What is a Customer Data Platform (CDP)?

What is a Customer Data Platform (CDP)?

A Customer Data Platform (CDP) is packaged software that collects customer data from many sources, resolves it into one persistent profile per person, and makes those profiles available to other systems. The CDP Institute, which popularised the term, defines the category as software that creates a persistent, unified customer database that is accessible to other systems.

The category exists because customer data usually lives in silos: analytics, CRM, email, support, ad platforms and the data warehouse each hold a partial copy with their own idea of who a "user" is. A CDP is meant to be the one place where those copies are stitched together and sent back out.

What a CDP does

Reading the definition closely gives four jobs:

• Collect. Ingest events and attributes from web and mobile SDKs, server-side APIs, webhooks, integrations with tools such as billing or support, and bulk imports.
• Resolve identity. Match anonymous IDs, user IDs, emails and device identifiers into one profile, so the same person on a phone and a laptop is not counted twice. This is the hardest part in practice, and it is never perfect.
• Store the profile. Keep events and traits per person in a database that the business can query and export, not one locked behind a single vendor screen.
• Activate. Send audiences, traits and events to downstream tools such as ad platforms, email tools, support systems, a personalization engine and the warehouse.

An example

A shop's visitor browses a product anonymously, adds it to a basket, then signs up by email on their phone a day later. The CDP receives the page views, the cart event and the signup, and merges the anonymous activity into the profile created at signup. Marketing builds a segment of "added to basket, did not buy, last seven days". The CDP pushes that segment to the email tool and to an ad platform audience, and when the person buys, the next event removes them from it.

Three common architectures

• Traditional CDP. The vendor collects, stores and activates the data in its own database. Quick to start, but it can become one more silo to keep in sync.
• Composable or warehouse-native. Your data warehouse holds the profile, and a tool reads from it and activates to other systems. It reduces data copies, but a data team has to build and maintain the profile logic.
• Suite-bundled. A CDP shipped with a marketing or CRM suite. Activation into that suite is strong, other destinations less so.

What a CDP is not

• Not a CRM. A CRM supports sales and support workflows around known contacts. A CDP unifies behavioural and identity data and feeds other tools, including the CRM.
• Not a DMP. A data management platform traditionally works with anonymous, mostly third-party data for ad targeting. A CDP works with first-party data about your own customers.
• Not an analytics tool. Analytics tools analyse behaviour. A CDP unifies it and sends it elsewhere, though the two overlap.
• Not your data warehouse. The warehouse stores everything for analysis. A CDP is the operational layer for customer-facing use.

Common mistakes

• Buying before defining use cases. Decide which segments and destinations you need first, then evaluate vendors against them.
• Assuming identity resolution is clean. Shared devices, email aliases and blocked cookies leave real gaps. Plan for partial matches.
• No event schema. Without agreed event names and properties, the CDP fills with inconsistent data nobody trusts.
• Letting the CDP and warehouse drift apart. Decide which one is canonical for each field.
• Ignoring consent. Collecting data does not give you the right to use it for every purpose. Pass consent state along with the data.
• Not watching the pipeline. If collection fails quietly, profiles go stale and segments shrink without any error.

CDPs, monitoring and load testing

LoadFocus is not a CDP. It matters only where a CDP touches your own infrastructure. If you run a collection endpoint or proxy that every page view and click goes through, API monitoring can check it on a schedule and alert when it errors or slows down (see API monitoring), and load testing shows whether it holds up at peak traffic, such as a sale (see load testing). A broader checklist is in our monitoring best practices article.

FAQ: Customer Data Platforms

Do I need a CDP if I already have a data warehouse?

Not necessarily. If you have a warehouse and a data team, a warehouse-native approach can cover profile building, and you only need a tool for activation. A packaged CDP helps most when you lack that team or need identity resolution and activation working quickly.

What is the difference between a CDP and a CRM?

A CRM tracks contacts and deals for sales and support. A CDP unifies behavioural and identity data from many sources and passes it to other tools, the CRM among them.

How is a CDP priced?

It varies by vendor. Common models charge by the number of monthly tracked users or profiles, by event volume, or by connected destinations. Ask each vendor how the price grows with your traffic before you compare.

Is a CDP GDPR-compliant by default?

No. A CDP can provide tools for consent, data export and deletion, but you have to configure them and decide the lawful basis for each use. Compliance depends on how you use it.

Why are CDPs tied to first-party data?

A CDP is built around data you collect directly from your own customers, such as sign-ups, purchases and in-product behaviour. That data is yours to use within your privacy notices, unlike third-party cookie data that browsers increasingly restrict.