Building the budget line by line

With a sales line for each month, the rest of the budget is mostly arithmetic — but only if you have first sorted your costs into the right kinds. Restaurant costs behave in exactly three ways, and once each line is labelled, twelve months of budgeting is a couple of formulas copied across. Get the labelling wrong and every month after a busy one will look like a disaster.

🧱 Three kinds of line, and no more

A fixed line is money that leaves whether you sell anything or not: rent, insurance, software, the accountant, the licences. In the budget it is the same figure in all twelve columns. A variable line moves with sales and is budgeted as a percentage of them: food and drink above all, plus card processing and takeaway packaging. And in between sits labour, which is part one and part the other, and which is where most restaurant budgets go wrong.

Here is the example restaurant's structure, taken straight from its statement in the financial reporting course.

Check it against the ordinary month you already know: sales of $50,000 give food of $16,000, labour of $8,500 plus $8,500 which is $17,000, rent of $5,000 and everything else at $9,000. Total $47,000, result $3,000. Same statement, now written as a machine that will take any sales figure you feed it.

Two shortcuts are worth knowing. The fixed money in this restaurant is $22,500 a month — rent, everything else, and the fixed half of labour — and the variable share is 49%, which leaves 51% of every peso to cover it. Those are the same two figures that gave the break-even of about $44,100 in the previous course, and they are about to do a lot of work.

👥 Labour is the line that is neither

Budgeting labour as a flat percentage of sales is the most common error in this whole exercise, and it is comfortable precisely because it always looks right on the annual total. It goes wrong month by month. A restaurant that opens its doors pays managers, a minimum crew and the hours it will roster whether forty people come or ninety; only the hours above that follow the volume. So in a quiet February, a flat-percentage budget promises a labour cost that no roster you could actually write would achieve, and the month misses its plan for a reason that was baked in before it started.

The split version tells the truth. In this budget February plans $8,500 of core labour plus 17% of $41,000, which is $15,470 — and $15,470 on $41,000 of sales is nearly 38%, against 34% in an ordinary month. Nothing has gone wrong there. That is what a quiet month does to a business with people in it, and seeing it in the plan is what stops you from firing a manager in March over a number that was always going to happen.

You can find your own split without any special data: look at the lowest-volume week of last year and ask what the payroll was, or add up the hours you would roster on your quietest realistic day. That figure, times the weeks in a month, is close enough to your core. The same idea applies to any line with a floor — a minimum utility bill, a service contract with a base charge — but labour is the one big enough to matter.

🎯 Profit is a line you decide, not what is left over

Add up the machine and this restaurant's year comes out at $620,000 of sales, $573,800 of costs and $46,200 of profit — about 7.5% of sales. That is the budget it built from the bottom up, and there is nothing wrong with it. But it is worth doing the exercise the other way round at least once, because the answer is usually surprising and always specific.

Suppose the owner needs $60,000 next year, because of a loan, a family plan or simply a decision that the business should pay better. The arithmetic runs backwards from there: fixed costs of $22,500 a month are $270,000 a year, so with 51% of every peso surviving to cover them, the sales needed are $270,000 plus $60,000 divided by 51%, which is about $647,000. That is $27,000 more than the plan, or $2,250 a month, or ninety covers a month at a $25 check — three more people a day.

Three more people a day is a target a floor team can hold in their heads, and it came from a number the owner chose. The same gap can also be closed from the other end: $13,800 a year of fixed cost removed — $1,150 a month of software, insurance, subscriptions and repairs — does exactly the same thing to the bottom line without selling a single extra plate. Usually the answer is some of each, and the point of running it backwards is that you find out in November rather than the following December.

One warning while you are here. Do not budget a food cost of 30% because 30% would be nice, when you have run at 32% for two years. A better percentage in the plan is a decision that something will change — a supplier, a portion, a recipe, a price — and that something belongs to the cost control course and needs to be named. An improvement typed into a spreadsheet without a cause behind it is not a plan, it is a wish, and it will show up as an unfavourable variance every single month.

📊 The year on one screen

Run every month through the machine and the plan looks like this. This is the sheet — the whole first template, and the thing topic 5 will compare against.

Two things jump off it that no single month's statement could ever have shown. January and February are budgeted to lose money — $570 and $1,590 — and they are not mistakes: this restaurant has always lost money in those months and always will, because $43,000 and $41,000 are below its break-even of $44,100. And more than half the year's profit is made in July, August and December, which between them plan $25,830 of the $46,200.

Both facts have consequences that have nothing to do with accounting, and they are the subject of the next topic: the money that survives February is money that was earned the previous August, and whether it is still there in February is a decision somebody makes — usually without noticing they are making it.