If you have started pricing custom software, you have already found the guides. They are all written by development firms, they all give a range, and the ranges do not agree with each other.
They are also all answering a narrower question than the one you asked.
Here is the short version, and then the part that the quotes leave out.
The short answer
For a small business — a focused internal tool that replaces one painful manual process, not a platform — published ranges from firms that do this work cluster roughly like this:
- Ortem Tech puts focused small-business systems at $15,000 to $60,000, shipping in 6 to 16 weeks.
- SOLTECH puts custom software generally at $100,000 to upwards of $400,000, with a small-business baseline of $75,000–$100,000.
- Most other published guides land somewhere between those two, depending mostly on how much the firm's own work looks like the low end or the high end.
That spread is not a contradiction. A single-purpose scheduling tool for a nine-person HVAC company and a multi-tenant platform with a mobile app are both "custom software," and one of them costs ten times the other. If you want a useful number, the question to ask a vendor is not "how much does custom software cost" but "how much would this cost, and what did you leave out."
Which brings us to the part that matters more.
What the quote covers, and where it stops
Every number above prices getting to launch day. Design, build, test, deploy, hand over.
We fetched the guide that ranks at the top for this question. It prices builds up to $400,000 across seven cost factors, an FAQ, and an estimating checklist. It does not contain a section on hosting, maintenance, or who runs the software after launch. Not a short one — none.
That is not dishonesty. It is an accurate description of what a project-based firm sells. The engagement ends at handover, so the pricing page ends at handover too.
The problem is that your use of the software does not end at handover. It starts there.
The four costs that begin the day you launch
None of these are optional, and none of them appear on a build quote.
1. Hosting and infrastructure. Servers or containers, a database, file storage, a domain, TLS certificates, email delivery, backups. For a small internal tool this is often genuinely cheap — tens of dollars a month, sometimes less. It is also a thing that now exists in your name, on a card that expires, in an account someone has to have the password to.
2. Maintenance. Not "improvements." Maintenance: security patches, library and framework upgrades, keeping up with the operating system and runtime versions that stop receiving security updates on a schedule you did not choose. Nothing about your business changed, and the software still needs work this quarter.
3. Dependency rot. Modern software is assembled from hundreds of third-party components. They get updated, deprecated, abandoned, and occasionally found to have a serious vulnerability with your name adjacent to it. Software that nobody touches for eighteen months is not stable. It is accumulating a migration that gets more expensive the longer it waits.
4. The person who answers. At 7am, when it does not load, and you have twelve technicians who cannot see their jobs. This is the cost with no invoice and the largest actual impact. If the answer is "email the agency and hope," you have not bought software. You have bought a dependency on somebody's calendar.
More detail on all four, and who actually does each one: who maintains custom software after it's built.
What the industry's own numbers say
You do not have to take our word for the size of this. The firms that build custom software publish their own estimating rules, and they are consistent:
- The long-standing rule of thumb, repeated across vendor estimating guides including CodeStringers, puts annual maintenance at roughly 15–20% of the original development cost, every year. Other guides quote 15–25%.
- Over a product's whole life, estimates of the maintenance share of total cost run higher still. CodeStringers collects several published figures: O'Reilly's "60/60 rule" at 60% of lifecycle expense, Schach at 67%, Galorath at 75%, Pigoski at over 80%.
Two honest caveats, because these numbers get quoted carelessly. First, they are rules of thumb and published estimates, not measurements of your project — they vary enormously by how integrated and how regulated the software is. Second, several of them are decades old. Treat them as an order of magnitude, not a budget line.
But the order of magnitude is the point: the industry that builds custom software agrees that most of what the software will cost you has not happened yet when the build invoice clears.
The five-year number nobody quotes you
Take the honest low end. A focused small-business tool at $30,000 to build.
Apply the industry's own 15–20% annual maintenance rule, add modest hosting, and hold everything else flat:
- Build: $30,000
- Maintenance at 15–20%/yr for four more years: roughly $18,000–$24,000
- Hosting for five years: call it $1,500–$6,000 depending on what it does
Five-year total: somewhere around $50,000–$60,000. The quote said $30,000.
We are showing the arithmetic rather than the answer, because your numbers will differ and you should run them yourself. Change the build figure and everything scales. The shape does not change: the number on the quote is roughly half to two-thirds of what the software costs you over five years, and the missing part arrives as a series of surprises rather than as a plan.
And that assumes everything goes well. The expensive version is the one where nobody does the maintenance for two years, then something breaks, and the firm that built it has moved on, raised rates, or closed — at which point you are paying someone new to learn a codebase before they can fix anything. That is the most expensive hour in software, and you pay for a lot of them.
This is the part worth a conversation before you commit to anyone. Book a call with one of us — you will get a founder, not a sales team, and we will tell you if the honest answer is that you should buy something off the shelf instead.
Three ways to pay for this, and who each one suits
One-off build, then you own the problem. Lowest total cost if you have somebody technical in-house who will actually do the maintenance. Genuinely the right answer for some businesses. Wrong for most businesses under twenty people, because the "somebody technical" is the owner, at night.
Build, then a maintenance retainer. The common upgrade. Better, and it fixes the maintenance gap. Watch two things: what the retainer actually covers versus what gets billed on top, and whether the firm's incentive is to keep your software simple or to keep the hours full.
Build and run it for one monthly price. No large up-front number; you pay monthly for working software that someone else hosts, patches and answers for. The trade is that you are in an ongoing relationship rather than owning a codebase outright — which is a real trade-off with real questions attached, and those questions deserve straight answers before you sign anything. We have written up how that model works, including what happens if you stop paying: custom software on a monthly price.
This is what Truss does, so treat that paragraph as interested rather than neutral. The arithmetic above is not interested — run it against any vendor, including us.
What to ask any vendor before you sign
Copy these. Ask all of them. The answers are more revealing than the quote.
- What is the total cost over five years, not three months?
- Who hosts it after launch, and whose account is the infrastructure in?
- Who applies security patches, and how often? What happens if the framework version we're on goes end-of-life?
- Who do I call at 7am, and what is the response time in writing?
- What is included in the ongoing price, and what gets billed separately? Give me the list, not the adjective.
- Who owns the code, the data, and the right to keep running it? (This one is worth its own conversation — see who owns the code when you pay someone to build your software.)
- What happens if we want to leave? What happens if you close?
Any vendor worth hiring will answer all seven without checking with anyone. If a question gets a vague answer now, it will get an invoice later.