Import the blueprint
Create a new scenario, choose Import blueprint and pick the JSON from the ZIP. Leave it disabled while you configure it. No connections or IDs are inside.
After a 503 or another failed append, you cannot tell whether the row landed. This scenario records each request key as pending in a ledger before the append, marks it complete with the target row, or holds it as needs_reconcile, so nobody retries a write that may already be there.
This page contains an affiliate link, marked “affiliate link”.
Run in Make on 9 Oct 2026 with fictional data. Only the failed-append path was tested: the failure was forced with a missing target tab. A real Google 503 was not triggered. In that run: about 5 credits per completed write, 2 for a repeat key (no second append), and 5 for a failed append that lands on needs_reconcile.
Runs in your own Make account with your own Webhooks and Google Sheets connections. Nothing runs through Flowpaja.
The scenario searches a Write ledger for the request key. A new key is recorded as pending, then one row is appended to Target rows, and the same ledger row is updated to complete with the target row number. Any existing ledger key stops a new run, so a replay does not append again.
If the append fails, the ledger row becomes needs_reconcile with the error type and message, and a Break handler stores the failed bundle with automatic completion off. Before retrying, search Target rows for the exact key once Sheets is reachable: one match means the append landed.
Only the failed-append path was tested, with a forced error. A real Google 503 was not triggered, so check in your own account that the handler also holds ConnectionError and 503-shaped failures. A Sheets search plus append is not atomic: keep this scenario the only writer to these keys.
The download includes a connection-free Make blueprint, setup README, fictional sample cases, a credit estimate and test notes.
Short path from the README. Full steps, sample cases and credits are in the download.
Create a new scenario, choose Import blueprint and pick the JSON from the ZIP. Leave it disabled while you configure it. No connections or IDs are inside.
Write ledger with request_key, status, error_type, error_message and target_row. Target rows with request_key and note. Keep them separate.
Point modules 2, 3, 5 and 7 at the ledger and module 4 at Target rows. Search Rows matches the key with Limit 2; module 3 continues only when Total number of bundles = 0. Modules 5 and 7 use the new ledger row number.
Module 6 (Retry): Automatically complete execution = No. Keep module 4 a single Add a Row with no automatic second append. Turn on Process data in order and Store incomplete executions.
Run the README cases one at a time in a disposable copy: new key, repeat, blank key and a forced append failure. Compare the ledger and Target rows with TEST_NOTES.md.
Credits measured in one fictional-data run; setup time is an estimate.
Measured in the 9 Oct 2026 run (scenario operations): 5 credits for a completed write, 2 for a repeat key, 1 for a blank key, and 5 for a failed append that lands on needs_reconcile. A real 503 was not measured. Check scenario history for your own numbers.
Setup takes roughly 20 to 40 minutes once your sheet is ready. That is a planning estimate, not a timed measurement. Make Free has 1,000 credits/month and two active scenarios.
Use this when duplicated or missing rows after a write error matter more than speed. To diagnose the error itself, read Google Sheets 503 service unavailable in Make.
For deduplicating incoming webhook data without a reconcile step, see Webhook to Sheets dedupe. If these rows feed a lead handoff, the paid Lead Router covers form to HubSpot and Slack.
More free templates: Free Overdue Digest · Webhook to Sheets dedupe · Lead capture with dedupe.
Questions: flowpaja.fi@gmail.com.
Tools we use for our builds: our tools page.
Make, Google and Google Sheets are trademarks of their owners. Flowpaja is independent and not affiliated with or endorsed by them. Flowpaja develops its templates with AI assistance.