What does custom software actually cost?

Almost nobody publishes this, which is why you’re reading a fifth search result to find it. Agencies want you on a call before they say a number, and freelance marketplaces show rates without scope, which is worse than useless.

So here are real ranges, what moves them, and how to tell whether a quote you’ve been given is realistic. These are market figures for small-business custom development in the US, not a price list — ours are on the services page.

The ranges

A marketing site with a CMS: $2,500 – $15,000

The low end is a handful of pages on an existing brand, with a content editor your team can use. The high end adds custom design, complex content structures, integrations, and a migration from an old site with redirects handled properly.

Below about $2,000 you are buying a template someone configured, which is a legitimate purchase — just know that’s what it is.

An automation or integration: $3,000 – $12,000

Connecting systems that don’t talk, or automating a repetitive manual process. Cost is driven almost entirely by how many systems are involved and how well documented their interfaces are. A well-documented modern API is cheap to work with; a legacy system with no API and a CSV export is not.

An internal tool or dashboard: $6,000 – $30,000

Replacing spreadsheets with one source of truth. The range is wide because “a dashboard” covers everything from one read-only view to a full operational system with roles, editing, and audit history. Data migration from messy existing files is routinely underestimated by both sides.

A customer-facing web application: $15,000 – $75,000+

Anything your customers log into. The jump from internal to customer-facing is real: authentication, permissions, payments, and error states all have to work for people you can’t train and can’t call.

Ongoing development: $2,000 – $10,000+ / month

A retainer for continuous work. Priced on capacity rather than deliverables, which suits a roadmap that changes month to month.

What actually moves the number

• How well you can describe the current process. The single biggest factor, and the one clients don’t expect. If you can walk someone through it step by step, scoping is quick and accurate. If you can’t, the first phase of the project is discovering it — and that gets billed.
• Integrations. Each system you connect to adds cost, and legacy systems without proper interfaces add a lot.
• Number of user types. One kind of user is straightforward. Admin, staff, and customer with different permissions is roughly three times the surface area to build and test.
• Data migration. Moving history out of spreadsheets is nearly always underestimated, because the mess only becomes visible once someone tries to import it.
• Decision speed on your side. Genuinely a cost driver. A project that waits a week for each approval costs more than the same project with a responsive contact, because idle time still has to be scheduled around.

What should make you suspicious

• A fixed price with no discovery. Either they’ve padded it heavily, or they’ll discover the real scope later and come back for more. Neither is good for you.
• A quote far below the ranges above. Usually means a different definition of “done” — no tests, no documentation, no handoff, and nobody available when it breaks.
• No written exclusion list. If nobody has told you whatisn’t included, you will find out by invoice.
• Hourly billing with no cap on an undefined scope. All the risk sits with you.

When you shouldn’t build at all

Plenty of the time, custom software is the wrong answer, and a developer who won’t tell you that is not one to hire. Buy off-the-shelf if an existing product does 80% of what you need, if your process isn’t unusual, or if you can’t yet describe what the software should do. More on that in buy or build.