W H I T E PA P E R

The Complete
Guide to the
Composable CDP
 The Complete Guide to the Composable CDP 2

Contents

04
A Quick History of CDPs

06
Problems with Traditional CDPs

08
The Rise of the Composable CDP

10
Benefits of a Composable CDP

11
Composable CDP Proof Points

12 Getting Started with a Composable CDP

 The Complete Guide to the Composable CDP 3

Preface
Organizations have more data than ever before at their disposal. Still, even with all this information,
many marketing teams find it challenging to drive outcomes that move the needle forward, while
many data teams struggle to tie their work to revenue-facing functions.

For years, marketing teams have relied on traditional Customer Data Platforms (CDPs) to power
personalization, targeting, and other customer-facing marketing efforts. However, with the rapid
adoption of cloud data warehouses, any organization, regardless of industry or size, can now store
near-infinite amounts of customer data centrally in their data warehouse with ease.

Rather than creating another data store in the CDP, leading organizations are now adopting a
Composable CDP that sits directly on top of their existing data warehouse. This new architecture
provides unparalleled flexibility, time to value, and security and is radically changing how
organizations across verticals manage their customer data at scale.

We have petabytes of data here at Bol.com. Moving this data out of our
data warehouse and pushing it to a traditional CDP would have been
completely illogical.

Rick Assiest

Group Product Manager Advertising

We didn’t choose an off-the-shelf CDP like Segment because a substantial

amount of engineering is required to get it up and running, and at the end of
the day, it is just a second source of truth. Why would we buy a CDP when all
of our data modeling is already being done directly in Snowflake?

Ryan Newsome

Head of Data & Analytics

 The Complete Guide to the Composable CDP 4

A Quick History of CDPs

A CDP is an all-in-one platform designed to help companies collect, store and unify customer data in
a centralized database so that marketing and business teams can easily build audiences and sync
data to downstream operational tools.

Traditional CDPs operate as black boxes, offering little to no configurability when it comes to product
features because everything is tightly integrated and bundled into a single product offering. As such,
traditional CDPs have five core components:

Packaged CDP

CDPs come equipped with proprietary SDKs (software development

kits) that organizations can use to collect behavioral events.
Event Collection

CDPs offer fully managed storage to house customer data and user
profiles directly within the CDP.
Data Storage

CDPs have built-in identity resolution features to stitch together user

actions and attributes across touchpoints.
Identity Resolution

CDPs provide audience management tools to help marketers build

user cohorts and orchestrate journeys across marketing channels.
Audience Building

CDPs provide out-of-the-box integrations with third-party APIs,

enabling organizations to send audiences directly to operational tools.
Data Syncing
A Quick History of CDPs The Complete Guide to the Composable CDP 5

CDPs have been around for a long time, with Segment pioneering the space in 2011. Since then, data
and marketing teams in almost every industry have relied on these platforms to power their
customer-facing use cases.

These solutions emerged during a time when data collection, management, and activation were
extremely challenging. Prior to CDPs, there was a huge rift between data and marketing teams
because data tools weren’t built for marketers, and marketing tools weren’t built for data teams. 

Moving data between various applications and systems was an engineering nightmare as it required
a combination of custom scripts, API integrations, or manual CSV files. More often than not, 

the end result of these efforts was a spiderweb of pipelines powered by intricate workflows within
various tools.

Traditional CDPs resolved this complex web of pipelines by introducing a single platform that
automatically integrated with downstream applications. This meant data teams no longer had to
worry about managing data flows, and marketing teams no longer had to worry

about data accessibility. Data and engineering teams could implement event collection with the
CDP, and marketing teams could then easily use it to self-serve and build audiences to power their
downstream digital channels.

Traditional CDP Architecture

Marketing Automation

Storage
Web & Mobile Events Advertising Platforms

CDP
Server Events Analytics

Two sources of truth?

ETL

Data Warehouse

Storage
 The Complete Guide to the Composable CDP 6

Problems with Traditional CDPs

Although CDPs set out to solve the correct mission, the platforms' tightly coupled and siloed
architecture introduces a number of complexities that greatly limit their effectiveness for modern
organizations.

Storage and Data Ownership

Traditional CDPs create a second source of truth for customer data. They force companies to store
and manage data within the CDP - outside of and decoupled from any existing data infrastructure.

Flexibility

CDPs lock customers into an opinionated and rigid data model. This makes it nearly impossible to
represent relationships between users and other entities like accounts, teams, workspaces, playlists,
or devices.

