A third of teams replaced a software subscription with their own build. Should you?

In February, Retool published its 2026 Build vs. Buy report. It surveyed 817 people in late 2025, from engineers to operations and finance staff, at companies from startups to the Fortune 500. The number that travelled furthest was this one: 35 percent of teams had swapped a SaaS subscription for something they built themselves, and 78 percent expected to build more internal tools this year. The build vs buy question looks different now, but the number needs care.

Read that number with care. Retool sells a platform for building internal apps, and the people it asked were its own customers and builders. They are far more likely to build than the average company. “At least one tool” could also mean a small one.

The report still shows something useful about where the change is happening. Retool names workflow automations and internal admin tools as the categories most often replaced. So the shift is mostly in the small tools a team runs on, not in anyone rewriting their accounting system.

Why this is happening now

AI coding tools have made a first version cheap. Retool found 51 percent of respondents had built production software their teams use today, and about half of those said it saved them six or more hours a week.

It also found the risk that comes with it: 60 percent had built something outside IT oversight in the past year. A tool that nobody in IT knows about has no backup, no owner and no one to call when the person who made it leaves. Building the first version was never the expensive part. Keeping it running is.

Build vs buy: when a custom web app makes sense

Before anyone writes code, we ask four questions.

1. Is the job specific to how you work? Payroll is the same everywhere, so buy it. The way you quote a job, schedule a crew or check a shipment may be what makes you different.
2. How often does it happen? A task done 50 times a day by five people is worth building for. A report made once a quarter is not.
3. Are you bending the tool? If your team keeps a spreadsheet beside the software because the software does not fit, the spreadsheet is your real system. That is the strongest sign we know.
4. Who owns it after launch? Someone has to fix it, update it and answer questions. If nobody can, keep paying the subscription.

If you answer yes to the first three and have a name for the fourth, building is worth a serious look. If not, the subscription is probably the better deal, even when the invoice annoys you.

Test the idea small

You do not have to decide on the whole thing at once.

• Write down one job, start to finish. Who starts it, what they enter, who checks it, what comes out. If you cannot write that on a page, you are not ready to build.
• Sit with the person who does the job. Watch them do it twice. The steps they skip when explaining it are usually the ones that matter.
• Build the smallest version that does that one job. Put it in front of real users for a few weeks, and keep the old tool running beside it.
• Count what changed. Hours saved, errors avoided or jobs finished. If the number does not move, stop there. You have spent little and learned a lot.

Our web apps team starts with this same build vs buy test, and will tell you if the subscription is the better deal. Read also what an AI-built app does not check.