What is a Customer Data Platform (CDP)?

By Mariah Blackmore

On Jul 14, 2023

Cloud Technologies
assorted electrical cables Photo by @barkiple - John Barkiple from Unsplash

Other Parts of the Series

  1. Part 1 - What is a Customer Data Platform (CDP) <-- This Article
  2. Part 2 - When do you need a Customer Data Platform (CDP)
  3. Part 3 - How to Get Ready for a Customer Data Platform (CDP) - Coming Soon!

What is a CDP?

More and more software out there is calling itself a Customer Data Platform, but what does that actually mean? What makes one piece of software an Email Marketing Platform and not a CDP?

For our purposes, we define a CDP as software that primarily focuses on doing the following things:

  1. It is used to centralize most, if not all, aspects of data collection into one single area.
  2. It can facilitate data collection, (some) processing, and distribution of collected data to a catalog of third party destinations and services, including cloud data warehouses.
  3. It can accept incoming data from a variety of sources, such as websites, mobile apps, server-side generated events, or even raw webhook data.

When we at Obsessive Analytics talk about CDPs, we refer to tooling that handles these three key aspects of customer data collection, processing, and distribution.

What can a CDP do?

Besides allowing you to own your own first-party data, CDP's main selling point tends to be the relatively frictionless distribution of event data to downstream destinations, removing the need to implement many tools over and over again in order to capture the data that you need. For example, common CDP use-cases are:

  1. Centralizing first-party data collection of event stream data such as pageviews, signups, purchases and more. This is often sold as "implement once, reuse many times".
  2. Handling ETL and Reverse ETL duties. ETL means to Extract, Transform, and Load data from third party sources such as CRMs, payment systems, and more into a data lake / warehouse. We refer to this as catalog data since it represents more a catalog of records and their current state, rather than a specific, singular event.
  3. Creation and maintenance of a data warehouse or data lake (or data lakehouse if you want to play buzz-word-bingo). A great benefit of data collection via a CDP is that you get an extremely standardized warehouse schema for analysts to work with. Where can you find information about people signing up to the webapp? - Try the webapp.signups table!

These use-cases can be accomplished with pretty much any CDP that fits our above definition. From here, many CDP vendors such as RudderStack, Segment, mParticle, Snowplow then have varying features built on top of this foundation.

Why choose a CDP?

The number of tools in a company's tech stack grows every year. Just check how many "marketing pixels" are on your website! A CDP can help reduce the time and effort involved in ensuring that everyone from product, to marketing, to sales and customer success can have access to the correct data in their tool of choice. A few other key benefits of CDPs include:

  1. With a CDP, and data warehouse you can have a single source of truth for data analysis and reporting.
  2. With a CDP, you don't rely on point-to-point integrations between tools, since the CDP handles every tool getting fed the same data.
  3. With a CDP, tracking code is generally implemented only once which reduces a lot of time spent setting up data pipelines and also improves your application's performance, since it reduces page bloat by sending data to your destinations server-side instead of having to do it all client-side.
  4. Last, but definitely not least, with a CDP, you can actually fully own your first-party data.


Stay tuned for parts two and three in this series to learn more about at what point having a CDP makes sense for your organization's use-case, as well as what kind of prep work might be required to successfully implement a CDP. Because remember, anyone selling you a piece of software, that magically solves all your problems, is probably less than 100% truthful!

© 2024 Obsessive Analytics Consulting. All rights reserved.