Time-to-Value

The average CDP implementation takes over six months to complete, and that’s not even accounting
for onboarding time. Additionally, trying to support new use cases or adding new data sources
introduces even more complexity because data teams are forced to build new data pipelines and
rearchitect the CDP any time a new attribute needs to be stored.

Incomplete Data

Since CDPs only collect and store behavioral data, they create a fragmented view of the customer.
Marketers struggle to access offline data or the custom data models, e.g., propensity scores, the data
team has built. Some CDPs have tacked on features to interface directly with data warehouses, but
these features are often lacking and separate from the core product offerings.

Cost

Traditional CDPs are extremely expensive because every feature is bundled together. This means
companies have to pay for collection, storage, and modeling, even if they’ve already solved for these
functions internally. Inevitably, many companies end up paying to store the same data twice (once in
a warehouse and once in a CDP.)
Problems with Traditional CDPs The Complete Guide to the Composable CDP 7

Ultimately traditional CDPs come in a “one-size-fits-all” model that makes them very rigid for hyper-
specific use cases. For example

With a traditional CDP, it’s possible to send an email to cart abandoners, but it’s almost impossible
to send that email with a local store incentive since store data is not available in a traditional CDP

With a traditional CDP, it’s possible to show ads to people who have visited product pages, but it’s
almost impossible to ensure those same ads aren’t shown when a product is out of stock without
another integration in place as catalog data is not natively available in a traditional CDP

With a traditional CDP, it’s possible to send an SMS about a new promotion, but it’s almost
impossible to test that same SMS on only the people who have a high propensity of redeeming it. A
CDP does not easily connect to all of your recommendation engines or data science models.

Each of these marketing use cases points out a fundamental problem: traditional CDPs have no
understanding of custom objects (e.g., local stores, products, propensity models) as they often boil
everything down to users, and events.

Traditional CDP

What about the rest?

Behavioral data Offline actions Data science Related Objects

App events Phone support Machine learning Households

Web Events Point of sales Experiment cohort Accounts

Support call Predicted

Session started Company name

records lifetime value

In-store
Purchase
Page viewed Company type
purchases propensity

Purchased Company size

The reality is businesses are much more complicated than behavioral data, and modern marketers
should have access to a true 360-degree view of the customer. That includes everything from
behavioral data to offline actions, to data science models and related objects. This level of granularity
is only possible with a Composable CDP powered by the data warehouse.
 The Complete Guide to the Composable CDP 8

The Rise of the Composable CDP

A Composable CDP has the same goal as a traditional CDP: activating unified customer
data to downstream tools. But rather than operating as a separate entity and storing data
outside of your current data infrastructure like a traditional CDP, a Composable CDP lets
you create audiences, personalize journeys, and send your existing customer data
directly from your existing data warehouse to your many marketing tools.

While CDPs set out to solve the correct problem, they did so with the fundamental flaw: they did not
build upon an immensely powerful technology rapidly being adopted by organizations as their
central store of customer data - the cloud data warehouse. 

The rapid adoption and investment in cloud data warehouses has caused many companies to realize
that they don’t need to purchase a separate platform to store data they already own in their data
warehouse. Instead, by simply integrating Hightouch with their existing data warehouse,
organizations are able to form what has since been coined the “Composable CDP” by cloud leaders
like Snowflake and Databricks.

This new architecture provides unparalleled flexibility, letting organizations tackle complex
challenges efficiently while equipping marketers with the tools they need to explore, analyze and act
on all of their customer data to create more personalized user experiences across channels.

Identity Resolution

Composable
CDP Architecture

Marketing Automation
A Composable CDP is a Web & Mobile Events

modular solution that

Server Events Advertising Platforms
collects, models and
activates customer data
ETL Analytics Tools
directly from your own
data infrastructure Data Warehouse

Storage
The Rise of the Composable CDP The Complete Guide to the Composable CDP 9

While many CDPs now claim composability by integrating with the data warehouse, a Composable
CDP must meet the four following requirements.

1 2 3 4

A Composable CDP A Composable CDP A Composable CDP A Composable CDP

runs on your own is schema-agnostic.
is modular and provides unbundled
infrastructure.
You have complete interoperable.
pricing.

The platform must control over how The platform must You should only pay
not store a separate your data is integrate with your for the features and
copy of your structured and existing technology capabilities you
customer data. stored. without locking you need.
into a single vendor.

The modularity of this architecture gives companies far more extensive capabilities and flexibility to
solve their most complex customer-facing use cases. This means organizations don’t have to
conform data to the requirements and constraints of another platform and can instead take
advantage of their existing data assets to drive value immediately.

