What is a Composable CDP?

Learn why Composable CDPs are seeing such rapid adoption, how they work, and why they’re replacing traditional CDPs.

CDPs, or Customer Data Platforms, have long been hailed as the all-in-one platform for marketing teams and have become essential tools for companies across industries to leverage their first-party customer data.

Regardless of your role, whether it be in sales, marketing, finance, or product management, you likely know that building and operationalizing a single view of your customer data is extremely difficult. CDPs work to fix this problem by offering non-technical business teams the capabilities needed to capture, model, and activate customer data quickly and at scale.

Although these all-in-one solutions quickly rose in popularity, the emergence of the cloud data warehouse has unlocked a far more flexible and performant CDP solution. With a Composable CDP, organizations are able to activate data directly from the cloud data warehouse, enabling them to take full advantage of their existing data investments while empowering marketing teams to maximize the value of their first-party data.

What is a Composable CDP?

A Composable CDP has the same goals and outcomes as a traditional CDP. However, the core difference lies in the underlying architecture. Rather than being bundled into a single product offering, a Composable CDP is built to leverage your existing data tools and technologies and activate data directly from the cloud data warehouse.

Composable CDPs address the exact same issues as their off-the-shelf traditional CDP counterparts but the architecture gives you far more extensive capabilities. The warehouse-native (composable) architecture enables you to select best-in-breed components for collection, modeling, and activation to uniquely address your specific use cases–all while providing better flexibility, security, governance, cost, metrics, and faster time-to-value.

Figure 1: Diagram showing how Hightouch can syncs data to downstream tools from your data warehouse.

How Does a Composable CDP Work?

While a composable architecture may seem more complicated at first glance, it’s surprisingly simple to understand when you look at it in the context of a modern data stack. You can break everything down into three pillars: data integration, data modeling/analysis, and business intelligence.

• Data integration is the process of moving data from the source to the destination.
• Data modeling/analysis is the process of transforming your data and building consumable data models for analytics.
• Business intelligence refers to visualizing your data models in a dashboard or report.

A Composable CDP adds a final layer to this architecture known as Data Activation. Data Activation uses a technology coined “Reverse ETL” to sync data to your downstream business tools–thus making it actionable by your business teams so they can tie their work directly to business-oriented outcomes.

Figure 2: Diagram of a Composable CDP architecture.

This approach differs from a traditional CDP because it establishes your data warehouse as the backbone for all your analytics and operational use cases, allowing you to easily leverage all of your first-party data and choose the best-in-class tooling for your architecture.

Rather than struggling to align your marketing team’s source of truth (the CDP) with your data team’s (the data warehouse), this best-in-breed architectural enables you to write to the data warehouse once and democratize that data anywhere across your organization.

A Quick History of CDPs

Traditional CDPs served as all-in-one platforms for data and marketing teams, with the goal of building an actionable, unified customer database. Data and marketing teams across industries have relied on CDPs to deliver personalized experiences for years. These platforms quickly rose in popularity because they uniquely solved a few complicated problems that had previously plagued marketing and data teams.

• Event collection: CDPs made it easy to collect clickstream data and capture web, mobile, and server events via proprietary software development kits or SDKs. This meant that you could capture the entire buyer’s journey from an ad impression and engagement, all the way down to a purchase or sign-up and categorize every action that happened along the way.
• Identity Resolution: Before CDPs, one of the biggest challenges was performing identity resolution and linking all of the actions of a user/customer into a single coherent profile. CDPs offer identity resolution to help resolve customer identities to this single actionable profile.
• Audience Management: One of the major draws for CDPs has always been their audience management capabilities. Having all your customer data in one place isn’t helpful unless you can model and build audiences for your downstream use cases.
• Activation: Audiences are only useful if you can activate them in your operational tools (e.g., Salesforce, Marketo, Iterable, Braze, etc.) Before CDPs, data engineers had to spend countless hours building custom integrations to sync data to various APIs. CDPs solved this problem by taking on all that underlying maintenance and creating the syncing capabilities needed to pass data to various third-party applications.

Figure 3: A digram showing how a CDP architecture.

The Evolution of CDPs

Previously, data warehouses were only accessible to large companies with dedicated data teams. However, the rapid rise of modern data warehouses like Snowflake, Databricks, and Bigquery have opened the door for any organization to spin up enterprise-grade data infrastructure to store, model, and analyze customer data with incredible flexibility and efficiency.

As organizations began to adopt these enterprise data warehouses, they quickly realized they now had two sources of truth: a complete 360-degree view of the customer in the data warehouse and a similar but incomplete view of the customer in their CDP.

This collision created a problem because marketing teams could only activate from a subset of customer data. In contrast, data teams were forced to try and maintain two sources of truth with two different sets of tools.

While CDPs have their benefits, they come equipped with a fundamental architectural flaw in that they create yet another source of truth–and an incomplete one at that. The only truly complete 360-degree view of your customer data only exists in your data warehouse because it houses all of your data sources–not just the clickstream events available in the CDP.

Benefits of a Composable CDP

While there are many advantages to this new Composable CDP architecture, they can fundamentally be bucketed into flexibility, scalability, and modularity. Each of these comes together to offer a far more powerful platform than that that can be achieved with any off-the-shelf CDP offering.

Flexibility

Traditional CDPs are built around rigid data models like users and accounts, creating a one-size fits all customer definition. In reality, customer data models aren’t so cookie-cutter. Many companies have custom objects like workspaces, playlists, carts, search history, tickets, etc., that need to be tied to customers.

