Custom software or off the shelf? A five-question test for small businesses

Every few weeks an owner tells us they need custom software. Sometimes they do. More often they need two existing tools connected, or a spreadsheet retired, or a process written down. The difference is worth thousands of dollars, so here is the test we run in our own heads before recommending anything.

The five-question test

Answer these honestly about the one workflow you are trying to fix. Not the business, the workflow.

1. Does the process work the same way every time? If three people do it three ways, stop. Custom software built on an unsettled process automates the argument. Off-the-shelf software fails for the same reason. Fix the process first. We wrote up how to tell a process problem from a software problem if you want the longer version.

2. Is this how you make money, or is it overhead? Payroll, bookkeeping, email, and scheduling are overhead for nearly everyone, and the off-the-shelf market for them is deep. Buy. The thing that makes your business different from the one down the street is where custom work might earn its keep.

3. Have you tried the tools you already pay for? Most small businesses own three or four subscriptions they use at ten percent. Before buying anything, check whether QuickBooks, your CRM, or your scheduling tool already does the thing. It is surprising how often it does.

4. Is the gap a missing connection or a missing feature? “Data gets re-typed from A into B” is a connection problem. It is cheap to fix with a no-code tool or a small connector. “Nothing on the market does what we do” is a feature problem, and it is the only answer that points toward real custom software.

5. Who will own it in a year? Off-the-shelf tools get maintained by the vendor. No-code automations need somebody on your team who will notice when they break. Custom software needs a developer you can reach. If you cannot name the person, you are not ready for that option.

Score it roughly. If questions 1 and 5 are shaky, fix the process and assign an owner before spending a dollar. If question 4 says “connection,” you want no-code or a small connector. Only when all five point the same way does a real custom build make sense.

Walk-through one: the landscaping company

Say you run a 20-person landscaping company. Crews get their routes from a scheduling app, the office bills from QuickBooks, and customers text the owner’s cell when something goes wrong. The owner wants “a custom app that does all of it.”

Run the test. Does the process work the same way every time? Mostly yes for scheduling and billing. No for customer texts, which get handled by whoever sees them first. Is it how you make money? Scheduling and billing are overhead. Customer communication is closer to the core. Tools you already pay for? The scheduling app has a customer portal nobody turned on. Connection or feature? The scheduling app does not tell QuickBooks when a job is done, so somebody re-types it. That is a connection. Owner? The office manager, who is comfortable with spreadsheets but does not want to learn a developer tool.

The honest answer is not a custom app. It is turning on the portal, connecting the scheduler to QuickBooks with a no-code tool the office manager can see and maintain, and writing a one-page rule for who handles texts. Total spend is a few hundred dollars a year and a few hours of setup. The “custom app” would have cost fifty times that and automated the text chaos.

Walk-through two: the electrical contractor

Say you run a 35-person electrical contractor doing commercial service work. Technicians fill out a paper checklist on every job that has to match a specific inspection standard. The office scans them, a coordinator types the results into a spreadsheet, and the spreadsheet generates a compliance report the customer’s insurer requires.

Run the test. Same way every time? Yes, because the standard forces it. How you make money? Yes. The compliance report is a large part of why customers pick this contractor. Tools you already pay for? The field service app has a generic forms module, but it cannot produce the report format the insurer wants. Connection or feature? Feature. The report logic is specific, and no vendor builds for it. Owner? The operations lead, who already maintains the spreadsheet and would rather maintain a small tool.

This one earns a small custom build: a mobile form that replaces the paper, and a report generator that produces the insurer’s format. It is not a platform. It is one form and one document, built on a process that was already stable. That is what custom software and automations should look like for a business this size.

Zapier, Make, or a small custom build

When the answer is “connect things,” you have three realistic options. Here is how they compare. Costs are market ranges as of September 2026 and will drift.

Option	Fits when	Rough monthly cost (market range)	Watch out for
Zapier	Simple two- or three-step workflows, non-technical owner	About $20 to $100 a month for most small teams	Per-task pricing climbs fast on high-volume workflows
Make	Someone on the team likes logic; more steps per dollar	About $10 to $60 a month for most small teams	Steeper learning curve; easier to build something nobody else understands
Small custom connector	Exceptions, retries, audit trail, or a tool with no integration	$20 to $100 a month hosting, plus a one-time build in the low thousands	Needs a developer you can reach when something changes

Our rule of thumb: start with Zapier or Make for anything that fits in ten steps and fails gracefully. Move to a small custom build when the workflow needs to handle what happens when step six fails at 2 a.m., or when one of the tools simply has no integration. We covered what automating a process actually costs in more detail, including what drives a build from the low end to the high end of the range.

One more thing about no-code tools. They are fantastic until they are load-bearing. What we have watched teams do is build thirty automations over two years, lose the person who built them, and then be afraid to touch any of it. If a no-code workflow touches money or customers, document it the day you build it.

When to stop and fix the process instead

Sometimes the test stops you at question one, and that is the most valuable result. A few signs you should put the wallet away:

• Two people describe the workflow differently and both are confident
• The “exceptions” are more than a third of the volume
• The workflow exists because of a problem that was solved years ago
• Nobody can say what “done” looks like for a single instance of the work

In each case the cheapest fix is a meeting, a written one-page process, and a month of doing it that way. Then run the test again. You will usually find the software question has shrunk, and sometimes it has disappeared.

This is why our Opportunity Assessment starts by watching the work rather than demoing a tool. The memo ranks what to delete, what to connect, and what, if anything, to build. Most of the time the build line is short.

Common questions

Is custom software ever the right call for a business under 100 people?

Yes, but rarely as the first move. It fits when the process is stable, it is specific to how you make money, and nothing off the shelf covers it without contortions. Even then, the right custom build is usually small: a connector or a single screen, not a platform.

Zapier or Make?

Zapier if the people maintaining it are not technical and the workflows are simple. Make if someone on the team is comfortable with logic and data shapes and you want more steps per dollar. Both are fine for two-tool connections. Neither is great once a workflow needs to handle exceptions, retries, and audit trails.

How much does a small custom build cost?

As a market range as of September 2026, a small custom connector or internal tool from an independent developer or small practice typically runs from about $3,000 to $15,000 to build, plus $20 to $100 a month in hosting. The wide range comes down to how many systems are involved and how clean the data is.

What if we already bought the wrong thing?

It happens to most businesses at least once. Before replacing it, check whether the tool is wrong or the process feeding it is. Often the tool is fine and the setup was done for a version of the job nobody actually does. Reconfiguring is cheaper than switching.