Unit economics
What has to be true for this to make money
A break-even calculation takes thirty seconds and nobody needs help with the arithmetic. The useful question is which of the numbers you fed it you actually need to be right about — because most business cases are one assumption wearing a spreadsheet, and finding that assumption is the whole job.
Modelled on a real engagement. The payload assumption is the one everything turned on — try moving it and watch the sensitivity order change.
Volume can fall 20.1% before this stops breaking even, and you are running at 80.0% of capacity.
The inputs
Where the money goes, per tonne collected
Which assumption the answer turns on
Each input moved down and up by the same amount, everything else held still. The order is the finding: argue about the top row, and stop arguing about the bottom one.
Where volume is heading, and how sure that is
Twelve periods of history, six projected. The band is the part worth looking at: it comes from the model's own errors on the history it has already seen, so a series that has been erratic produces a visibly wide band rather than a confident line.
- 272
- central projection, month 18
- 233 – 310
- 95% interval by then
- Yes
- clears break-even even at the bottom of the interval
What this cannot do: anticipate a contract starting or ending, a price change, a competitor, or a season it has never seen. It extrapolates the shape of the past. Every forecast does, and the ones that do not say so are the ones to distrust.
What each number means
- Contribution margin per unit
- Price per unit minus every variable cost per unit. What one more sale actually leaves behind once the costs that scale with it are paid.
- Break-even volume
- Fixed costs divided by contribution per unit. Below it you are funding the business; above it you are running one.
- Margin of safety
- How far volume can fall before break-even. The number worth putting in a board paper, because it answers "how wrong can we be".
- Sensitivity analysis
- Each input moved by the same amount to see which one moves the answer. Usually one does and the rest barely matter.
The collection preset is a real engagement
An operating plan showed a loss and the conclusion drawn from it was that the contract was unviable. The plan rested on a payload figure nobody had checked against the vehicles in the yard: it had been understated by nearly a third, and payload sits upstream of route counts, fuel, driver hours and disposal trips, so correcting it moved everything at once. The corrected model showed a profit.
Nothing about that required a better spreadsheet. It required noticing which input the answer turned on, which is exactly what the sensitivity ordering above is for. Load the collection preset and move the costs around — the top row changes, and the argument changes with it.
Where a model like this stops
A single average unit is a simplification. Real businesses have a mix, and the mix moves. If one product subsidises another, one blended contribution margin will hide it.
Fixed costs are only fixed over a range. Double the volume and you buy another vehicle, another shift, another site. The step is invisible to a straight line through break-even.
The forecast extrapolates shape, nothing more. It cannot know about a contract starting, a price change or a season it has never seen. The widening interval is the honest part of it.
Sensitivity is one input at a time. Costs move together in practice — fuel and driver hours rarely rise alone — so treat the ordering as a guide to where to look, not a complete risk model.
When the real model is bigger than one page
Multiple sites, a product mix, step costs, a rate to negotiate, a board paper that has to survive questions. We build those so the assumptions stay visible and every figure traces to a source — which means the answer changes when an assumption changes, rather than needing a rebuild.
Questions people ask
- What is the break-even formula?
- Break-even volume equals fixed costs divided by contribution margin per unit. Contribution margin per unit is the selling price minus every variable cost per unit. So a business with 11,500 of fixed costs a month and a contribution of 60 per unit breaks even at 192 units a month.
- What is contribution margin, and how is it different from gross margin?
- Contribution margin is price minus the costs that vary with each unit sold, so it tells you what one more sale leaves behind. Gross margin subtracts cost of goods sold, which usually mixes in costs that do not vary per unit. For a break-even question, contribution margin is the one that matters, because only it scales with volume.
- What is margin of safety?
- How far volume can fall before you stop breaking even, expressed as a percentage of current volume. If you sell 240 units and break even at 192, the margin of safety is 20 per cent. It is the number worth putting in a board paper, because it answers how wrong the forecast can be before the decision changes.
- What is sensitivity analysis and why does it matter more than the break-even number?
- Sensitivity analysis moves each input by the same amount and records how far the answer shifts, which ranks the inputs by how much they matter. Break-even tells you the answer under one set of assumptions; sensitivity tells you which assumption to go and verify. Usually one input dominates and arguing about the rest is wasted effort.
- Can a business have no break-even point?
- Yes. If contribution margin per unit is zero or negative, every sale loses money before any fixed cost is covered, so no volume breaks even. Growth makes that worse rather than better: the price or the variable cost has to change.
- How accurate is a demand forecast like this?
- The projection here uses Holt's linear trend and reports a 95 per cent interval derived from its own errors on the history it has seen, so a series that has been erratic produces a visibly wider band. No method of this kind can anticipate a contract starting, a price change, or a season absent from the history.
Try it on your own file first
Every tool here runs in your browser and uploads nothing, so you can see what we would see before telling us anything at all. When it is worth a conversation, send the real extract and the question you actually need answered.