Figure 4: The lack of flexibility for CDPs when it comes to data models.

Since traditional CDPs are designed solely to handle customer events and clickstream data, it can often be difficult to answer key questions:

• What is the average life-time-value for accounts with 250 employees?
• Who are my most active users?
• Which accounts have created a workspace in the last 30 days?
• Which accounts are at risk of churning?

Your data warehouse is the only platform that is powerful enough to answer these questions. Without the ability to model and query arbitrary and relational data from other sources, CDPs can only address basic marketing funnel use cases (e.g., shopping cart abandoners, suppression lists, etc.)

Single Source of Truth

Since most CDPs primarily focus on collecting event stream data, you can’t leverage additional data sources unless they’re built around an event-based model. CDPs claim to be a single source of truth, but they only give you access to a subset of your customer data (e.g., events and users), and they lock you into a single vendor, meaning they’re not future-proof and are painful to replace.

With a Composable CDP and warehouse-first approach, you have access to all your customer data, giving you more control and flexibility when transforming and modeling your data.

Figure 5: The difference between the type of data CPDs and the data warehouse stores.

Cost

It is no hidden fact that traditional, off-the-shelf CDPs are an expensive investment. With high contract costs, long implementation times, and seemingly endlessly growing customer record storage costs, organizations often struggle to see any significant ROI.

With a Composable CDP, you’re able to capitalize on your existing cloud data warehouse investments rather than paying for double storage of customer data across the CDP and the warehouse.

Data Ownership and Governance

Traditional CDPs operate as their own separate entity outside of your cloud infrastructure, so you’re subject to the whims and changes of your CDP in terms of how you can use your data. CDP vendors own your data and impose restrictions on how far back you can access historical data, which means you don’t have complete control over how it’s managed.

A Composable CDP approach gives you full ownership and governance over how your data is used, providing complete transparency and access control–thus ensuring you can easily comply with all security and regulatory requirements like GDPR, HIPAA, or CCPA.

Time-to-Value

When purchasing a basic CDP, you’re forced to evaluate everything upfront and locked into a single platform. In most scenarios, implementing a CDP will take you over six months and can extend to a year, and that’s not even factoring in the onboarding time it will take to train your data and marketing teams on how to use the new tool.

Taking a composable approach and adding an activation layer like Hightouch, which simply runs on top of your existing warehouse, means you can skip the implementation time and activate your customer data immediately because you can take advantage of your current technology stack.

Reliable Metrics

Running a CDP parallel to your warehouse creates an added level of complexity because your data teams have to model and transform your data in two places, which means you have no standard business logic.

Using your data warehouse as your CDP means your data teams can reuse all of your same models and code and only have to define your metrics once. Using your data warehouse as your CDP means your data teams can define your core metrics once using existing models and code, creating a “write once, use anywhere architecture.” With a Composable CDP, every time a business metric or model changes in your warehouse, you can reflect that exact change in your downstream operational systems.

Arguments Against a Composable CDP

Three common arguments arise when discussing a Composable CDP:

1. Data warehouses don’t support real-time use cases.
2. Data warehouses are too technical for marketers.
3. Data warehouses don’t support identity resolution.

These statements used to be accurate, but thanks to advances from Data Activation platforms and Reverse ETL, you can uniquely address each:

1. You can now fetch warehouse data with sub-30 millisecond latency for personalization use cases.
2. Your marketers can use visual audience builders on top of your warehouse to build their target segments without ever having to write a single line of SQL.
3. You can manage identity resolution in your warehouse entirely with SQL or leverage third-party tools that specialize in merging data sets.

Composable CDP Use Cases

While a Composable CDP addresses all of the conventional use cases of a traditional CDP, it also powers a much more comprehensive range of use cases with a much more scalable and flexible approach.

Custom Audiences

Audience building is probably the most prevalent use case for traditional CDPs. With a Composable CDP, you can go beyond traditional audience building and leverage all of your existing data models (e.g., shopping cart abandoners, LTV, recent purchasers, etc.) to build rich audience cohorts.

Using a Data Activation tool like Hightouch on top of your existing data warehouse, you can easily sync that data directly to your frontline marketing tools to optimize your marketing campaigns and increase your return on ad spend (ROAS.)

Object Syncing

Object syncing is the process of syncing data from one table to another. Common use cases include sending product usage data (e.g., last login date, active users, new signups, etc.) from your warehouse tables to business applications like Salesforce, Marketo, Hubspot, Braze, or Iterable. With a Composable CDP, you’re not limited to specific objects like users and events and have the full flexibility to sync any and all of your data directly to any of your end tools.

Event Syncing

Event syncing refers to syncing event data to destinations. For example, if you want to help your marketers personalize a campaign, they need access to crucial event data like (pages viewed, purchases, orders, shopping carts, etc.,) so you can build personalized experiences in near real-time.

Real-Time Data Syncing

A Composable CDP is also much more suited for real-time use cases because you can now pull data directly from your warehouse Hightouch’s Personalization API, meaning your business teams can build customer experiences in real-time.

How to Build a Composable CDP

The easiest way to turn your warehouse into a CDP is simply to leverage a Data Activation tool like Hightouch. Hightouch queries directly against your warehouse and syncs data to your chosen destination–all without ever storing your data.

Best of all, there’s a visual audience builder for your non-technical users who don’t know how to write SQL. The first integration with Hightouch is free, so you can set up your own Composable CDP after creating a new workspace.