We've discussed in other blog articles how Customer Data Platforms (CDPs) have been part of the data marketing ecosystem for about twenty years. Often criticized, they have nevertheless endured and remain the standard for collecting, storing, and managing customer data.
The emergence of Composable CDPs since 2021 shows that the CDP market still has promising days ahead.
If you've already read various resources on traditional CDPs (also known as packaged ones) vs. Composable CDPs, you probably don't need another article explaining which approach is better than the other.
This guide aims to assist decision-making processes by providing a clear vision and understanding of the different components of a CDP, and how this applies to traditional CDP and composable CDP.
Definition of Customer Data Platforms
Many players position themselves in the CDP market without necessarily worrying about the definition of a CDP.
👉🏼 To understand the nuances between different solutions, let's first define concretely what a Traditional / Packaged CDP is.
Traditional CDPs
A Packaged CDP is a ready-to-use platform that offers predefined and standard functionalities for collecting and storing data from various sources, for transforming and aggregating it, building audiences, and sending data to marketing destinations.
A traditional CDP may also offer features to ensure data quality, adhere to governance protocols, and comply with privacy regulations.
Reading this definition, you can realize several things:
A traditional CDP must store the data it collects to aggregate and unify profiles (ID resolution). This means that your customer data is stored (or duplicated) in a third-party tool.
CDP functionalities are predefined and are often sold in "blocks". So, you can buy package by package, but this implies an additional cost each time. Since the functionalities are predefined and designed for the majority of companies, some limitations in terms of customization and scalability may be encountered.
🤔 What about a Composable CDP then?
Composable CDPs
A Composable CDP is a modular solution that collects, models, and activates customer data from your existing data infrastructure. It relies on a set of tools (open-source or software) to perform some or all of the functionalities of a traditional CDP: collecting, storing, and activating customer data.
So, there are several considerations to keep in mind:
A Composable CDP is not required to have all the functionalities of a traditional CDP: it only offers the bricks that meet the needs of a company at any given moment. It allows starting a CDP strategy quickly, implementing the first use cases in a few days, and then gradually advancing on the project.
👉🏼 Discover our "modular" approach to CDP
A Composable CDP only exists by assembling tools that serve the same purpose as the bricks of a packaged CDP: a tool for collection, a tool for storage, a tool for activation, etc. That being said, this allows obtaining a Composable CDP that seamlessly integrates with your existing data environment (rather than operating as a separate entity): you don't have to change tools for all functionalities.
Selecting different tools allows flexibility in data schemas and how the purpose is achieved. You have complete control over your data (and how it is collected, stored, and activated). In my opinion, that's the real advantage of a Composable CDP.
Let's compare the two options below!
Comparison guide
Our One-pager highlighting the differences between Traditional CDP and Composable CDP
Components of a Customer Data Platform
In this section, we mention the main components of a CDP and how these functionalities are covered by a packaged CDP and a Composable CDP.
Attention, the selected functionalities are indicative: not all publishers necessarily have them or manage them in the same way. And that's okay: a CDP doesn't have to have all these components to meet your needs.
Schematic diagram of a CDP
Data Collection in Proprietary Tools: A CDP, whether packaged or composable, necessarily relies on customer data. It is therefore necessary to collect this data, which can come from various "first-party" sources (application, website, etc.).
In the case of a Packaged CDP, the Customer Data Infrastructure (focusing on collecting and storing customer data) allows sending data to third-party tools, without having to store it in a data warehouse.
In the case of a Composable CDP, companies adopt Customer Data Infrastructures as "stand-alone," which have data warehouses as primary destinations. Few third-party tools are supported as additional destinations.
Integration of Third-party Data: Customer data can also come from third-party data sources, such as advertising platforms, payment platforms, support platforms, etc. To collect and store the data in a base that can then be exploited by the CDP, it is essential to go through an ETL (Extract, Transform and Load) process.
In the case of a Packaged CDP, third-party data is first stored in the CDP's database (i.e., outside of your company's data infrastructure), and then it can be integrated into other proprietary destinations. If the destination is not supported by your CDP publisher, you must create your own data pipeline.
In the case of a Composable CDP, ETL solutions (such as Fivetran or Airbyte) are used to send data from "third-party" sources to the data warehouse.
Illustration of ETL process
Data Storage: A key feature of a CDP, it allows for temporal studies and segments based on past actions.
In the case of a Packaged CDP, the data is stored in a database owned by them. It copies and duplicates data that could exist in your infrastructure.
In the case of a Composable CDP, the data is stored in a data warehouse. The most well-known actors in the market are BigQuery, Snowflake, or Redshift.
🏆 DinMo is the only European CDP that is Google Cloud BigQuery Ready!
Identity Resolution: This component allows unifying the different information collected about a customer from various sources, with the aim of creating a unique profile.
In the case of a Packaged CDP, identity resolution is done directly by the publisher, either using probabilistic or deterministic methods*.
đź’ˇ *Probabilistic Approach: uses statistical models to manage uncertainty and variability, predicting whether it is (or not) a unique person.
*Deterministic Approach: relies on fixed rules and relationships where each person is predictable with certainty.
In the case of a Composable CDP, this identity resolution is done directly by the company (either in the data warehouse with SQL joins, or on activation platforms, such as DinMo).
Audience Building: This essential functionality for CDPs allows creating audiences that will be shared with destinations. This mostly corresponds to a "Visual Builder" that allows adding segmentation conditions via "drag and drop."
In the case of a Packaged CDP, the "Visual Builder" is a full-fledged feature, based on the predefined "Data Model." However, these models are rigid, and users cannot build any type of object.
In the case of a Composable CDP, this functionality is not always available, as it depends greatly on the type of Reverse ETL that has been chosen to build the CDP. Some Reverse ETLs are only used as data pipelines, and in these cases, audiences must be created by coding. Conversely, some Reverse ETLs have added no-code Visual Builder features and are now designated as data activation platforms.
Our SQL-free segment builder
Reverse ETL: This process refers to transferring data from a data warehouse to third-party destinations (ads, CRM, support, etc.).
In the case of a Packaged CDP, this process, known as orchestration, involves moving data from the CDP's database to useful destinations.
In the case of a Composable CDP, a packaged Reverse ETL tool (or a data activation tool) is used for this process.
Illustration of Reverse ETL process
A CDP, whether packaged or composable, may very well have other feature bricks. The most common are "data quality" (to ensure the consistency and accuracy of shared data) and "governance" (especially to ensure compliance with privacy regulations).
With the emergence of Artificial Intelligence, CDPs now offer new features: predictive attributes, predictive segmentation, and decision support. This component is often offered as an add-on for both packaged CDPs and Composable CDPs.
đź’ˇIf you want to see DinMo's AI capabilities, check our Galeries Lafayette customer story.
Packaged CDP vs. Composable CDP: Which One to Choose?
Before choosing between a Packaged CDP and a Composable CDP, it is essential to evaluate the specific needs of your company.
If you are looking for a turnkey solution and don't necessarily have a dedicated data team to build your stack, the packaged CDP can be considered. Conversely, if your company has complex needs and is looking for a fully customizable and scalable solution, the Composable CDP will probably be a better option.
Evaluate the existing tools in your company (tools already used, available resources, etc.) and the goals to be achieved to choose a solution that is suitable for today as well as tomorrow.
It is recommended to consult experts in customer data management to get personalized advice and make the best decision for your company. Do not hesitate to reach out to our partners.
Feel free to contact us if you want to learn more about our approach to Composable CDPs.
