A prospect once showed me three quotes for the same system: R450 000, R1.2 million, and R2.8 million. All three vendors had seen the same brief. None of them were lying. They were pricing three different depths of the same iceberg, and the brief was vague enough to allow it.
After nine years of quoting, building, and maintaining custom software, here is the honest breakdown of where the money actually goes, so you can compare quotes on substance instead of size.
What drives the build price
The build price is mostly a function of four things: the number of screens and workflows, the number of integrations, the number of user roles, and the messiness of the data being migrated in. A quote that does not ask hard questions about all four is a guess, and guesses get revised in your direction.
| Project shape | Typical range | Timeline |
|---|---|---|
| Focused internal tool, one workflow, no integrations | R350k - R700k | 8-12 weeks |
| Customer-facing MVP, payments, one integration | R700k - R1.5m | 12-16 weeks |
| Operations platform, multiple roles, 3+ integrations | R1.5m - R4m | 4-9 months |
| Enterprise system replacement with data migration | R4m+ | 9-18 months |
These are 2024 South African agency ranges. Offshore rates can halve them, and rework can double them back. The cheapest quote is only cheap if it survives contact with production.
The costs that are not in the quote
- Running costs. Hosting, monitoring, backups, and licences typically land between R5 000 and R40 000 per month depending on scale and compliance needs.
- Maintenance. Budget 15-20% of the build cost per year. Dependencies age, browsers change, and the integration you rely on will change its API on a Friday.
- Your own people’s time. A build needs 4-8 hours a week from a decision-maker on your side. Projects stall more often from absent product owners than from slow developers.
- Change. The product you launch is version one. The features users actually request after launch are where the real value gets built, and they need a budget line.
Why the cheap quote gets expensive
The R450 000 quote from my opening story was real, and the client took it. Eighteen months later they came to us with a system that could not pass a security review, had no tests, and needed a partial rebuild before it could take the load of their actual customer base. The final bill crossed R2 million, and the expensive part was not the money. It was the year of market time they lost.
Low quotes usually save money in the places you cannot see at demo time: error handling, security, tests, documentation, and data integrity. The demo looks identical. Month six does not.
How to keep the cost honest
- 1Pay for a short discovery phase before committing to a build price. A fixed quote on a vague brief protects nobody. We covered our version of this in how we run discovery workshops.
- 2Ask every vendor the same three questions: what is excluded, what happens after launch, and what does a change cost. Compare the answers, not the totals.
- 3Insist on shippable increments every two to three weeks. Progress you can click is the only progress that counts.
- 4Ring-fence 15% of the budget for the things discovery cannot see. If you do not plan the contingency, the project plans it for you.
When custom is worth it
2-3x
build cost, the realistic five-year total
15-20%
of build cost per year for maintenance
15%
contingency worth ring-fencing on day one
Custom software is expensive. It is worth it when the process it automates is your competitive edge, when off-the-shelf tools force your team into workarounds that cost real hours, or when licence fees for a big suite exceed the cost of owning exactly what you need. If you are weighing that decision, talk to us before you collect quotes. Thirty minutes of scoping will make every quote you receive comparable.