The easiest way to understand this difference is to view them as two restaurant menu offerings:
Traditional CDPs follow a fixed menu option where the entire meal is offered at a set price.
Conversely, a Composable CDP follows an à la carte model with individual pricing where people can
choose items tailored to their specific preferences.

For example, Hightouch allows teams to select from a suite of warehouse-native, modular
components, including event collection, identity resolution, audience activation, and more. These
features combine with their cloud data warehouse to form a Complete, and Composable CDP.

Hightouch Unlocks a Complete and Composable CDP

PREPARE

Identity Resolution Enrichment

ACTIVATE

Reverse ETL

COLLECT
Audience Builder
Event Collection
Data Onboarding

Personalization API

your data warehouse

 The Complete Guide to the Composable CDP 10

Benefits of a Composable CDP

The benefits of a Composable CDP can be summed up into four primary categories: flexibility,
security, modularity and time-to-value.

Flexibility Security
A Composable CDP gives organizations Because a Composable CDP simply sits
immediate access to all of their data– on top of existing infrastructure,
not just clickstream events. Ultimately organizations don’t have to worry about
this means marketing teams can data leaving their cloud environment or
leverage any data type to easily duplicate stores of customer data.
accommodate their complex use cases.

Modularity Time-to-Value
Companies have full control to choose A Composable CDP enables
what technologies and processes they organizations to circumvent long
use for data collection, storage, implementation times and activate their
modeling, and activation. This allows existing data assets immediately to drive
them to tailor their architecture to the business outcomes.
specific organizational outcomes
they're looking to drive.
 The Complete Guide to the Composable CDP 11

Composable CDP Proof Points

This modular architecture works with any organization regardless of maturity, and businesses across

industries are seeing the immediate benefits of building on top of their data warehouse.

Within six weeks, we implemented Hightouch on Snowflake and got out

hundreds of audiences, which made marketers and agencies happy.

Warner Music Group Itay Rahat

Powers 1000+ custom

VP of Product, Audiences, and
audiences
Marketing Technology
Read full story

We evaluated CDPs but quickly realized that no solution would allow us

to activate our existing data in Snowflake. Hightouch bridges the gap

between data and marketing so we can access our first-party data and

use it to build audiences.

The Zebra boosts
advertising audience
Match Rates by 170%
Cynthia Caridad

Director of Lifecycle
Read full story

Hightouch lets us build really complex audiences in minutes, giving us

everything we need to run A/B testing, measure performance, and

optimize them so we can drive value immediately.

bol.com Replaces
3rd-Party Cookies to Rick Assies

Increase CTR by 33%

Group Product Manager Advertising
Read full story

Access to first-party data has made our data engineers have more
empathy for our marketers, and our marketers are now more willing to

provide insight into the campaigns they are running.

AutoTrader Builds a
Composable CDP
and Increases
Engagement for New Darren Haken

Car Buyers by 20% Head of Engineering

Read full story
 The Complete Guide to the Composable CDP 12

Getting Started with a Composable CDP

Most organizations think they need to reach a certain level of data maturity before they can fully take
advantage of a Composable CDP. In reality, this is a misconception. Hightouch offers the same
capabilities of a traditional CDP without the tightly coupled architecture and can introduce these
products one step at a time.

Hightouch Events
Customer 360 Toolkit

Collect and load events directly into your data Stitch your existing customer data into rich,
warehouse. Hightouch Events is an easy-to-use actionable profiles directly in your warehouse.
event collection product. It’s fully customizable Link all of your user actions and attributes to a
and purpose-built to centralize your data in the single comprehensive profile without writing any
warehouse.
of code.

Data Warehouse
CRM
Unified profile
hightouchevents.track(

“Order Completed”, {
 Web Identity
sku: “142A1”,
 Resolution
total: “79.99”,

}); Commerce Service
Buy
Database
Order Completed

Product Viewed

3rd party
Page Viewed

Customer Studio

Build and activate rich audience cohorts to Your avorite companies use
f

power your most complex personalization use Hightouch to activate data

cases. Perform A/B testing and cross-channel securel at scale
y .

orchestration and easily analyze audience

breakdowns and performance.

Define your audience

Filter down customers using anything in your underlying data.

...
Has property Lifetime Value greater than 100

OR Ungroup

Has custom trait Favorite Category Order Completed equals Shoes

...

Where...

OR Ungroup

+ Add filter

AND

Performed Order Completed at least 1 time

Within 6 months ago

+

...

Where Brand equals Nike, Adidas Schedule a demo Learn ore

M
+

Where event property is... Then performed...