Nobody publishes real numbers for this. Agencies say "it depends" and ask you to book a call. It does depend, but you deserve the ranges before the call, so here they are.
I'm a director at a software development company in Selangor, and I also handle our finances, so I read software quotes from both sides of the table. This is the guide I wish our clients had before they collected quotes.
The short answer
For custom software built by a Malaysian team in 2026, these are realistic ranges:
- A focused internal tool (one workflow, a handful of users, admin screens): RM 40,000 to RM 90,000
- A customer-facing web application or MVP (accounts, payments, dashboard, launched properly): RM 90,000 to RM 250,000
- A mobile app alongside a web platform (iOS and Android via one React Native codebase): add RM 60,000 to RM 150,000
- A regulated or integration-heavy platform (eKYC, insurer or bank integrations, compliance review): RM 250,000 upwards, sometimes well upwards
If a quote sits far below these ranges, the missing money is usually hiding in one of the places covered below.
What actually drives the price
1. The number of screens is a distraction. The number of decisions is the cost.
A five-screen app where every screen shows different data to different roles costs more than a twenty-screen brochure app. When you scope a project, count the decisions the software has to make (who can see what, what happens when a payment fails, what an admin can override), not the screens.
2. Integrations are priced by their worst day
Connecting to a payment gateway with good documentation is quick. Connecting to an insurer or a government system with a PDF spec from 2018 is not. We built a motor insurance API that connects more than ten Malaysian insurers. Each one exposed different interfaces and rules. Integration work is quoted high because the agency is pricing the worst insurer, not the best one.
3. Compliance multiplies everything it touches
If your platform needs eKYC, AML checks, or a regulator's sign-off, every feature it touches costs more to build and test. Our team has shipped two BNM-regulated insurance platforms, and the honest advice is: bring compliance requirements into the first architecture conversation. Retrofitting them later is the most expensive version of the project.
4. Launch is not the end of the bill
Software has a monthly cost after it ships: cloud hosting, monitoring, security updates, fixes. For a typical SME platform on AWS, expect RM 1,500 to RM 8,000 monthly depending on scale, plus a maintenance retainer if you want the building team on call. Any quote that doesn't mention running costs is incomplete. Ask. For what it's worth, we maintain every system we deliver. An agency that won't offer maintenance on its own work is telling you something.
5. Who is building it
A freelancer, a local agency, and an offshore team will quote the same project differently because they carry different risks. Freelancers are cheapest until they disappear mid-project. Offshore teams look cheap per hour until you price the communication overhead in your own time. A local agency costs more per hour and less per outcome, if (and only if) they can show you systems still running years later.
Why five quotes differ by 5x
When owners show me competing quotes, the spread is almost never dishonesty. Each agency priced a different set of assumptions: different user counts, different admin tooling, different support periods, different corners quietly cut.
The fix costs you one page of writing. Before collecting quotes, write down: who uses the system, what each type of user does in it, roughly how many of them exist, which systems it must connect to, and what happens in the first six months after launch. Give the same page to every agency. Now the quotes describe the same project, and the differences tell you about the agencies instead of their assumptions.
Where projects overrun, and how to prevent it
Overruns are not inevitable. We have delivered every project on time, and the habits that make that possible are the same ones that protect your budget. The overruns we see elsewhere come from the same few places:
- Scope discovered late. The workflow everyone forgot until week eight. Prevented by a proper discovery phase before the build quote is fixed.
- The second system. "While we're at it, can it also..." Each addition on its own is fine. The sum is not. Prevented by a change process both sides respect.
- Data migration. Moving years of records out of the old system is a project of its own. Price it as one.
- The missing decision-maker. If the person who can approve screens only appears at launch, you will rebuild screens. Put them in the demo loop from sprint one.
What we'd tell you in the first call
Some projects shouldn't be custom software at all. If an off-the-shelf product covers 80% of your workflow, buy it, and spend a fraction of the budget connecting it to the rest of your tools. We say this in first calls regularly. Custom software earns its cost when the workflow is your competitive advantage, when integrations demand it, or when no product fits a regulated requirement.
If you're scoping something now, our case studies show what real projects delivered, with numbers. And if you already have quotes from other agencies and want an honest second opinion on them, get in touch. Reading proposals is the fastest work we do.