CDP Meaning: How Customer Profiles Get Built
In marketing and customer data management, CDP means customer data platform: software that keeps customer information connected over time and makes it available to other systems, as defined by the CDP Institute.
What does a customer data platform do?
The CDP Institute’s definition centers on three properties of the customer record:
- Persistent: information is retained so customer behavior can be tracked over time.
- Unified: data from different sources is linked to the same customer.
- Accessible: other systems can use the assembled information for analysis and customer interactions.
How customer profiles get built
1. Collect data from customer interactions
A CDP ingests information from connected sources, including three common categories described in Oracle’s CDP overview: behavioral data from websites, apps, and other channels; transactional data such as purchases and returns; and demographic data such as names and addresses.
Much of this is first-party data, meaning information collected through a company’s own systems and customer interactions, according to Oracle’s explanation.
2. Organize attributes and events
In Adobe’s profile data model, customer attributes describe a person through fields such as names and email addresses, while time-series events record actions such as cart additions, link clicks, and video views at particular times.
Adobe describes the combined profile as customer record data accompanied by related events.
3. Link identifiers through identity resolution
Identity resolution connects identifiers associated with a customer across sources. Twilio Segment’s documented identity graph supports cookie identifiers, device identifiers, email addresses, and custom identifiers, matching them to a persistent profile identifier.
The same documentation describes configurable identity rules and merge protection that control associations and address problems with non-unique identifiers. Identity resolution is a rule-based part of profile construction; the identifiers and matching behavior belong to the particular implementation.
4. Resolve conflicting values
Adobe’s merge-policy documentation explains how its platform prioritizes information when records from different sources contain conflicting attribute values.
Adobe supports two merge methods: dataset precedence, which ranks source datasets, and timestamp ordering, which favors the record most recently ingested into the platform. For timestamp ordering, recency refers to when the record reached the platform.
5. Maintain the profile as data changes
Oracle describes ongoing ingestion as the process that keeps customer interaction history updated as new information arrives.
For changes to existing attributes, Adobe documents profile corrections through batch ingestion into a profile-enabled dataset configured for updates.
6. Build audiences and activate data
Segmentation groups profiles according to shared attributes or behaviors. Adobe applies rules to profile data to produce audiences, with membership recorded against qualifying profiles through its profile service.
Activation makes that data usable in a destination system through connections such as the integrations in Adobe’s destinations documentation, which export audiences or datasets to marketing automation, advertising, CRM, and other platforms, with controls for selecting the fields to export.
What is a CDP used for?
Oracle lists CDP uses including segmentation, personalization, product recommendations, and connecting online and offline interactions.
A CDP can also supply a broader customer view to a CRM: Salesforce describes CDPs and CRMs as complementary systems, with the CDP unifying data and the CRM supporting relationship management.
CDP, CRM, DMP, and warehouse answer different questions
These systems can all handle customer information, but their main purposes differ:
| System | Full name | Main purpose |
|---|---|---|
| CDP | Customer data platform | Maintains unified customer records for other systems to use, according to the CDP Institute. |
| CRM | Customer relationship management | Organizes direct customer interactions, sales activity, service requests, and follow-ups, as Salesforce explains. |
| DMP | Data management platform | Builds audiences for advertising, traditionally using cookie-based data, as Oracle describes. |
| Warehouse | Data warehouse | Consolidates current and historical data for analysis and reporting, according to Google Cloud. |
Cloud data warehouses can also support streaming and real-time analytical workloads, as Google Cloud explains.
Packaged and composable CDPs
The CDP Institute’s current definition includes packaged, warehouse-native or composable, and dual-mode architectures; storage, identity resolution, governance, and activation services can be native or supplied by connected components.
In that definition, the CDP retains responsibility for customer identity, record structure, and downstream usability regardless of where those services run.
Profile timing and consent are separate parts of the workflow
Adobe supports batch, streaming, and edge audience evaluation, so an audience’s update behavior depends on the evaluation method.
Downstream timing also depends on the receiving platform: Adobe’s destination guidance explains that audience removals follow scheduled data flows and that their effective speed can vary with destination ingestion and processing.
Consent can be represented in the profile and used to determine collection, audience export, or channel participation, according to Adobe’s consent-processing documentation. Storing those preferences and enforcing them are separate functions: the documented Adobe Web SDK behavior automatically enforces collection consent, while more granular preferences require enforcement in downstream processes.