βΆWhat's the difference between Census Sync and a webhook or API call?
Webhook = event-triggered one-way push (user signs up, fire email). Census Sync = bidirectional, SQL-powered, scheduled or real-time syncs. Census handles backfill (retroactive updates), transformations (dedupe, aggregate), and partial syncs. Webhook is simpler but requires custom code; Census is more powerful and requires no engineering.
βΆCan I sync deletes/updates or only inserts?
All three. Census Sync tracks row changes (UPSERTs, deletes). If a customer unsubscribes in Salesforce, Census can mark them as opted-out in your warehouse. Deletions are tricky though, most platforms don't support deleting records, only marking as inactive/archived.
βΆHow long does a sync take from warehouse to CRM?
Real-time syncs trigger within 30-120 seconds of data warehouse refresh. Scheduled syncs run on intervals (hourly, daily, every 15min). CRM API rate limits (Salesforce = 15 calls/sec) can queue jobs. Bulk operations backfill faster than one-by-one. Typical: 10k records = 2-5 minutes.
βΆWhat if the downstream system (CRM) is down during sync?
Census retries with exponential backoff (3-5 retries over 1-2 hours). Failed batches are queued and can be manually retried. Partial syncs are rare; most syncs all-or-nothing. Build monitoring alerts so you know if a sync stalls for >30min.
βΆCan I sync from Postgres or only from data warehouses?
Both. Census works with operational DBs (Postgres, MySQL) and warehouses (Snowflake, BigQuery, Redshift). Postgres is lower-latency but needs SSL/VPN setup. Warehouse syncs are easier to audit (query source in SQL directly). Most companies use warehouse for audit/compliance.
βΆHow do I debug if a sync failed for some rows?
Check Census > Logs > Failed Rows. Common issues: (1) duplicate key in destination (non-unique email in CRM), (2) required field missing from source SQL, (3) field mapping mismatch (expecting int, got string). Fix the query, resume from failure point. Census shows row-level errors, not just batch-level.
βΆWhat's the salary progression from analytics engineer to reverse-ETL specialist?
Analytics engineer ($90-130k) β Reverse-ETL/Activation engineer ($110-160k) β Data ops lead ($140-200k). The jump comes from owning end-to-end customer data activation (not just BI dashboards). Scarcest skill: SQL + workflow design + business intuition. Companies desperate for this post-Series A.