Martech, explained · Data foundations · 2:23
Reverse ETL, explained
The data exists, it just isn't where anyone works: what reverse ETL moves, its three parts, and the real work of keeping it right
Embed this film
Paste this where the film should play. It links back here.
The short answer
Reverse ETL takes what the warehouse knows and puts it back into the tools people use: the CRM, the email tool, the ad platforms. It has three parts: a model that says who or what to send, a mapping from column to field, and a sync that runs in batches and sends only what changed.
Questions this answers
What is reverse ETL?
ETL moves data from your tools into the warehouse: extract, transform, load. Reverse ETL runs the other way. It takes what the warehouse knows, a customer's product use, a lead score, a segment, and puts it back into the tools people work in: the CRM, the email tool, the ad platforms.
What are the three parts of a reverse ETL sync?
A model: a query, or a segment built without code, that says who or what to send. A mapping: which warehouse column goes into which field, in each tool. And a sync: how often it runs. Each run sends only what changed since the last one.
Where does reverse ETL sit in a composable CDP?
It is the activation piece. The warehouse decides who, with the model; reverse ETL delivers it to the tools. Some engagement and ad platforms now read the warehouse tables directly, without a copy, which means fewer copies to keep in step, but the segment's definition still has to live somewhere.
Is reverse ETL real time?
No. It runs in batches. Every hour is common, but it can be every few minutes or once a day. That is fine for most segments, lead scores and churn risk, where data from an hour ago still works. It is not for an abandoned cart that needs a reply in the next minute; that needs a stream that skips the warehouse.
What usually breaks in reverse ETL?
The sync itself rarely breaks. A column gets renamed in the warehouse, a field in the CRM goes empty, with no error, and nobody touched the CRM. That is the real work: not moving the data, but keeping each definition in one place, so the CRM and the email tool agree, and making sure someone hears about it when a column changes.
Transcript
Every chapter as text. Click a time to play the film from there.
0:00 Data that isn't where anyone works
A sales rep opens a customer in the CRM.
The warehouse knows that customer used the product forty times this week. The CRM shows nothing.
The data exists. It just isn't where anyone works.
0:16 What reverse ETL means
ETL moves data from your tools into the warehouse: extract, transform, load.
Reverse ETL runs the other way.
It takes what the warehouse knows and puts it back into the tools people use: the CRM, the email tool, the ad platforms.
0:35 Model, mapping, sync
It has three parts.
A model: a query, or a segment built without code, that says who or what to send.
A mapping: which column goes into which field, in each tool.
And a sync: how often it runs. Each run sends only what changed.
In a composable CDP, this is the activation piece.
The warehouse decides who. Reverse ETL delivers it.
1:05 Batches, and what they're for
It runs in batches. Every hour is common, but it can be every few minutes, or once a day.
That's fine for most segments, lead scores and churn risk, where data from an hour ago still works.
It isn't for an abandoned cart in the next minute. That needs a stream that skips the warehouse.
1:25 Reading the warehouse without a copy
And the category is changing.
Some engagement and ad platforms now read the warehouse tables directly, without a copy.
Fewer copies to keep in step. But the definition still has to live somewhere.
1:40 What usually breaks
The sync itself rarely breaks.
A column gets renamed in the warehouse, and a field in the CRM goes empty, with no error.
Nobody touched the CRM.
1:53 The real work
That's the real work. Not moving the data, but keeping each definition in one place.
And making sure someone hears about it when a column changes.
Back to that rep. Forty sessions this week, right there in the CRM.
Nobody typed it. The warehouse finally reached the tool where the work happens.
Learn more at Martech Stack Builder, or build your own.