Build an Internal Tool or Buy a SaaS Subscription?
By CodexierPublished 6 min read
Every growing company reaches the moment where the tool it pays for per user per month no longer fits how it works, and someone asks whether building its own would be cheaper. Sometimes it is. More often the honest answer is buy, and configure it properly. This guide gives a method for comparing the two over five years, the signals that custom is justified, and a scorecard you can fill in before calling anyone.
What off-the-shelf does well
A subscription gives you in an afternoon what took the vendor years: a tested product, security work, integrations with Fortnox or HubSpot, support and a roadmap someone else pays for. For processes every company has, such as bookkeeping, payroll, CRM or document signing, that is unbeatable. A custom build of a standard process is paying full price for a worse version of something you could rent.
Where workarounds become costly
The signal that a subscription no longer fits is not dislike but labour. Count the spreadsheets that exist because the tool cannot do something, the double entry between two systems, the manual exports on Friday afternoon, and the seats paid for people who log in once a month to click one button.
- Your process is a differentiator: how you quote, schedule or price is why customers choose you, and the tool forces you into its version.
- Seat pricing punishes you: forty field staff need one screen each, and the vendor charges the full seat.
- You run three tools plus glue: the integration between them is now the real system, and nobody owns it.
Five-year cost comparison
Put both options in one table with the same rows, and fill it with your own numbers. The shape of the result is more reliable than any single figure.
| Cost line | Buy (subscription) | Build (custom) |
|---|---|---|
| Year one | Seats × price × 12, plus setup and training | Build price, plus hosting and a small change budget |
| Years two to five | Seats grow with headcount; vendor price rises | Hosting, maintenance and change requests; no per-seat cost |
| Risk | Price change, feature removal, vendor exit | Developer leaves, code rots without maintenance |
| Exit | Data export; re-implementation elsewhere | You keep the system; you fund its upkeep |
As a scale check: a tool at 490 kr per user and month for ten people costs, over five years, close to six times the fixed price of our MVP build at 49 990 kr. The build still needs hosting and upkeep, but the seat count no longer matters.
Two honest corrections to that check. A custom tool needs a maintenance budget every year, or it becomes the old website nobody dares touch. And a subscription's setup and training costs are real but small next to five years of seats, so do not let year-one numbers decide.
Ownership, data and exit
Subscription: own the data
Check the export format, whether attachments and history come with it, and the notice period. A vendor without a full export is a lock-in, not a tool.
Build: own the code and hosting
The contract assigns the source code to you, the repository and hosting accounts are in your name, and a second developer could take over from the documentation.
Both: own the process
Write your process down independently of any tool. That document is what makes switching, or rebuilding, possible later.
A build-or-buy scorecard
Score each statement from zero (false) to two (clearly true). Ten or more points to build, under six to buy, in between run a small paid discovery before deciding.
- The process is specific to us and part of why customers choose us.
- We pay for many seats that use a small part of the product.
- The manual workarounds cost more than a day a week across the company.
- The requirements have been stable for a year; we are not still discovering what we need.
- We can fund yearly maintenance without resentment.
- No off-the-shelf tool covers the core flow without heavy customisation.
- We have someone who will own the tool internally, specifying and prioritising changes.
A high score does not mean building everything. The usual right answer is a small custom core for the differentiating flow, with bought tools around it for bookkeeping, mail and documents. A planning blueprint scopes that core before any code; the MVP build delivers it at a fixed price listed on the pricing page.
When you should not build with us
If your requirements changed twice last quarter, you are still discovering the process; keep buying, or use a no-code tool, until it settles. If the scorecard lands under six, we will say so on the first call rather than sell a build. And if there is no internal owner for the tool, a custom system will decay just like the subscription workarounds did, only with you paying for it. For a first version of a product you intend to sell rather than use, see what an MVP is and is not.
Frequently asked questions
Is no-code a middle option?
Yes, for internal tools with modest data volumes and a clear owner. Platforms like Airtable, Retool or Power Apps let you build a fitting tool quickly, at a subscription cost and with their own lock-in. They suit the in-between scorecard result; a custom build suits a process that is core and stable.
How do we protect ourselves if the developer disappears?
Source code assigned to you in the contract, repository and hosting in your own accounts, documentation good enough for a second developer, and a maintenance agreement or at least a known contact for one. Ask any builder, including us, to show how a handover would work before you sign.
Can we start by building only part of it?
That is the recommended route. Build the one flow that the subscription handles worst, integrate it with the tools you keep, and run both for a quarter. If the custom part proves itself, extend it; if not, you have lost a small fixed price rather than a year.
Want an honest read on your scorecard?
Bring your current tools, seat count and the workarounds list to a short call. We tell you whether to reconfigure, switch, build a small core or leave it alone, and what a fixed-price build would cost if it makes sense.
Book a free 15-minute call