Skip to main content

Do you let Zaps pay invoices without someone clicking approve?

  • Nov 6, 2026 5:15 PM UTC
  • 3 replies
  • 14 views

Curious how people handle this in practice. If you have a Zap that pays supplier invoices or sends bank transfers (Wise, Stripe payouts, bill pay, bank API, whatever):

 

- does it pay on its own, or does someone approve each one?

- if some go out on their own, where do you draw the line (amount, known supplier, new bank details)?

- has anything gone wrong, like a double payment or changed bank details?

- has a client or accountant ever asked who approved a payment?

 

Also keen to hear from people who decided not to automate payments at all and why. Asked the same in the n8n forum and got really good answers, so trying here too.

3 replies

WebhookEngineer
Forum|alt.badge.img+1

I’d handle payment approval separately from preventing duplicate payments.

My default would be to require approval for new recipients, changed bank details, or payments outside explicit per-payment and daily limits. Knowing the supplier wouldn’t be enough on its own. Tie approval to the exact recipient, bank details, amount and currency. If any of those change, the approval should no longer be valid. Record who approved that version and when.

Retries need their own protection. If a payment request times out, treat the outcome as unknown and reconcile with the provider before submitting another payment. Use a stable payment ID and provider-supported idempotency where available. Don’t rely on a “have we paid this?” lookup alone, since concurrent runs could both pass that check.

I’d first check whether the payment provider can enforce approval and keep the audit trail, with the Zap preparing the requests.


For repeat vendors, I’d automate the checks but keep a human in the release step: match invoice number, vendor, and amount against an unpaid ledger and purchase order, then hold duplicates, changed remittance details, or totals outside tolerance. Make approval expire if the invoice or bank details change, and require the approver to be different from whoever edited the vendor record. Log the source document and approval decision for audit. That keeps routine preparation fast while exceptions stop before money moves; I’d avoid a fully unattended first release.


One angle the replies above don't cover: checks on the supplier itself, before the first payment goes out.

What I'd put in the Zap:

  1. Trigger on a new or updated vendor (QuickBooks Online "New Vendor", or a Find step that compares the bank details stored on the vendor with the ones on the invoice).
  2. Paths by Zapier: Path A when the vendor is new or the bank details changed, Path B for known vendors with unchanged details under your amount limit.
  3. On Path A, screen the supplier name first. The US Consolidated Screening List has a free API (trade.gov) you can call with a Webhooks by Zapier GET step; for EU and UK names the EU Sanctions Map and the OFSI consolidated list are free to check. Save the result and the date on the vendor record.
  4. Then a Human in the Loop "Request Approval" step (or a Slack message with an approval link) that shows the approver the screening result, the old and new bank details and the amount. Nothing is paid until it comes back approved.
  5. Confirm changed bank details outside email: call the supplier on the number you already had on file, not the one printed on the invoice. A spoofed "we changed banks" email is the most common way these payments go wrong.

Log approver, timestamp and screening result in a Google Sheet or in the accounting app, so you can answer the accountant's "who approved this?" in seconds.

Disclosure: I work at Unicorn Private; our Zapier integration Unicorn Intelligence does this check, but the steps above work with any tool.