2 of 4
Composable CDP
A composable CDP does the same four jobs as a packaged one, split into separate tools around your data warehouse, often from different vendors. You own the data and the choices, and part of the wiring between the tools is yours too.
- Best for
- A marketing team that wants more control, has a data team to run the warehouse, and is willing to take on the responsibility that comes with the flexibility.
- The trade
- The work a packaged vendor kept out of view lands with your data team: the identity rules, the schedules between tools, and noticing when one of them fails.
How it works
Break it
The boxes, and who holds each
- Website, app, store and CRM Marketing
Where customers leave a trace: page views, app events, purchases, account changes.
For example: HubSpot
- Event pipeline Your data team
Captures events and lands them in the warehouse, in a shape the data team decided.
For example: Twilio Segment, RudderStack, Snowplow
- Data warehouse Your data team
The hub. Every customer record lands here, and every other tool reads from here.
For example: Snowflake, BigQuery, Databricks
- Audiences and reverse ETL Your data team
Defines audiences on warehouse data and syncs them out to your channels.
For example: Hightouch, Census, GrowthLoop
- Email, ads and website Marketing
Where the audience is put to work: a send, an ad, a personalized page.
For example: Braze, Klaviyo, Meta Ads, Google Ads
Quick check
Did it land?
1. What does composable remove from the four jobs of a CDP?
2. How fast does reverse ETL usually get an audience from the warehouse to a channel?
3. A sync fails overnight. Who finds out first?
Build this architecture
Open it in Stack Builder with the template on the canvas, then make it yours.
Open in Stack Builder →