Note: This article was drafted for my employers’ blog but never quite made it for various reasons; I present it for your private enjoyment. ed.
When you’re choosing a customer data platform, should you strap on the tool belt and go DIY or look for a full-featured solution? We’ll explain the difference and help you decide.
Composable customer data platform (CDP) is a term that puzzles a lot of mar-tech buyers. What does it mean? Is a composable CDP truly a new invention, just clever marketing, or something in between? Let’s break it down – and figure out what kind of CDP is right for you.
Currently, you’ve got two major CDP options to help you store and organize your customer data: full-featured or composable. What’s the difference? Adopting a full-featured CDP is like having furniture delivered to your home; a composable CDP is more like getting the kind of furniture you assemble yourself.
You can use full-featured CDPs to access data from key sources, harmonize it, manage identity, build and activate audiences for marketing, service and more. Composable solutions claim to provide a flexible architecture that can be adapted to your business requirements.
The good news is that you can combine the benefits of full-featured CDPs and composability without compromising on features or flexibility.
For example, you can maintain a modern cloud data warehouse and still have push-button access to customer relationship management (CRM) data, as well as all the features you require for CDP use cases, without having to select and maintain a portfolio of applications.
How can you do this?
Short answer: adopt a full-featured CDP that has composable benefits like modularity.
Longer answer: in this blog post, I’ll review what composability offers and the requirements for a full-featured CDP. Then we’ll talk about how to get the best of both.
But first, let’s define what a composable CDP is and how it differs from a full-featured CDP.
- What is a composable CDP and how does it work?
- What is a full-featured CDP and how does it work?
- Full-featured or composable CDP: Which is better for customer experience?
- Can a full-featured CDP be composable?
What is a composable CDP and how does it work?
A composable CDP runs on a data warehouse and requires users to assemble the features they would like to include, much like building blocks. Components that the CDP does not offer (such as advanced identity management) can be supplied by other vendors. It is common for composable CDPs to have multiple vendors co-existing in the end solution, each performing different functions.
Composability is a concept in software design, and it’s widely used. For example, Wix helps customers build websites from different pieces – that’s composable. Data scientists rely on packaged libraries of functions that plug into Python and R — that’s composable.
When thinking about a composable CDP vs. a full-featured CDP, it’s important to keep in mind that composability is an architecture, not a product.
What does composability mean for CDPs? To gain the benefits of composability, you need to look for a CDP that’s built using an architectural approach that supports modularity. It needs to be flexible, with robust application programming interfaces (APIs), and a lot of options for configuration and use case support.
The advantage of a full-featured CDP like Salesforce Data Cloud built using composable principles is it gives customers a wider range of options for implementation and deployment. I know customers of Salesforce Data Cloud who do their own identity management in a homegrown solution, relying on the CDP for segmentation and activation.
The same kind of modularity helps customers who want to do their own segmentation and audience-building, say, rather than using the CDP’s built-in tools.
What is a full-featured CDP and how does it work?
A full-featured CDP is a tool designed to organize customer data from different sources and provide an up-to-date, unified view of each customer. Unlike composable CDPs, full-featured CDPs contain all the CDP capabilities in a single product, including data ingestion, modeling, identity management and segmentation.
Full-featured CDPs arose to fill a need in the market: combining disparate data to create a connected customer experience across marketing, service, sales, commerce, and beyond. They also provide a great foundation for emerging generative AI applications.
That’s why there’s so much investment and innovation in the full-featured (also known as packaged) CDP sector. According to the analyst firm IDC, the packaged CDP market is expected to surpass $5.7 billion by 2026, growing about 18% per year.
As is natural in a fast-moving sector, misconceptions arise. I am happy to report they are nothing to worry about:
- Full-featured CDPs are not “legacy” systems
- They do not create another customer data silo
Full-featured CDPs are far from being last generation’s technology. On the contrary, full-featured CDPs like Salesforce Data Cloud are built using cloud architectures that avoid the challenges of legacy databases, both in speed and data requirements.
And saying that CDPs create yet another silo is kind of like saying the Google search engine creates just another website. Rather, full-featured CDPs take siloed data and make it available to the business. It’s the key to your customer data, not the lock. And some full-featured CDPs such as Salesforce Data Cloud provide access to data in data warehouses like Snowflake without copying.
