Ask "how much does custom software cost?" and you usually get one of two answers:
- "It depends." (True, and useless.)
- A number so big you close the tab.
So let's do this properly. Plain numbers, plain words, no sales fog.
The short answer
Industry guides put custom business software in roughly these ranges:
| What you build | Typical range | Example |
|---|---|---|
| One focused module | $20,000 to $50,000 and up | An approval workflow, a quoting tool, an inspection app |
| A multi-department system | $75,000 to $300,000+ | Orders, inventory and customer portal working together |
| Yearly upkeep | 15 to 20% of the build cost | Hosting, fixes, small improvements |
Two honest warnings. First, most of these numbers come from firms that sell custom builds, so treat them as a rough guide, not a quote. (SPEC INDIA, Merehead) Second, AI is pushing the low end of every range down fast. More on that below.
What actually decides the price?
Not the number of features. These seven things do:
- How many screens and steps. A five-screen approval app is a weekend chore compared with a fifty-screen order system.
- How many systems it talks to. Every connection to your ERP, CRM or finance system adds work. This is usually the biggest cost driver.
- How many types of users. Planners, managers, customers and auditors each need different screens and permissions.
- Moving old data. Ten years of messy spreadsheets don't clean themselves.
- Rules you must prove you followed. Audit trails, approvals and regulator reports add real effort.
- Phones and offline use. Field teams in basements and substations need apps that work without a signal.
- How fast you make decisions. Projects get expensive when nobody can say "yes, that's right" for three weeks.
What does renting cost instead?
This is the part most "build vs buy" articles skip. Let's do the math on a typical per-seat SaaS plan.
| Example | |
|---|---|
| Users | 300 |
| Price per user, per month | $165 (Salesforce Enterprise list price after its 2025 increase) |
| Cost per month | $49,500 |
| Cost per year | $594,000 |
| Cost over five years | $2,970,000 |
And that's before implementation partners, add-ons, extra storage, AI features and future price increases. Those increases are real:
- Salesforce raised list prices by about 6% in August 2025. (Tech.co)
- Microsoft 365 E3 went from $36 to $39 per user per month in July 2026. (UC Today)
After five years of renting, you own nothing. If you stop paying, it's gone.
Where do custom projects blow the budget?
Custom software gets expensive for predictable reasons:
- Fuzzy requirements. "We'll figure it out as we go" is the most expensive sentence in software.
- The big bang. Trying to replace everything at once instead of one module at a time.
- Hidden extras. Training, change management and post-launch support can add 15 to 25% if nobody plans for them. (SPEC INDIA)
- No documentation. The team that built it leaves, and nobody knows how it works.
Notice that none of these are about code. They're about planning.
When does building win?
Build when most of these are true:
- The process is part of what makes your business different
- You'd be paying per seat for hundreds of users
- The product forces you to change how you work to fit it
- You're paying for a huge product but using a small part of it
- It needs to talk to your ERP and other systems in your own way
When should you keep buying?
Building everything is just as silly as buying everything. Keep renting the boring, standard stuff:
- Email, chat and office tools
- Payroll and basic accounting
- Anything where your way of working is the same as everyone else's
Rent the commodity. Own what makes you different.
How AI changed the math
Until recently, the expensive part of custom software was people: analysts, architects, developers and testers, working for months. AI now does a large share of that work. The biggest software companies say AI already writes 20 to 30% or more of their code. (We covered this in AI just rewrote the rules of enterprise software.)
But AI alone doesn't fix the budget-killers above. Fuzzy requirements and missing documentation are planning problems, and AI that skips planning just makes the wrong thing faster.
That's how Defender works: AI agents do the heavy lifting, from writing the requirements and the architecture to building, testing and deploying the module. Your team approves each step. You get the speed of AI with the planning that keeps projects on budget, and you own all the code at the end.
How to get a real number for your module
Stop asking for quotes on a vague idea. Get the requirements written first. Then the number stops being a guess.
Get a real cost for your first module
Register for Defender and we'll write the full requirements for the module you want, for free. You'll see exactly what it takes to build, before you spend a dollar.
Register for Defender
