1) Reconcile record counts and timestamps between systems
Pick a short window (yesterday or the last 24 hours) and export simple counts and the newest timestamps from each system involved — e.g. number of new contacts, orders or invoices and the most recent created/updated time.
Compare totals and the latest timestamp side-by-side: missing records suggest loss, a newer timestamp in the source but not the target suggests delay. This is a quick signal you can get in 10–20 minutes with basic exports or saved reports.
Common root causes and quick fixes:
- Filtering or date-range errors in the integration: widen the window and re-run the transfer or remove the filter temporarily.
- Schedule delays or processing queues: restart the sync or increase its frequency while you investigate.
- Dedupe rules rejecting records: pause downstream dedupe or mark a few test records to see how they're handled.
Low-effort controls: keep a daily snapshot (a simple spreadsheet row of counts and last timestamp), set a simple email alert if counts differ by more than a small threshold, and assign a data owner to check these once a day.
2) Sample key fields across systems to catch mapping or default-value issues
Pick 15–30 recent records that should have moved between systems and open them in both systems. Check the fields your automations rely on: email, order ID, status, owner, product code and any dates.
Look for common problems: fields that are blank in the target, truncated text, wrong default values, or values shifted into the wrong field. Pay special attention to fields used for routing (owner, status) because errors there cause downstream process failures.
Typical causes and immediate fixes:
- Field name/type mismatches: correct the mapping in the integration tool or change the field type in the target.
- Default values overwriting source data: change the target's default or add a conditional rule (only set default when source is blank).
- Truncation or format differences (e.g. dates/timezones): normalise format at source or add a small transformation rule.
Low-effort controls: keep a one‑page mapping cheat‑sheet for each integration, make a short acceptance checklist (three fields to verify) before turning on a change, and schedule a weekly 10‑minute sample check by the process owner.
3) Inspect error logs, retry queues and recent changes (apps, API keys, permissions)
Open the integration platform logs, the app’s webhooks or API activity page, and any retry or dead‑letter queues. Look for repeated error messages, spikes in errors, or a backlog of retries — these usually point to an obvious fault.
Also check for recent changes: new app installs, changed API keys, permission changes for service accounts, or software updates. Human changes are often the trigger, and they’re quick to spot once you know to look.
Immediate actions and controls:
- If you see auth errors: rotate or re‑enter the API key and reauthorise the connection, then monitor retries.
- For rate‑limit or timeout errors: reduce batch size, add short pauses between requests, or move to a less‑busy sync window.
- For repeated transform errors: pause the integration to stop further damage, fix the mapping, then replay or reprocess the failed queue.
Prevent recurrence with small controls: add an error‑count email alert, keep a one‑page runbook (who to call, how to reauthorise, where the logs are), and log any config change in a single place so it's quick to audit.
If you want a short runbook template or a quick health‑check you can run this afternoon, Optira can help prepare one for your team.