2. 2
McKinsey & Company
A unified technology platform that
takes in disparate first-, second-,
and third-party data sets, provides
normalization and segmentation on
that data, and allows a user to
push the resulting segmentation
into live interactive channel
environments
A data management
platform (DMP) is…
SOURCE: Forrester
3. 3
McKinsey & Company
A DMP collects, segments, and activates customer and prospect data across channels
SOURCE: Adobe Experience Cloud, internal analysis
DMP process flow
1 DSP = Demand Side Platform (Same as a stock broker - aggregating demand from various advertisers and feeding the requests to the exchange)
Profile consolidation Activation
Audience creation
Data collection
DMP
3rd party
data
3rd party
data
CRM Website
1st party data
Social
1st party data
1st party data
Analytics
▪ Ingests multiple
data sources:
– Collects website
traffic
– Collects paid
advertisement
engagements
(e.g., display,
search)
– Ingests first-,
second-, and
third-party data
▪ Identifies and
consolidates data
from multiple
sources into
unique profiles
(users or prospects)
▪ Matches devices into
unique profiles
▪ Generates a
complete view of
each target
▪ Creates accurate
and comprehensive
profiles based on
online and offline
insights
▪ Enables customized
segmentation based
on marketer needs
▪ Enables DSPs1 to
optimize ad buys
by leveraging
customer segment
information and
insights
▪ Helps enable
retargeting in
digital ad
campaigns
▪ Collects results of
DSP activation
(e.g., display clicks,
landing page
arrivals)
▪ Reports results for
custom audiences
(through
dashboards), helping
marketers identify
opportunities for
improvements
4. 4
McKinsey & Company
A DMP enables or improves a number of important digital marketing use cases… NOT EXHAUSTIVE
There are additional use-case variations and combinations that
help address specific marketing and business priorities
What it is Example
Use case
Targeting
▪ Show this ad only to people who
visited my website in the last x weeks
and live in the Boston area
▪ Target display advertising to a
very specific custom audience
Prospecting
▪ Target ads to people who have the
same characteristics as my best
customers
▪ Find new prospects on the
internet
Suppression
▪ Exclude people who just became
customers in the last x weeks from
our ad campaign
▪ Exclude certain audiences
from seeing my ads
Personalization
▪ Show our email marketing recipients
the same ads online as were included
in the email
▪ Provide the same message to
people across channels
5. 5
McKinsey & Company
…and impacts across the customer decision journey to drive marketing effectiveness
Use cases across the customer journey
1 Anonymous behavior data could either be directly collected by the DMP or purchased from 3rd parties
Reach more
targeted prospects
Target existing customers
across channels
Personalize landing
pages
Retarget interested prospects
Develop rich profiles
of converters and
high value customers
Personalize relevant
digital experiences
Reduce churn by
targeting potential
churners
Reach prospects as they travel across
the decision journey
Advocate
Buy
Experience
Consider
Evaluate
Bond
6. 6
McKinsey & Company
A high performing DMP has a set of distinct features (1/2)
Feature
Cross Device
Identification
Experience
Optimization
Online & Offline
Data Ingestion
Omni Channel
Integration
Real Time Data
Activation
Description Example application (in telco)
▪ An prospect who abandons his cart, after shopping for
an iPhone 7 on his work computer, is retargeted on his
mobile phone later that evening, where he completes his
purchase
▪ Enables people-based marketing by matching devices to
unique visitors, therefore powering seamless cross-device
targeting, increasing reach, and improving campaign
performance
▪ Exposes users to content that has performed the best with
other similar users (supports look-alike modelling)
▪ Company X modifies the messages and layout of its
website to better match the characteristics of the visiting
user based on what previous campaigns have shown to
be successful for individuals with similar traits
▪ Provides a view of the entire customer journey by syncing
offline data (e.g., CRM, consumer insight, sales data) with
online data
▪ A day after expressing interest in Product A to a
customer service representative (who then updated her
record in the CRM system), Jane receives an email
featuring Product A
▪ Provides a 360-degree view of customers and prospects
▪ Deploys audience data to multiple channels (e.g., display,
mobile, offline, email, video, social)
▪ Ensures that customers receive cohesive and relevant
messaging across touchpoints
▪ An customer who has unsubscribed from an email list
(an indicator of high churning risk) is re-engaged with
segment-specific banner ads presenting an attractive
offer to help drive retention
▪ Collects data through tags, api integrations, or on
boarding, and makes it available for audience creation or
activation as soon as it is ingested
▪ Allows for increased media efficiency and targeting based
on real time marketing signals
▪ A new customer who just signed up for a new unlimited
plan no longer sees unlimited plan advertisements when
visiting various websites
7. 7
McKinsey & Company
A high performing DMP has a set of distinct features (2/2)
Feature Description Example application (in telco)
Universal ID
Implementation
▪ As the DMP tags are implemented across paid and owned
channels, they apply a consistent ID that allows
aggregation of data across these channels
▪ DMP provides service (or partners with a service provider)
to aggregate device IDs into person and household IDs
▪ Statistical analysis allows daily scoring of propensity to
churn, call for support, upgrade phone, add a package,
or add a line. Data is activated through the DMP to send
targeted messaging to the same person across devices
(e.g. phone, home, work computers)
Audience
Extension
▪ The DMP marketplace gives access to hundreds of 3rd
party data partnerships that can provide specialized data
and/ or audiences
▪ Audiences can be tested for match rate before purchase
▪ After an advanced segment analysis reveals a higher
propensity of Asian males to purchase Product A, a
centralized DMP business team establishes new
partnerships with Asian media publishers for audience
data
Universal DMP
Utilization
▪ All campaigns, down the marketing funnel and across
marketing channels, are activated through the DMP to
reach well defined audiences, ensuring that the customer
experience is consistent, and that sequential messaging is
tailored to reinforce the customer journey
▪ A newly acquired audience of prospects is served
Company X upper funnel messaging for some weeks
before the same audience is pushed down the funnel
with feature and offer campaigns
Advanced
Segmentation
▪ Continuous development and refinement of customer data
segments for a better understanding of users and
attributes impacting engagement, conversion, churn, etc.
