Build vs buy: should you own your business software?

The honest framework I use to decide whether a business should buy off-the-shelf SaaS or own a custom system, and where owning almost always wins.

Build versus buy is usually framed as a cost question, and that framing is where most owners go wrong. Buying is cheap to start and expensive to keep. Building is expensive to start and cheap to keep. Which one wins is not about the first invoice, it is about how long you will run the business and how central the workflow is to how you actually make money.

I sit inside owner-operated businesses and make this call for a living, so here is the actual framework I use, not the vendor version of it.

Buy when the capability is a commodity

Some capabilities are genuinely solved and identical for everyone. You should almost always buy these, at least at first. The test is simple: if your version of the workflow is the same as every other business your size, and the tool does not touch the thing that makes you money, rent it and move on. Paying a small subscription to avoid building undifferentiated plumbing is a good trade.

The trap is that "commodity" is a moving line. A tool that was commodity at five people becomes a liability at fifty, because per-seat pricing turns a rounding error into a real line item, and because your process quietly grows more specific than the template allows.

Build when the workflow is how you win

The workflows worth owning are the ones that are actually a differentiator: how you quote, how you dispatch, how you onboard, how you deliver. When you rent these, you are forced to run your business the way a generic template imagines a business runs, and you spend weeks bending the tool to fit and still not quite getting there.

When you own them, the software fits because it was built to fit. There is no per-seat meter growing against your headcount, no field you need that "is not possible," and no vendor sitting between you and your own data. The build costs more up front and then the cost stops, which is the opposite of a subscription.

What the math actually looks like

Almost every build-versus-buy spreadsheet I have been handed compares a subscription's monthly price against a build's quoted price and stops there. That is the wrong comparison, because the two costs behave differently over time. A subscription is a flat line that never ends and tilts upward every time you add a person, add a tool, or hit a renewal. A build is a spike that ends. The real question is not which number is bigger this month, it is where the two lines cross and whether you will still be running the business after that point.

Three things move that crossing point, and none of them appear on a pricing page. Headcount, because per-seat meters compound across every tool at once rather than one at a time. Tool count, because eight subscriptions each doing one job cost more than the sum of their invoices once you count the glue holding them together. And time, because a build cost stops and rent does not.

The costs that never reach the spreadsheet

Bending your process to fit a template, paid in your team's hours every week instead of in a monthly invoice.
Integration work between tools that were never designed to talk to each other, which you maintain and the vendors do not.
Switching cost, which grows quietly the entire time you rent, because leaving means extracting your data in whatever shape the vendor allows.
The requests your team stops making because the answer has been no often enough that they route around the tool instead of through it.

The three questions I actually ask

Is this workflow a differentiator, or plumbing? Differentiators lean build; plumbing leans buy.
How does the price scale? Flat pricing is friendlier to rent; per-seat and per-usage pricing punish you exactly as you grow.
Who owns the data and the logic? If losing the tool would strand your records or your process, that is an argument for owning it.

What owning looks like in practice

This is not theoretical. In my own company I rebuilt the business into one system it owns: 16 departments, 14 live bases in production, 21 SaaS tools replaced, and $0 ongoing license cost, with 25+ AI agents doing real work inside it. That was roughly a 200-person operation, and almost none of it needed a subscription once the workflows lived in one place we owned.

The point is not to build everything. It is to be deliberate: buy the commodity, own the differentiator, and stop paying rent forever on the parts of your business that are actually yours.