The SaaS you cannot cancel is usually one week of work
There is a line item on your card statement that renews every January, costs somewhere between $200 and $800 a month, and exists because in 2022 you needed one specific thing done. Your team still uses that one specific thing. They do not use the other forty features, the analytics module, the workflow builder, or the mobile app. You have thought about cancelling it twice and both times stopped, because nobody could say with confidence what would break.
That hesitation is the product. It is what you are paying for.
Rent versus own, done honestly
The usual build-versus-buy comparison is rigged in both directions. The vendor's version ignores that you pay forever and the price only moves up. The engineer's version ignores that software you own is software you maintain. Do it with both columns and the picture is less dramatic and more useful.
- Keep renting
- Build it once
Show the data
| Year | Keep renting | Build it once |
|---|---|---|
| Y1 | $4,800 | $9,000 |
| Y2 | $9,840 | $9,900 |
| Y3 | $15.1K | $10.8K |
| Y4 | $20.7K | $11.7K |
| Y5 | $26.5K | $12.6K |
The interesting part of that chart is not that owning wins eventually — it is where the lines cross. Somewhere in year two, for most tools in the low four figures. If the tool is sticking around longer than two years, the rent column is the expensive one. If it is a bet you might abandon next quarter, keep renting; that is what rent is for.
Four costs the spreadsheet leaves out
- <strong>Seat creep.</strong> The price per seat is fixed; your headcount is not. A tool that made sense at six people is a different conversation at twenty, and the vendor knows exactly when you crossed the tier.
- <strong>The workflow tax.</strong> Your process now has the vendor's shape, not yours. Every workaround your team invented to fit their model is a cost you have already paid and stopped counting.
- <strong>Your data, at their latency.</strong> If the numbers you need live in someone else's system, every report that touches them is an export, a sync, or a paid API tier.
- <strong>Cancellation risk.</strong> Not the risk that you cancel — the risk that they do. Sunsetting a plan, tripling a price, or being acquired is entirely outside your control, and the migration lands on the worst week of your year.
Do not build these
I make a living writing software and I will still tell you to renew most of your stack. The tools worth renting are the ones where the vendor is absorbing a problem you genuinely do not want:
| Keep renting | Build and own | |
|---|---|---|
| Regulatory surface (payments, payroll, tax) | Yes — let them carry it | No |
| Security and compliance burden | Yes — SOC 2 is a real cost | No |
| Your team uses most of the product | Yes | No |
| The vendor's roadmap matters to you | Yes | No |
| It encodes something specific to your business | No | Yes |
| You use exactly one screen of it | No | Yes |
Payments, payroll, email deliverability, identity, accounting — rent all of it. Nobody should be writing their own tax engine to save $400 a month. But the internal tool that tracks your approvals in your stages, the dashboard that reads your own database, the integration between two systems you already pay for: those encode your business, and renting your own business logic is the expensive way to run a company.
Replace one screen and watch
- 01Write down what your team actually does in the tool.
Not what the tool does — what they do. Usually three or four actions. If the list is long and everyone's list is different, keep renting.
- 02Export your data before you decide anything.
Two reasons: you find out now whether a clean export exists, and you learn what the tool has been storing that you assumed it was not.
- 03Build the one screen, run both in parallel.
For two weeks, the team does the work in your version and the vendor's. Disagreements between them are the requirements you missed, and they surface in days rather than after cancellation.
- 04Cancel at the renewal date, not before.
You already paid for the term. Use it as the free parallel-run window, then let it lapse.
- 05Write down who owns the thing you built.
A name, and where the code lives. The failure mode of owning software is not maintenance cost — it is that nobody knows it is theirs until it breaks.
Do this once and something changes in how your team thinks. The next renewal email is not a reflex, it is a question with an answer. Most of the time the answer is still renew — but it is a decision you are making rather than one being made for you.
Pick the tool you would cancel if you were sure. We build the part you actually use, run it beside the subscription, and the code is in your repo by Friday.
See how the free week works →