What custom software development costs in South Africa in 2026

Every week someone asks us some version of the same question: what does it cost to build custom software in South Africa? The honest answer is "it depends", and the dishonest answer is a single number. This guide sits between the two. It explains the bands we quote in, what pushes a project from one band to the next, how the engagement model changes the bill, and what a proper proposal should contain so you can compare quotes from different companies on equal terms.

Two caveats before the numbers. Everything here is in rand and reflects what we see in the South African market for professional engineering teams; offshore marketplaces and freelancer platforms price differently and carry different risks. And the bands are ranges for whole projects, not price lists. Your project lands somewhere inside a band because of the drivers described further down, and a written scope is the only thing that turns a band into a price.

The four bands we quote in

We use the same bands on our proposal form that we use internally, because they map to genuinely different kinds of work.

Under R500 000. A focused product with one core workflow: a customer portal, a booking system, an internal tool that replaces a spreadsheet, a marketing site with a few integrations, or a rescue of an existing app that mostly works. One or two engineers for two to three months. The discipline at this level is saying no to scope, not adding it.

R500 000 to R2 million. A complete product for a real business: web and mobile front ends, a proper backend, authentication and roles, payments, reporting, and integrations with the systems you already run. A team of two to four for four to nine months. Most of the platforms in our case studies started in this band.

R2 million to R5 million. A platform with several user types, regulated data, real-time features or high transaction volumes, often with a mobile app and a web application sharing one backend. Four or more engineers over most of a year, with architecture, security and compliance work as first-class line items rather than afterthoughts.

R5 million and up. Multi-year products, platforms serving hundreds of thousands of users, or programmes that replace core systems in stages while the old ones keep running. These are relationships rather than projects, and are almost always run as a retainer with a dedicated team.

What moves a project between bands

The number of screens is the least useful predictor of cost. These are the drivers that actually matter.

• Integrations. Every external system you connect to (payment gateway, accounting package, CRM, ID verification, courier, SMS) adds design work, error handling and testing. Three integrations is normal; ten is a different project.
• User types and permissions. A single kind of user is cheap. Customers, staff, admins, partners and auditors, each seeing different data, is where authorisation logic and testing multiply.
• Regulated data. Personal information under POPIA, financial data under FICA, or health data changes how the system is designed, hosted and audited. It is not a surcharge; it is a different set of requirements.
• Mobile. Native iOS and Android apps roughly double the front-end effort compared with a responsive web app. Flutter or React Native narrows that gap considerably when the app does not need deep platform features.
• Real-time and scale. Live updates, chat, notifications, and systems that must stay fast at thousands of concurrent users need infrastructure and testing that a simple request-and-response app does not.
• The state of what exists. Building on a clean slate is predictable. Building around a legacy system, a half-finished codebase or an AI-generated prototype means an audit first, because nobody can price what they have not read.
• How decisions get made on your side. One empowered product owner who answers questions the same day is worth a great deal of money. Committees and long feedback loops are the most common reason a project drifts a band upward.

How the engagement model changes the bill

Fixed scope. You get a written scope, a fixed price and a timeline. We carry the risk of estimating badly; you carry the discipline of not changing the scope halfway. Best for well-understood projects and for rescues after an audit.

Retainer. A dedicated team for a monthly fee, with a shared backlog you control. Best once a product is live and evolving, or for programmes where the destination is clear but the route will change.

Team augmentation. Our engineers inside your team, your rituals, your tools, billed per person per month. Best when you have engineering leadership and need capacity or a specific skill.

We do not publish an hourly rate card. Hourly billing rewards slow work and hides risk in change orders, which is exactly the "typical outsourcing" pattern we exist to avoid. For retainers and augmentation we agree a monthly rate per engineer that depends on seniority and commitment, and it is written into the proposal before anything starts.

What a proper proposal contains

If a quote you receive is missing any of these, you are comparing a number, not a project.

1. The scope in plain language, including what is explicitly out of scope.
2. The team: who, at what level of seniority, for how many days a week.
3. A timeline with milestones you can see working, not just a delivery date.
4. The price, and the rule for what happens when scope changes.
5. What you own at the end: repositories, cloud accounts, credentials, documentation.
6. What happens after launch: support, monitoring, and how bugs versus new features are handled.

Where the money goes

On a typical project in the middle band, roughly a fifth of the effort is discovery, design and architecture, more than half is building and integrating, and the remainder is testing, security, deployment and handover. Teams that quote noticeably below the market usually do it by skipping the first and last parts. You find out when the system meets real users, real data and a real auditor.

Where QLTech fits

We quote fixed scope or transparent retainers, we write the proposal in the form above, and we start every engagement on an existing system with an audit so the price is based on what is actually there. If you want a number for your project, the quickest route is the proposal form: four questions, and a written scope back within three business days. If your starting point is an app that was built with AI tools and has stalled, start with the AI app rescue page instead; the audit is a fixed price.