Asking what custom software costs is a bit like asking what a building costs. A shed and an office tower are both buildings. Still, owners deserve better than "it depends," so here are the ranges we would give a friend, the reasons a project lands at one end or the other, and how to set a budget you won't regret.
One caveat up front. These are ranges from our own experience working with small and mid-size businesses in the US. They are not survey data, and a large agency or an offshore team will quote differently.
Typical ranges in 2026
- A single automation or integration (two or three tools connected, a scheduled report, an approval workflow): roughly $5,000 to $25,000.
- An internal tool or dashboard used by your own team, such as a quoting tool, a job tracker or a finance dashboard: roughly $15,000 to $60,000.
- A system that runs a core process, with several user types, customer-facing screens or a replacement for a product you outgrew: roughly $50,000 to $150,000, sometimes more.
- A marketing website built properly, with real copy and search basics: roughly $5,000 to $20,000.
If someone quotes far below these numbers, ask what is left out. If someone quotes far above them for a small business tool, ask what you are paying for.
What drives the cost
The number of screens matters less than people expect. These matter more:
How many outside systems it touches
Every connection to another product (your accounting system, CRM, payment processor, a supplier's portal) adds work. Some have clean, well-documented APIs. Others are old, slow or barely documented, and those can double the effort for that piece.
How settled the process is
If your team agrees on how the work should flow, building it is straightforward. If three people do it three different ways, part of the project is getting to one answer. That is valuable, but it is time, and time is what you pay for.
Data cleanup
Moving years of records out of spreadsheets or an old system is often the most underestimated part of a project. Duplicates, inconsistent names and missing fields all have to be dealt with before the new system can trust the data.
Who uses it
An internal tool for eight people can be plain and practical. Anything your customers see needs more design, more testing on phones, and more care with permissions and security. Expect customer-facing work to cost noticeably more for the same features.
Why it got cheaper
A few years ago, a modest internal system often meant a small team for several months. Today, in our experience, an experienced developer using AI coding tools can often deliver the same thing in 6 to 12 weeks. That has moved custom software into budgets where it used to be out of reach.
The savings are uneven, and it helps to know where they come from. AI tools are very good at the routine parts: standard screens, forms, data plumbing, first drafts of tests and documentation. They are much less help with the parts that decide whether a project succeeds. Understanding how your business makes money, deciding what to leave out, cleaning messy data and checking that the numbers tie out are still human work.
This is also where the risk sits. AI-written code that nobody experienced reviews can look finished and still be wrong in ways that surface months later. When you talk to a builder, ask how generated code gets checked before it ships. A vague answer is a warning sign.
For small and mid-size businesses, the practical effect is simple. Problems that used to be "live with it" because a fix cost six figures are now worth solving. A process that eats 15 hours a week across your team was never worth a $150,000 project. At $20,000 to $40,000 it often is.
Costs after launch
The build is not the whole bill. Plan for these:
- Hosting and services. Servers, databases, email sending, file storage. For most small business tools this runs from about $20 to a few hundred dollars a month.
- Maintenance. Security updates, fixes when a connected product changes its API, and small changes as your business shifts. Our rule of thumb is to budget 15% to 20% of the build cost per year.
- Your team's time. Someone on your side should own the tool, report problems and decide on changes. A couple of hours a month is usually enough.
Software with no maintenance plan doesn't stay the same. It slowly breaks as everything around it changes.
One example: how we price
Here is how our own pricing works, as one reference point among many. We bill by the hours of development. A short discovery maps the process and your data and tools, and ends with an estimate of the hours the build will take. Care plans, which cover hosting, monitoring, updates and changes, are billed monthly. A website typically takes a few weeks, and a custom system roughly 6 to 12 weeks. The details are on the pricing page, and you can see the kinds of tools those numbers buy on the examples page.
Other good builders price differently. The structure matters more than the numbers: a short step to define the work, then a clear estimate before the build starts.
How to budget
This is the part where our finance background shows. Start with the cost of the problem, not the cost of the software.
- Estimate the hours the current process takes each week, across everyone involved. Multiply by a loaded hourly cost.
- Add the cost of errors: late fees, missed invoices, refunds, a bad decision made on stale numbers. Be conservative.
- Decide what payback period you would accept. For most owners we talk to, 12 to 24 months feels right.
- That gives you a ceiling for the build plus two years of upkeep. If the quotes come in well under it, the project is probably worth doing.
- Hold back 10% to 20% as contingency. Something always turns up in the data.
If the math doesn't work, don't build. Fix the process on paper, or use an off-the-shelf tool, and revisit it in a year.
Common questions
Should I pay hourly or a fixed price for custom software?
Either can work. A fixed price for a vague idea is usually padded heavily to cover risk, or followed by change orders. Hourly billing with a clear estimate up front means you pay for the work your project actually needs.
Does AI-assisted development mean lower quality?
It can, if nobody experienced reviews the output. Used well, it frees developer time for the parts that matter: understanding the business, testing, and checking the edge cases. Ask any builder how AI-written code gets reviewed before it ships.
Do I own the code when the project is done?
You should, and it should be written into the contract. Our clients own the software we build for them, and we host and manage it. If a builder keeps ownership or licenses it back to you, price that in, because it ties you to them for as long as you use it.