Knowledge base
What is a Customer Data Platform (CDP)?
A CDP is the tool that turns a pile of scattered customer data into one trustworthy profile per person, and makes that profile available to every other tool. Here is what that means and where the line sits.
7 min read · Updated June 28, 2026
The canonical definition
The CDP Institute, the industry body that coined the category, defines a CDP as packaged software that builds a persistent, unified customer database that is accessible to other systems. Three words carry the weight.
- Packaged: it is bought and run by the marketing team, not built as a custom engineering project.
- Persistent and unified: it stores data over time and resolves it into a single profile per real person, not a momentary snapshot.
- Accessible: the profiles are not locked in a silo. Other systems, especially the sending platform, can read them.
What it actually does
A CDP has four core jobs. The CDP Institute formalizes these into a seven-point "RealCDP" test, but these four are the heart of it.
- Ingest data from any source: the website, the app, the email tool, the warehouse, support tickets.
- Resolve identity: stitch the anonymous browser, the email address, and the logged-in user ID into one profile.
- Segment: let a marketer build audiences from that unified data without writing code.
- Activate: push the profiles and segments out to the tools that send messages or buy ads.
Identity resolution, briefly
The hard part is connecting a person to themselves across devices and sessions. Deterministic matching uses confirmed identifiers (a shared email or hashed ID) and is near-certain but narrow. Probabilistic matching uses statistical signals (device, behavior) for wider but less certain reach. Most CDPs blend both.
What a CDP is not
Most confusion about CDPs is really confusion with three neighbors.
- Not a data warehouse. A warehouse is a general-purpose, engineering-owned store for all company data. A CDP is purpose-built for customer profiles, with identity resolution baked in and a marketer at the controls. See the side-by-side.
- Not a DMP. A data management platform handles anonymous, cookie-based audience segments for ad targeting. A CDP holds known, first-party, persistent individual profiles.
- Not an engagement platform. The sending platform (the category Iterable is in) acts on the data by delivering messages. The CDP assembles and holds the data. One is data-out, the other is data-in.
The category is shifting
The classic definition says "packaged software," but the CDP Institute now also recognizes the composable or warehouse-native CDP, where the warehouse is the source of truth and CDP capabilities are assembled on top of it. Do not treat "a CDP must be packaged software" as settled.
Frequently asked questions
Is Iterable a CDP?
No. Iterable positions itself as the cross-channel engagement layer, not the data layer. It receives unified data from a CDP or warehouse and acts on it. The vendor guidance is to pair the platform with a CDP, not to use one in place of the other.
Do I need a CDP and a data warehouse?
Often you have both. The warehouse is the broad analytical store; a packaged CDP adds marketer-friendly identity resolution and activation. A composable CDP instead builds those capabilities directly on the warehouse so there is only one copy of the data.
What is identity resolution?
The process of merging scattered identifiers and events into one profile per real person, including attaching a previously anonymous visitor to their known identity once they log in.
Keep reading
Sources
Part of the MarTech and customer engagement field notes on this site. Written by Antoni K Pestka.