Custom software or SaaS? A plain answer for a 5 to 50 person *business*

Is it cheaper to build or buy software for a small business?

For a few users, or a full replacement at 5 to 15 employees, off-the-shelf SaaS is usually cheaper. A targeted custom tool that saves 10 or more hours a week can pay back within a year, and a client portal becomes cheaper to own than rent past roughly 120 users (vendor analyses, read 2026-10-16).

We build software for a living, so read this with that in mind. The honest answer starts with where renting wins.

When does SaaS win?

If your process is ordinary, a product built for thousands of customers beats anything we build for one. Accounting, payroll, email and a standard CRM are solved problems. A software consultancy models a ten-person business running a stack of SaaS tools at $1,620 to $1,820 a month. Over five years it finds $97,200 to $109,200 for the SaaS stack against $140,000 to $270,000 to replace it with custom software. Its verdict: SaaS wins at this scale, and it is not close (PCG, read 2026-10-16).

Small teams also get things a custom tool has to earn: new vendor features every quarter, someone else's on-call engineer, and a product you can drop when it stops fitting. Scheduling tools start near $16 a month. At that price a build is the wrong answer.

When does a targeted custom tool pay back?

The same guide makes an exception for a highly specific workflow that no product handles well. A targeted tool of $15,000 to $30,000, not a platform replacement, can pay for itself within a year if it saves 10 or more hours a week of manual work. For a larger custom platform at 50 to 200 employees it puts break-even at 18 to 30 months after launch (PCG).

Benchford's Focused build ($4,950) and Complete build ($9,850) sit below that range on purpose. They fix one workflow, or one customer-facing surface and the staff view behind it, instead of replacing a stack.

The arithmetic is short. Spread over 24 months, a Complete build with Standard care is about $709 a month, before tax. If the rented tools you would drop add up to less, you need recovered hours or customers to make it pay. About 36 hours a month of front-desk time at $20 an hour does it, or seven more memberships at $110 a month. Those are break-evens, not forecasts.

What does renting versus owning look like for a client portal?

Portals show the crossover clearly. A portal guide models 40 client-side users on portal software at $25 a user per month: $39,000 to $41,000 over three years, against $85,000 to $95,000 for a custom portal with approvals and several roles. At 40 users, renting is cheaper, and it is not close. At 80 users the software bill reaches $24,000 to $28,000 a year and the gap closes. Past about 120 users, custom is cheaper (0arch.io, read 2026-10-16).

The build in that example is $65,000, a different scale from ours. The shape of the curve is the point: per-seat fees grow with your clients, and a built system does not.

What is a build-versus-buy analysis?

It is a comparison over one time span. Add up what you pay to rent the tools for the years you plan to use them. Add the hours lost to workarounds. Then set that total against the build price plus hosting and upkeep for the same years. The side with the lower total wins, unless one side cannot do the job at any price.

When should a small business build custom software?

• • You have tried three products and each one needed a workaround.
• • The process is what makes you different from your competitors.
• • Staff or customers use it every day, and per-seat fees grow with them.
• • It has to connect to systems you already run, and no product does that well.

What about lock-in, data export and API access?

Cost is not the whole price. Read three things before you sign anything: how the contract renews, what you can export, and whether there is an API. Toast, for one, states that a point-of-sale contract renews automatically for one-year periods and that you must give at least 30 days' written notice not to renew (Toast support, read 2026-10-16). Jane, for another, states that it has no open API and no plans to add one (Jane, read 2026-10-16). Neither is a flaw. Each limits what you can build on top.

With custom software the questions flip. You own the code and the data, in accounts in your name, and a written runbook lets another developer take over. That is ownership, not magic. It also means you pay for hosting and upkeep.

How do you test it with your own subscription bill in 15 minutes?

1. 1. List every subscription tied to the process, with the price per user and any add-ons.
2. 2. Multiply by your users for 36 months, and add 5% a year for price rises.
3. 3. Add the hours your team spends on workarounds each week, at what an hour of their time costs.
4. 4. Compare that total with a fixed price plus care for 36 months.
5. 5. If the rented tools win, keep them. If the gap is hours and not dollars, a targeted tool may pay back.

Fifteen minutes with a spreadsheet beats an hour of anyone's sales pitch, ours included.

What do we tell people who should not build?

Most small businesses should not build their core system. If a configured tool or a spreadsheet covers the job, we say so on the call and point to the tool. A freelancer at $2,000 may be right for a simple brochure site. We add a fixed price, a written ship date, a launch guarantee, a 12-month warranty and a runbook, and that is worth paying for when the system matters to the business. See every price and what we build, and read what custom software costs for the ranges.

Sources, all read on 2026-10-16: PCG, 0arch.io, Toast support and Jane. Prices change. Check each source on the day you decide. Last updated October 16, 2026.