Software Margins Are Coming Home
Software companies run gross margins around eighty percent. That number is quoted admiringly in investor decks, and it should be read differently by the people paying the bills.
If your vendor earns eighty points of gross margin on your subscription, then roughly eighty cents of every dollar you send is not covering the cost of serving you. It is profit — theirs. For twenty years that was a fair trade, because the alternative was hiring engineers you could not afford.
When the alternative gets cheap, that eighty cents becomes contestable.
- A vendor’s gross margin is, from your side of the table, the premium you pay to not build it.
- That premium was fair when building was expensive. It is now negotiable in narrow categories.
- Internal software converts a recurring operating expense into a one-time build plus small maintenance.
- The saving is real but must be booked honestly — maintenance is a cost even when it is invisible.
- This changes the shape of the P&L, not just its size.
The arithmetic, plainly
| Rented tool | Built tool | |
|---|---|---|
| Year one cost | Full annual subscription | Build effort, then near zero |
| Year three cost | Same or higher, plus seat creep | Maintenance only |
| Cost direction | Rises with headcount and usage | Falls per unit as you grow |
| Who owns improvements | Vendor roadmap | You, immediately |
| Hidden cost | Price rises at renewal | Attention and key-person risk |
The second row is where it compounds. Rented software gets more expensive as you succeed, because pricing is indexed to seats or usage. Built software gets cheaper per unit as you grow, because the build cost is already spent. Those two curves diverge for as long as you own the business.
Renting software means your cost base grows with your success. Owning it means your cost base stays still while your revenue moves.
The honest accounting
It would be dishonest to book internal tooling as free. It is not. It consumes attention, it creates key-person risk, and when it breaks there is no support contract. Those are real costs that simply do not arrive as an invoice — which makes them easy to underestimate right up until they matter.
We book them explicitly. Any internal tool carries a named owner and an assumed maintenance load. If we cannot name both, we treat the build as more expensive than the subscription, because it is.
What it does to the shape of the business
The interesting effect is not the saving. It is that operating leverage improves permanently. A brand carrying a large stack of per-seat tools has a cost base that scales with the team. A brand running mostly owned tooling has a cost base that scales with almost nothing.
For a holding company that is decisive, because the tooling built for one brand carries to the next at close to zero marginal cost. The second brand inherits the machine. That is the entire compounding argument, and it only works if you own the machine.
LAMPWORK operating analysis. Gross margin figures reflect commonly reported SaaS benchmarks rather than any specific vendor.
Own a brand this applies to?
We talk to founders at every stage — long before they're ready to sell.
Start a Conversation →