Custom software vs off-the-shelf: an honest comparison
Most companies should buy most of their software — and a firm that sells custom development should say so plainly. Off-the-shelf wins the commodity categories, and it isn't close. The question worth your time is the narrow slice where it doesn't: the one or two processes that are the reason customers pick you.
When off-the-shelf wins
The default answer, and rightly so. If any of these describe the function, buy it and move on.
The function is a commodity
Accounting, payroll, email, support desks, standard CRM. Thousands of companies need exactly the same thing, so vendors amortise serious engineering across all of them. You will not out-build that — and shouldn't try.
The process doesn't differentiate you
If a competitor ran your identical workflow tomorrow and lost nothing by it, the workflow isn't your edge. Buy the tool, accept its defaults and spend your attention where it compounds.
The team is small and the fit is close
A ten-person team bending slightly to a tool's shape costs far less than any build. Custom software earns its keep on volume — of transactions, of seats, of exceptions.
When custom wins
Four signals that the subscription has stopped being the cheap option — usually arriving quietly, one budget cycle at a time.
The process is the advantage
When how you quote, schedule, route or price is the reason customers pick you, renting the same workflow as every competitor puts a ceiling on it. That is the core case for custom software: the system encodes the way you win.
A spreadsheet layer grew around the SaaS
Exports, macros, the one workbook everyone fears touching. The tool stopped fitting long ago — the workaround layer is your real system now, and it's unowned, untested and one resignation away from collapse.
Per-seat pricing at scale
At ten seats the subscription is a rounding error. At two hundred it's a build budget every few years — recurring forever, with an annual price rise. The cost guide shows where the lines cross.
Integration spaghetti
Five tools, none of them the source of truth, and people re-keying data between them daily. At some point the connective tissue matters more than any single tool — and it's the one part nobody sells off a shelf.
The comparison, dimension by dimension
Neither column wins every row — which is exactly why "it depends" is the honest headline and the rows below are what it depends on.
| DIMENSION | OFF-THE-SHELF | CUSTOM |
|---|---|---|
| Upfront cost | Low — a subscription and some setup | Higher — you fund the build before the payoff |
| 5-year cost shape | Climbs with seats, tiers and add-ons — indefinitely | Front-loaded, then flattens to hosting and maintenance |
| Fit to process | You adapt to the tool's model of the work | The tool adapts to how you actually work |
| Switching cost | Grows quietly — data, habits and integrations accumulate | You own the code and the data, so the leverage stays with you |
| Ownership | Rented — the vendor's roadmap decides what changes | Yours in full — IP, data model and roadmap |
| Speed to value | Days to live, if the fit is close | First milestone in weeks; a focused v1 in 2–3 months |
Buy the commodity. Build the edge. Integrate both.
For most companies the honest answer isn't a side — it's a split. The build-vs-buy question is rarely all-or-nothing, and treating it that way is how businesses end up either renting their advantage or hand-building their accounting.
Buy the commodity
Accounting, payroll, email, support desk — categories where a vendor's thousands of customers subsidise engineering you would otherwise fund alone.
Build the edge
The one or two systems where your process is the product — the quoting engine, the scheduling logic, the pricing model. That is where custom software pays for itself.
Integrate both
Connect the bought and the built so data flows without re-keying. Integration engineering is where hybrid setups quietly succeed or fail.
A five-question checklist
Answer yes or no for the specific process you have in mind — not for your software estate as a whole.
- 1
Is this process part of why customers choose you?
If a competitor could run the identical workflow and lose nothing, buy. If the workflow is the moat, keep reading.
- 2
Has a spreadsheet layer grown around your current tool?
Exports, macros and manual workarounds are the strongest single signal that the fit has run out.
- 3
Do licence costs scale with headcount in a way finance has started to notice?
Per-seat pricing is fine at ten seats and a build budget at two hundred.
- 4
Is data re-keyed between two or more systems every week?
Re-keying is a tax with an error rate — and it compounds as tools accumulate.
- 5
Would you run the process differently if the software allowed it?
A yes here means the tool is shaping the operation rather than serving it.
Three or more yeses and custom deserves a priced evaluation — an evaluation, not a commitment. See what drives the price in the cost guide, or bring the process to a free diagnosis and get the answer with numbers attached.
Common questions
Isn't custom software just more expensive?
Upfront, usually — you fund a build instead of a subscription. Over five years the picture often inverts: per-seat licences scale with headcount, add-on tiers accumulate and the manual work around a misfitting tool carries a payroll cost. The cost guide shows how to put real numbers on both sides.
How long before custom software delivers value?
Longer than a subscription, shorter than reputation suggests. A focused v1 ships in 2–3 months, with the first working milestone in production well before that. Off-the-shelf is live in days — but "live" and "fitting your process" are different milestones.
Can we start off-the-shelf and go custom later?
Often the right sequence. Buy the tool, learn what the process actually needs and watch for the signal that the fit has run out — usually a spreadsheet layer growing around the tool. By then you're specifying the replacement from evidence rather than guesswork.
Won't you just tell us to build? You sell software.
Roughly a third of our free diagnoses end in a written "don't build" — an off-the-shelf tool, a process change or a smaller automation instead. Recommending builds that shouldn't happen is how estimates stop holding, and ours have to hold.
The generic answer ends here.
A free diagnosis answers build vs buy for your specific process — and a third of the time, the answer is don't build.