Reverse ETL
Why it matters
A warehouse that only feeds reports tends to become a place where good analysis goes to be admired. Reverse ETL closes the gap between knowing and doing. It also keeps logic in one place. If the definition of an "active customer" lives in the warehouse, the CRM, the email tool and the support desk all show the same answer. If each tool calculates its own, they drift apart and people argue about whose number is right.
How to apply it
- Define the metric or segment once, inside the warehouse, and test it there.
- Choose the one destination where a person will act on it, rather than syncing to every tool.
- Match the sync schedule to how fast the fact changes. A health score that moves weekly does not need a sync every minute.
- Treat the destination field as read-only, so nobody edits a value that the next sync will overwrite.
- Alert on failed syncs, because a stale score looks exactly like a current one.
What it is
Most data movement runs one way. Tools such as the CRM, the billing system and the website send their records into a data warehouse, where they are cleaned and combined (see ETL / ELT). Reverse ETL runs the other way. It takes a result computed in the warehouse and writes it back into a tool where somebody works. The result is often a score, a segment or a flag, such as "customer health: at risk" or "lead score: 82".
A simple example: the warehouse knows that an account logged in twice last month, filed three support tickets and has an unpaid invoice. That combined picture exists nowhere else. Reverse ETL writes it into a field on the account record in the CRM, so the account manager sees it without opening a dashboard.
Common mistakes
- Syncing raw tables instead of finished results, which fills the CRM with fields nobody reads.
- Writing to a field that sales reps also edit by hand, which causes silent overwrites.
