10 Ways Customer Data
Platforms Differ From Data
Management Platforms
hey music loversle
Table of Contents
What’s a customer data platform again? ..................... 3
CDP vs. DMP: The bascis ............................................ 4
10 ways CDPs and DMPs are different ........................ 5

We get this question a lot (like, a LOT): How is a Customer Data Platform
like BlueConic different from a Data Management Platform?
This is a totally legit question. To provide a descriptive response requires a certain level of
sophistication when it comes to customer data management. There are a number of valuable
similarities between customer data platforms, like BlueConic, and data management platforms,
like BlueKai, Krux, and Lotame, to name a few. There are also a few key differences that affect
the utility, implementation, function, and value of each. So, how are CDPs different from DMPs?
This ebook will provide some clarity so that you know what you’re getting and why.
First, let’s revisit what a customer data platform is, courtesy of the one and only David Raab. Back
in 2013, Raab defined a CDP as “a marketer-controlled system
that builds a multi-source customer database and exposes it to
external execution systems.” In particular, Raab wrote that the
name works because:
1. Customer shows the scope extends to all customer-related
functions, not just marketing
2. Data shows the primary focus is on data, not execution
3. Platform shows it does more than data management while
supporting other systems.
3
4
“Anonymity is essential to the DMP’s role as a way to exchange information about
audiences without violating personal privacy”
On the surface this CDP concept is pretty straightforward, but it’s not all that hard to find yourself embarking
on an inadvertent thought exercise: which technologies from other categories of martech qualify in some
way as a CDP, blurring the lines that Raab has so convincingly provided? The blurriest convergence is
between CDPs and data management platforms (DMPs) because:
1. Both do customer data management,
2. Both help marketers do their jobs better, and
3. Both help provide richer insight about your consumers.
Raab further states, “The distinction has always been clear: CDPs work with both anonymous and known
individuals, storing “personally identifiable information” such as names, postal addresses, email addresses,
and phone numbers, while DMPs work with almost exclusively with anonymous entities such as cookies,
devices, and IP addresses. Indeed, anonymity is essential to the DMP’s role as a way to exchange information
about audiences without violating personal privacy. What’s changed is that CDPs are integrating more often
with advertising systems, and thus storing more DMP-type information such as cookie IDs with audience
tags. Some DMPs are also storing personal identifiers, although these are carefully isolated from situations
where anonymity is still important. But just adding personal identifiers doesn’t give a DMP the advanced
identity matching and flexible data storage built into CDPs. So it will be hard for most DMPs to match full
CDP functionality.” To learn more about Raab’s work, visit http://raabassociatesinc.com/.
Thanks are also due to Marty Kihn for his original post on http://blogs.gartner.com/ discussing the 10 secrets
of DMPs which we will now use to unfold the key differences between CDPs and DMPs.
Despite those high level similarities, here are the key differences:
1. DMPs store data in two different ways.
CDPs don’t. Whereas the DMP has two different data stores – one for #allthedata and one for really fast
utilization of a subset of that data, the typical CDP database is all one, so there isn’t a subset of the data
that lives separately. The key value store and the total insight lives in a single, massive place (which is an
Apache Cassandra cluster, to be specific in BlueConic’s case). The database is massive, and can scale,
but it also features lightning fast reading capabilities.
2. DMPs collect data like everyone else – with tags, APIs, and uploads.
For CDPs, it’s the same mechanism but richer data. Any marketing technology worth its salt today
had better be able to bring in data in the formats it’s most commonly captured. So while a CDP does
use those three means of collecting data, it’s not just about how to collect data; it’s arguably more so
about the depth and extent of data brought into the database. In BlueConic’s case, this means that the
activity tags capture far more than just clicks and swipes and the associated URLs. With a universal tag
approach (no custom JavaScript), BlueConic tags enable form input collection, whether or not the forms
are submitted. Text from the content of pages viewed and values from cookies and JavaScript objects
are all fair game to be stored in user profiles in addition to URL components. When loading data via
APIs and uploads, marketers can overwrite user attributes, or add values to them such that BlueConic
acts as a lifecycle system of record, maintaining all of the values that have been stored for the user in
addition to the most recent.
5
6
3. DMPs are in the business of labeling (and relabeling) people.
But when and how are pretty distinct. Kihn summarized this section by writing,
“the data provider must organize its data into categories and subcategories. So all
the data sent to the DMP is basically a list of users and an associated list of which
predefined categories/subcategories they belong to.” But we think you should
be able to bring enormous lists of individual users and their attributes into the
platform without predefined categories. Ideally, you will create a taxonomy with
segment definitions you can create before or after loading the data, and can help
you discover new ways of classifying segments (see image). Individuals move in
and out of segments in real-time based on their behaviors or changes in status.
