Every business past its first few hires runs on rented software. Email, files, chat, CRM — per seat, per month, cancel any time. The model is so normal that asking “should we own any of this?” sounds almost eccentric.
It shouldn’t. It’s a real decision with real trade-offs in both directions, and most writing on it is sales copy for one side or the other. Here is the version we’d want to read.
What renting genuinely gets you
Start with the honest case for SaaS, because it’s strong.
Speed. You can open a Google Workspace account this afternoon and have working email, calendars, and storage before dinner. No servers, no configuration, no decisions you’re not ready to make.
No maintenance. Patching, uptime, spam filtering, security response — handled by vendor teams larger than most companies’ entire staff. You will not out-operate Google at running mail servers, and you don’t have to.
Cheap at small scale. At three seats, per-seat pricing is a rounding error, and the ability to cancel next month is worth more than any optimisation.
Everything connects. Mainstream tools integrate with each other out of the box, because every vendor’s roadmap includes the others.
If you’re a three-person company that might pivot twice this year, renting is probably right. That’s not a concession; it’s the honest read.
What compounds against you
The same model reads differently at twenty seats and five years in.
The bill scales with headcount, not usage. Every hire adds a licence on every tool, and no licence ever gets cheaper. Owned infrastructure has a different shape: a server doesn’t care whether ten people or forty log into it.
You don’t set the price. Subscription prices are revised at the vendor’s discretion, and your negotiating position weakens the deeper your data and workflows are embedded. You can’t credibly threaten to leave a system your whole company lives in.
Your data lives under their terms. Export tools exist, but they define what you can take: files usually travel well; sharing structures, version histories, and integrations usually don’t. (That’s a big enough topic that it gets its own article.)
Features migrate upward. A capability you rely on can move to a higher tier at renewal. Nothing dishonest about it — it’s the model — but it means your costs can rise without you adding a single seat.
What owning actually costs
Ownership isn’t free after setup, and anyone who implies otherwise is selling something.
There’s a real one-time engineering cost to set up private email, file, and hosting infrastructure properly. There’s a monthly server bill — typically fixed, and indifferent to hiring. And there’s maintenance: updates, backups, monitoring. Someone has to do it — your team, or an engineer you keep on a support agreement. Skipping it is not an option; badly-run owned infrastructure is worse than well-run rented infrastructure.
The honest summary: owning replaces a permanent, headcount-linked operating cost with a front-loaded cost plus a smaller fixed one. It trades “someone else’s problem” for “your asset, your responsibility.”
The two questions that decide it
Most of the decision collapses into these:
1. What does your cost curve look like? Seats × tools × months, projected at the headcount you expect in three years. Compare it with a fixed infrastructure cost plus maintenance. At three seats the comparison is boring. Past ten or fifteen, it usually stops being close — but run your own numbers, not anyone’s marketing table.
2. Where must your data live? If confidentiality obligations, regulation, or plain principle require your files and mail to sit on infrastructure you control, the rental model is a bad fit at any price — and if none of that applies, ownership has to win on economics alone.
Neither answer makes the other side wrong. A design studio with heavy client churn and a consultancy holding privileged documents can both be reasoning correctly and land in different places.
If you’re at the size where the numbers deserve an hour of attention, give them the hour. Ask what your stack costs at your projected headcount in three years. Ask what a full export of everything would look like this Friday. The answers vary from business to business; the questions don’t.