This is the decision underneath most automation spending, and it is usually made by default rather than on purpose.
Licence when the job is common to your whole industry and being the same as everyone else is fine. Build when the job is how you are different, because that is the only part a competitor cannot buy next quarter.
The test is not cost. It is whether doing this particular thing better than your rivals is worth anything to you.
A licence is easy to approve because it is small and monthly. A build is hard to approve because it is larger and arrives at once. That asymmetry, rather than any analysis, decides most of these choices.
The comparison worth doing is simple and you can do it on paper. Take the per seat price, multiply by the seats you will genuinely have in three years rather than today, and multiply by thirty six. Compare that with a build plus three years of running it. Do not take our word for which way it comes out, because it depends entirely on your seat count and it will not be the same answer for a team of eight as for a team of eighty.
What the arithmetic will not tell you is the part that actually matters, so do it and then set it aside.
Licensed software is available to your competitors on the same terms, at the same price, tomorrow. That is not a criticism of it. For work that is common to your industry, having the same capability as everyone else is exactly right, and building your own version of a solved problem is a waste of money and attention.
But it means a licence can never be a source of advantage. It can only stop you being behind.
So the question is not what this costs. It is whether doing this particular job better than your rivals is worth anything. If it is, that job is the one to own.
| If the job is | Then | Because |
|---|---|---|
| Payroll, accounting, email, calendars | Licence it | Solved, identical everywhere and nobody has ever won on it |
| Common to your sector and done the same way by everyone | Licence it | Parity is the goal, and parity is cheap |
| How you actually differ from the firm down the road | Own it | It is the only part a competitor cannot buy |
| Built on data only you hold | Own it | The data is the asset and the system is how you use it |
| Something you would struggle to describe to a vendor | Probably own it | If it were standard, a product would already exist |
Owning is not free after the build. Somebody has to maintain it, somebody has to own it inside the business, and the run cost grows with usage. A firm that sells you a build without saying that is not being straight with you.
The honest version of the pitch is this: you are trading a renewing cost you cannot escape for an asset you have to look after. That is a good trade when the thing is genuinely yours and genuinely differentiating. It is a poor trade when you have rebuilt something you could have rented.
The worst outcome is not choosing wrong. It is building something you then cannot maintain, which leaves you with the cost of both options and the benefit of neither. Three things prevent it:
Do those and the decision becomes reversible, which is worth more than getting it right first time.