Build vs buy: when Zapier is the right answer

Two hosts take the side that loses business where it's true — most 20-person firms should start with the boring tool — and get specific about the thresholds that flip the decision toward custom.

A: Someone asks for a custom integration roughly once a week, and the honest answer, most of the time, is: you don't need me for this. You need a tool you're already paying for, and about forty minutes.

B: Which sounds like a strange thing for someone who sells custom builds to say out loud.

A: It's not false modesty, though. It's the same scoping discipline as everything else in this season — decide what a workflow actually needs before deciding who builds it.

B: A twenty-person business paying for custom software to reproduce a form-to-CRM connector isn't buying capability. It's buying a maintenance obligation it didn't need.

A: So when's the existing tool the right call?

B: When the flow is a standard shape. Form submission to CRM. New row to a message. Calendar event to email. If you can describe it in one sentence and a stranger would recognise the shape, someone's already built the connector.

A: And when volumes are modest. These platforms price per task or per run, and at low volume that's cheaper than owning infrastructure — even infrastructure that costs nothing to run, because someone still has to maintain it.

B: Don't pay anyone to reproduce that. Including us.

A: When does custom actually earn its cost, then?

B: When logic branches on your own rules. "Route to the senior tech if the job's over a certain amount and the customer's called twice this month" — that's not a standard connector shape. Once branching depends on knowledge that only lives in your business, off-the-shelf tools run out of road fast.

A: When data can't leave your environment — some contracts and some regulators require that, and there's a real, separately priced option for it.

B: And when you need telemetry the tool won't give you — a run log with per-step outcomes, visible retries. Most platforms give you a green checkmark or a red one and not much in between.

A: So what are the actual thresholds that flip it?

B: Four signals. Volume — low stays comfortable on a platform, high enough that per-task pricing dominates is worth calculating against ownership. Number of systems touched — one or two well-supported favours the tool, several or one with no API favours custom.

A: How much the logic changes — rarely favours the platform's own interface, branching on rules in someone's head favours code. And who maintains it — a platform interface someone already knows, or a run log someone can actually debug.

B: No single signal decides it alone, worth saying. A high-volume simple workflow often stays comfortable on a platform. A low-volume workflow with real branching can be worth building anyway.

A: Most real projects turn out to be a mix, honestly — the platform for the mechanical connection, custom logic for the specific part that needs it. Purity isn't a virtue here.

B: One honest limit — we have an obvious incentive to say "build it." This argument's priced deliberately against that incentive, which is worth knowing as you listen to it.

A: One thing to check this week.

B: Describe your workflow to a stranger, in one sentence. If they recognise the shape, price the platform first. If the sentence needs an "unless," that's the signal worth pricing against custom.

A: Written version's at atinnovators.in, under insights.