Build vs buy: when Zapier is the right answer

Someone asks us to build a custom integration roughly once a week, and the honest answer is usually "you don't need me for this — you need Zapier, and about forty minutes."

That is not false modesty. It is the same scoping discipline as everything else on this site: decide what a workflow actually needs before deciding who builds it. A 20-person service business paying for custom software to reproduce a form-to-CRM connector is not buying capability. It is buying a maintenance obligation it didn't need.

Use an existing tool when

• The flow is a standard shape. Form submission to CRM record. New row to Slack message. Calendar event to email. If you can describe the flow in one sentence and a stranger would recognise it, someone has already built the connector.
• Volumes are modest. Existing platforms price per task or per run, and at low volume that is cheaper than owning infrastructure, even infrastructure that costs nothing to run — someone still has to maintain it.
• The integrations you need already exist. Check before assuming otherwise. The catalogue of pre-built connectors is larger than most owners expect.

Do not pay anyone to reproduce Zapier. That includes us. If a scoping call ends with "actually, just use the tool you're already paying for," that is not a failed sales call — it is the correct outcome, and it is one we reach more often than you'd expect from a company that sells custom builds.

Build custom when

• Logic branches on your own business rules. "Route to the senior tech if the job is over $2,000 and the customer has called twice this month" is not a standard connector shape. Once branching depends on knowledge that only exists inside your business, off-the-shelf tools run out of road fast.
• Data can't leave your environment. Some contracts and some regulators require it. Self-hosted inference is a real, separately priced option for exactly this case — a model running on infrastructure you control, not a third-party platform processing your customer data as a side effect of routing it.
• You need telemetry the tool won't give you. A run log with per-step outcomes, visible retries, a state you can reason about after a partial failure — see automation you can actually inspect. Most off-the-shelf platforms give you a green checkmark or a red one and not much in between.
• Per-task pricing has crossed over ownership cost. There is a real crossover point where volume makes the platform's per-task fee more expensive over a year than building and running the equivalent yourself. Calculate it before assuming either direction — owners guess wrong in both directions about as often.

The thresholds that actually flip it

Signal	Off-the-shelf	Custom
Volume	Low to moderate	High enough that per-task pricing dominates
Number of systems touched	One or two, well-supported	Several, or one with no usable API
How much the logic changes	Rarely	Branches on rules that live in someone's head
Who maintains it	You, using a tool's own interface	Needs a run log and someone who can debug a failure

No single row decides it. A workflow that is high-volume but simple often stays on a platform quite comfortably. A low-volume workflow with branching business logic can be worth building anyway, because the cost of a wrong routing decision is high even if it happens rarely.

Most real projects are a mix

Purity is not a virtue here. A common shape we build: the platform handles the mechanical connection — new form submission triggers a webhook — and custom logic picks up from there, because the branching and the telemetry are the part a connector can't give you. Paying for infrastructure to reproduce the trigger would be waste; paying for a connector to encode branching business rules would be a worse and slower version of what a short custom step does cleanly.

Honest limits

• We have an incentive to say "build it." We priced this article deliberately against that incentive — read it as the argument that costs us business where it is true, not marketing copy.
• "Off-the-shelf" is not free. Per-task platform pricing at real volume is a genuine cost, not a rounding error, and it is worth calculating rather than assuming.
• Migrating later is real work. A workflow that starts on a platform and later needs custom logic is not a wasted step — it usually means the platform earned its cost while the volume was low — but the migration itself is a project, not a toggle.

How to tell if this is your decision

Describe the workflow in one sentence. If a stranger recognises the shape, price the platform first. If the sentence needs an "unless" or a "depends on," that branching is the signal that custom is worth pricing against it.

← Back to AI for owner-led service businesses