Sync Expensify with Google Sheets using n8n
Build a production Expensify ⇄ Google Sheets sync in n8n — auth, dedupe, retries and monitoring.
Key takeaways
- Keep Expensify the source of truth, mirror to Google Sheets.
- Dedupe on stable IDs — never on display names.
- Alert on drift daily, not per-event.
- Version the workflow JSON in Git.
Syncing Expensify with Google Sheets is one of the highest-leverage automations you can build in n8n. Done right it removes a whole class of copy-paste work; done wrong it creates duplicate records for a decade. This guide shows the exact n8n pattern that survives production.
Pick a source of truth
Before touching n8n, decide which system owns each field. In most spend setups Expensify owns identity while Google Sheets owns lifecycle state — but yours may differ. Write it down, get sign-off, then build.
A one-page conflict table pays for itself the first time an edit collides.
Wire up n8n
Create credentials for both Expensify and Google Sheets. Add a webhook trigger where the platform supports it, polling elsewhere. Normalise both payloads to a common shape with a Set node before any writes.
Add an idempotency key based on the Expensify record ID so replays never double-write.
- Trigger
- Normalise
- Lookup counterpart record
- Upsert
- Log
Two-way vs one-way
Start one-way, Expensify → Google Sheets. Prove correctness for a week. Only then add the reverse direction, with a loop-breaker (e.g. a last-updated-by flag) so events don't ping-pong.
Monitor, don't hope
Send a daily drift report — count of records in Expensify minus matched records in Google Sheets. Alert if drift crosses a threshold. Silent syncs erode trust faster than loud failures.
Frequently asked questions
- Do I need queue mode?
- Only above roughly 10k events per day. Below that, a single n8n instance handles Expensify ⇄ Google Sheets comfortably.
- How to handle deletions?
- Soft-delete in Google Sheets with a tombstone field. Never hard-delete on receipt of a Expensify delete — you'll lose audit trail.
- What if schemas drift?
- Fail closed — the workflow should stop, alert and wait for a human. Silent schema drift is the #1 source of bad data.
- Can I self-host this?
- Yes. Any n8n deployment (Docker, Coolify, Railway) works for a Expensify ⇄ Google Sheets sync of this shape.