with three things written next to it — what it is for, who can release it, and what counts as a genuine surprise rather than a miss in estimating. A surprise is a vendor failure or a regulatory change. A miss is the growth assumption nobody tested. Contingency covers the first. It should never quietly absorb the second.
What to do in one sitting
You do not need a task force or a new system. You need the draft budget, last year's actual spending, and about ninety minutes.
- Set the budget next to the actuals. Put twelve months of real spending beside each proposed line. Circle any line that came in twenty percent or more under budget two years running. That gap is an assumption asking to be interviewed.
- Ask the re-buy question. Pull every subscription and recurring service off the card or out of payables. For each, ask the person closest to it: if this renewed tomorrow at full price, would you buy it again today? "No" and "I don't know" both mean the line goes to zero until someone makes the case.
- Zero-base your three categories. Software, contractors, and one more of your choosing. Rebuild each from need, not from last year's actuals.
- Make every growth claim say its reason out loud. Next to each growth number, write in plain words what has to be true for it to happen. If nobody can write that sentence, the number is a wish, and wishes get a smaller line.
- Gather the padding into one line. Strip the quiet ten percents out of the individual lines and park the total in a labeled contingency with a named owner and written release rules.
Do those five things and your budget stops being last year's spreadsheet with a new date on it. It becomes what a budget is supposed to be: a short, honest list of what you believe, written down while there is still time to check whether it is true.