What negative unit economics means in practice
Take one customer, one order or one delivery. Add everything that transaction genuinely costs, including the variable people cost that startups habitually exclude. If the revenue from that unit does not cover it, the unit is negative.
Marketplaces, delivery and anything with a physical leg are the classic cases. Software is not immune. Support heavy products, generous free tiers and inference costs on AI features all create per unit losses that look like fixed costs on a spreadsheet.
- Delivery or fulfilment costs that do not fall with volume
- Support load that scales linearly with customers
- Third party costs per transaction, such as inference or payment fees
- Discounting used to drive growth that never gets withdrawn
Why scale usually makes it worse
The standard hope is that volume fixes the margin. Sometimes it does, when the cost is genuinely fixed and being spread. Often it does not, because the dominant cost is variable and simply multiplies.
Worse, growth can raise variable costs. Expanding beyond a dense first city raises delivery cost per drop. Moving beyond early adopters raises support cost per account. The economics degrade exactly when you are least able to stop.
Selling a business whose model did not work
The honest framing is usually the most saleable one. The concept was proven and the economics were not, in this market, at this density, with this cost base. That is a genuine invitation to a buyer with a different cost structure or a denser geography.
The technology is often entirely sound. Routing, matching, recipe or logistics software built for a model that failed still works, and a buyer can point it at a model that does.
Common questions
Is negative unit economics always fatal?
No, but fixing it requires a structural change such as pricing, cost base or geography. Volume alone rarely rescues a genuinely negative unit.
How do I present bad unit economics to a buyer?
With the actual numbers and the reason. Buyers who can fix the economics are looking for exactly this, and they will find the numbers in diligence regardless.
Does this affect what my startup is worth?
It moves the valuation from a revenue multiple to an asset basis. The code, users and data are still worth real money.