A Customer Data Platform (CDP) is software that ingests customer data from every source, resolves identity across systems, and activates segments to downstream tools like ad platforms, email, and sales engagement. It is the operational layer between your data and your marketing tools.
What a CDP actually does
Three jobs define the category:
- Ingestion. Collect events from web, mobile, product, and server-side, plus batch loads from CRM, billing, and support.
- Identity resolution. Stitch anonymous device IDs to known emails to account IDs into a single profile. In B2B, this also means mapping people to accounts.
- Activation. Push segments and traits in near-real-time to ad platforms, email tools, sales engagement, and on-site personalization.
Anything that does not do all three is not a CDP. Tag managers, MAPs, and CRMs all touch customer data; none of them are CDPs.
The vendor landscape
The market splits into three groups. Packaged CDPs like Segment (now Twilio), mParticle, and Tealium own the original event-collection-and-activation pattern. ABM-flavored CDPs like 6sense and Demandbase add account resolution, intent data, and predictive scoring on top. Composable CDPs like Hightouch and Census skip storage entirely and run on top of your warehouse, doing only identity and activation.
When to buy
Buy a CDP when marketing operates more than five activation channels, when identity resolution across anonymous and known users is blocking campaigns, or when consent and compliance need a single source of truth. In B2B specifically, buy an ABM-flavored CDP when account-level signal is the work and you have no data team.
Skip the CDP if your stack is small enough that HubSpot or Salesforce native segmentation handles it, or if you have a warehouse and a data team and can run a composable pattern.
Pricing reality
Most CDPs charge by monthly tracked users or events. Costs scale aggressively with traffic. Get the activation list right before signing; vendors price the third destination very differently from the tenth.
Common pitfalls
- Treating the CDP as the warehouse. A CDP is for activation, not analytics. Keep the warehouse for analytical workloads.
- Underestimating identity work. Identity rules look simple in the demo. Real-world identity needs governance and testing every quarter.
- Activating before you’ve cleaned the source data. Garbage in, garbage at scale.
Related
- Data warehouse vs CDP — the build-versus-buy decision
- Reverse ETL — the alternative pattern
- 6sense — ABM-flavored CDP