“QUOTE TBDThe full-featured CDP “enables companies to smoothly unite their data, driving better customer insights and experiences. It’s shaping the future of data and AI-driven success,” said Rahul Auradkar, executive vice president and general manager of unified data services & Einstein at Salesforce.
Full-featured or composable CDP: Which is better for customer experience?
Full-featured CDPs, unlike composable CDPs, have certain capabilities that make them foundational for a more connected customer experience. For example, when a major international racing organization needed to develop profiles of their 500 million fans around the world and connect with customers across channels like mobile, the web and advertising, they needed all the features of Salesforce Data Cloud.
Our customers tell us that CDPs – at minimum – need to be able to do at least five things:
- Access data easily from sources like websites and warehouses
- Harmonize the data so it’s consistent
- Handle identity management
- Create and explore segments and audiences
- Activate audiences to channels like email, advertising, call centers, etc.
Which brings us to a challenge of composability: it’s not a set of features. So I always recommend that any technology buyer be absolutely sure that any product calling itself a composable CDP can do all the things that a good CDP does. That’s just common sense.
Companies that decide to go the do-it-yourself composable CDP route often find themselves taking on more technical overhead than they anticipated. They may have to implement and manage multiple tools from multiple vendors, without a common user interface or service-level agreements. The requirements for ongoing maintenance can be significant.
Another characteristic of full-featured CDPs is that they are generally designed to be business user-friendly, using clicks, not code. This distinguishes them from legacy tools and some warehouses that require a more technical user, raising talent and training costs.
Can a full-featured CDP be composable?
Luckily, a requirement of full-featured CDPs is not that you have to settle for a partial solution. It is possible to work with a CDP that’s also composable in principle — that combines a full CDP feature set with the added flexibility of a composable architecture.
One way to think about the different types of CDPs is to compare their different approaches to standard CDP capabilities, such as data ingestion and analytics. In general, cloud warehouses and vendors of so-called reverse-ETLs target a more technical, do-it-yourself user.
Salesforce Data Cloud | Cloud Warehouse | |
Access to CRM data | Easy point-and-click direct access to objects in CRM | Requires connector |
CRM Access to CDP data | Direct zero-copy access to data in CDP from CRM | Requires connector |
Access to Data Warehouse | Direct zero-copy access or file transfer | Usually included |
Ingest from other apps | AppExchange marketplace, APIs, transfer | Marketplace, connectors or custom |
Identity resolution | Advanced probabilistic capabilities built-in | Basic features or via partners |
Identity graph | Unify any customer or entity data (e.g., events, model scores, attributes) | Managed in Data Warehouse |
Schema used | Flexible, fully-customizable schema | None |
Storage | Hyperforce multi-substrate cloud | Multi-cloud |
Analytics | Point-and-click, Einstein & BYOM* | SQL based |
Data activation | AppExchange, direct to destination or via transfer | From Data Warehouse |
Cost | Consumption based; pay for what you use | Pay for features |
Compliance | GDPR, CCPA and HIPAA compliant | May require BAA for HIPAA compliance |
User personas | Designed for personas from the business user (no code) to developers (pro code) | Targets technical users (pro code) |
Of course, there’s more to composability than just the CDP. Profile data should sit within the context of an enterprise architecture that enables you to build a more connected customer experience across your entire enterprise — from advertising through conversion, loyalty, service, win-back and more.
This more ambitious approach requires real flexibility at the platform level.
For example, Salesforce Data Cloud is built on the Salesforce core platform, which is metadata-driven and highly configurable. And it all sits on top of Hyperforce, our multi-substrate cloud infrastructure that provides trust, scale and governance capabilities across the enterprise (not available in all geographies).
Win-Win for the CDP Buyer
Composability is a great principle and we embrace it in our technical designs.
So it makes a lot of sense to ensure any full-featured CDP you’re considering has the flexibility and modularity advantages that you might find in a composable CDP.
It also makes sense to scrutinize any product – whether it’s called “composable” or not – to make sure it’s going to do what you need it to do. In the case of CDPs, your requirements likely include data access, identity management, analytical workflows, and activation.
Believe it or not, it is now possible to have your data layer and consume it too.