4. DMPs use outside partners to help map data to users.
Sure – whatever the customer wants! The cleaning, assembly, and organizing
that data onboarders like Acxiom LiveRamp do for customers is extensive, and a
CDP both enriches that data set and makes that data actionable. In some cases,
BlueConic gives marketers insights about their consumers that they
had previously relied on an outside partner for. Just as often, a CDP
ingests data from the onboarder and put it to use by powering more
relevant engagement.
7
5. A DMP “user profile” is not supposed to be a complete customer profile.
A CDP’s profiling mechanism enables this without needing to be it. At BlueConic, we’ve
got a fairly unconventional (dare we say unpopular?) point of view about “single” and
“360º” views of customers which is that they are both unrealistic and impractical. For
example, one marketer we spoke to told us that his company had twenty-seven (27!!)
separate instances of Salesforce CRM alone. That doesn’t include any other systems
that, of course, also have customer data flowing in and out of them. The whole point
of this holy grail, this single customer profile, is so that you’d have all the information
about a person – historical, contextual, demographic, behavioral, etc. – on hand to
inform the nature of a brand interaction with that person. The issue is not that there
isn’t a system that can be your end-all, be-all customer data repository, but rather that
there are so many options that they get in the way of getting the data where it needs
to be, when it needs to be there – when the brand is about to initiate an interaction
proactively or in response to a consumer’s action. So a
DMP can’t be and a CDP shouldn’t aspire to be.
8
6. What DMPs call an “audience” is just a segment.
This is a legacy of the DMP’s original and primary purpose (which is discussed in
bullet 7); advertising, or dealing with new and unknown customers. Segments are
typically more associated with direct marketing and ipso facto groups of known
customers. The benefit of being able to create both audiences and segments, as
CDPs can, is the ability to syndicate (to use Kihn’s word) the data to any kind of
partner on either side of the known/anonymous equation – adtech or martech.
7. DMPs were designed to build targets for advertising.
To be a truly valuable part of the marketing tech stack, a CDP needs to be
able to take any data and get it wherever it needs to get – regardless of where
in the customer life cycle (or marketing org) the other systems live. That can
mean passing segments of rich first party data on to advertising partners like
DoubleClick or Facebook for look-alike modeling or more effective retargeting, as
well as any other stop in the customer journey.
9
8. DMPs were also designed to personalize websites.
+mobile apps, +campaigns, +emails and manage the delivery.
Kihn is absolutely right that it seems like today “everyone and their dance
partner does personalization, including landing page optimizers, testing
platforms, content management systems, recommendations engines and so on.”
Nonetheless, despite this apparent abundance of options, very few marketers are
actually doing personalization well and certainly not across channels. It’s critical
to put the tools in the marketers’ hands to bring data and capabilities together
that will allow a linked, consistent brand experience as the customer moves
through their decision journey. The wealth of traits that a CDP collects and the
persistence of the profile for that individual across channels and sessions takes
“personalization” to a whole new level.
9. DMPs make many decisions based on predefined rules.
That’s part of it. CDPs deal in rules and triggers too – “if this,
then that” statements, combinations of criteria, A/B tested
decisions. But where there is a leap from the DMP’s capabilities
is in the fact that actions taken by a user are the only trigger,
whereas a user’s identity or other non-behavioral attribute can
result in an immediate decision or action. Furthermore, a CDP is
happy to share the rules and triggers that are set up with other
systems – no black box here.
10
10. DMPs are better at counting than analytics.
The key takeaway of this, most provocative statement in Kihn’s post is the
following: “The purpose of the DMP is to label people,” and our own addendum
to that, “for the primary purpose of advertising.” In contrast, the role of a CDP is
not to create an additional or separate source of truth about customers. It is to
facilitate the synchronization of customer data so that at any given touchpoint or
interaction, the marketer can recognize that person and engage with the fullest
extent of insight about that individual possible. By bridging previous barriers
between known/unknown data management, leveraging both contextual and
historical information, and remaining an agnostic and flexible tributary for data
in and among the marketing tech stack, a CDP fulfills the idealized ambition of a
single customer view with an entirely practical and marketer-driven solution.
One caveat to this comparison is that on some technical
dimensions, like data storage mechanism, some companies
that are considered CDPs don’t fulfill the criteria the same
way we describe here which is based on BlueConic, so not
everything we compared and contrasted will apply to every
vendor in the category.
Boston, USA (Headquarters)
207 South Street, Suite 671
Boston, MA 02111
T: +1 (888) 440-2583
Nijmegen, NL (Europe)
Wijchenseweg 101, 6538 SW,
Nijmegen, The Netherlands
T: +31 24 205 1000