▪ The customer analytics team provides a loyalty score for
all customers. Through online audience analysis, the
attributes correlating with loyalty are determined, and the
learnings are used to improve segments
8. 8
McKinsey & Company
Case example: how a DMP functions in the case of a telco client
Saw a family
plan upgrade
ad 2 days
ago
Visited the
company
website today
and looked into
unlimited data
plans
Upgraded to
a new
iPhone last
month in-
store
Saw a promo
offer on
“Dataplan” when
reading the
news on ESPN
Searched “best
data plan for
family” on Google
yesterday
Owns house in
New York
DSPs receives
command and
make targeted
media buys
accordingly
Email with
“$25 off loyal
customer
special offer”
to
DMP collects data
e.g., people who
visited a section of
your website
Marketer uploads
1st party data
e.g., email list
Marketer
purchases data
e.g. people
interested in
sports
Gathers data Builds user profiles
Creates custom
audiences Transfers data1
Send personal-
ized messages
ID#
123456
Customers who
have clicked on a
“$20 off Dataplan”
offer but did not
upgrade their plan
A
Prospects who
have researched
data plans online
this week and
subscribe to Sports
Illustrated
B
Customers who
have a family
wireless plan,
clicked on the
“Dataplan”
Facebook ad today,
and live in the
North East
C
A
“Dataplan”
Display ad
with a Sport
creative
to B
Video ad on
“Dataplan
connection
with family”
to C
1 DMP can feed an email engine but it requires a setup with the right platform (involving ID swapping and hashing), as the DMP does not hold email addresses
9. 9
McKinsey & Company
DMPs provide customer privacy protection and data security
How it works
▪ DMP’s do not store personally identifiable information – no names, phone numbers, addresses, emails, social
security numbers or the like
▪ Customers are assigned an anonymous DMP cookie ID when their data is collected and analyzed by the DMP
– Example: DMP only knows that ID#12345 has visited website X but does not know that ID#12345 is John Smith
▪ Customers are targeted at a small group level instead of at an individual level
– Example: Marketer ABC wants to target their ad at people who visited product page X. Once the DMP group of
‘people who visited product page X’ grows to a minimum number (such as 50 or 100), marketers can use that
criteria to target (e.g., show ad to people who visited product page X)
– Example: when John Smith visits a website that has an advertisement from Marketer ABC, the ad-server reads
John’s anonymous DMP cookie ID and quickly checks if user ID# 12345 is part of the group that visited page X
to determine whether or not to show the ad
Customer
privacy
protection
▪ DMPs are cloud-based systems, often hosted in public clouds such as Amazon, and to operate in a safe
manner there, all kinds of security measures are taken – including encryption and secured connections
▪ There is a wide variety of data privacy and security regulations that apply to DMPs, such as the Federal
Communications Act in the USA, the Personal Information Protection and Electronic Documents Act in Canada, the
European Union's Data Protection Directive and many others. DMPs must comply with these regulations in the
countries that they operate in – or they would not be in existence there for very long
▪ DMPs provide privacy and security information on their websites:
– BlueKai example: www.oracle.com/legal/privacy/marketing-cloud-data-cloud-privacy-policy.html
– Krux example: www.krux.com/privacy/
▪ You can check what DMPs know about you (as an anonymous ID) on sites like www.bluekai.com/registry/ and
http://youradchoices.com
Data security
11. 11
McKinsey & Company
DMP is a core component of the martech stack that enables data to be used effectively for targeting
Data management Campaign execution Analytics and performance Focus of this document
Functionality description
Tech stack elements
Enable real-time access / interconnectivity among data
systems to create a 360° customer view
Marketing data platform (MDP)
1
Leverage customer & prospect data to create discrete,
targetable customer segments
Data management platform (DMP)
2
Decisioning engine to personalize marketing content across
channels
Content personalization engine
3
Create and manage dynamic, personalized, triggered
campaigns to customers
Campaign management platform (CMP)
4
Create, manage, and deliver content across all channels and
touch points (e.g., web, mobile, campaign, social, paid media)
Content/Asset management platform
5
Deliver, track, and optimize digital media (e.g., search bid
manager, ad server, affiliate platform)
Ad serving/bid management platforms
6
Manage and optimize programmatic media buying
Demand-side platform (DSP)
7
Manage all tags with a Tag Container to reduce IT
dependency
Tag management System
8
Test and iterate website elements to drive improved
conversion rates
A/B testing platform
9
Analyze web and mobile traffic, content engagement
patterns, and conversion
Web/Mobile analytics
10
Analyze digital marketing channels to optimize spend & MROI
Attribution modeling tools
11
Customer
data platform
2nd/3rd party data
DMP
data collection
and analysis in-
the-cloud
1st party data
Demand-side
Platform (DSP)
DMP allows marketers to use comprehensive
customer data to connect ad audience with the
optimal marketing messages without
compromising customer privacy
12. 12
McKinsey & Company
A strong martech stack enables leading digital marketers to reach excellence across
a range of areas …
What leading digital marketers do
What more sophisticated
companies do
What most companies do
Marketing
technology and
data
▪ Clear vision and strategy for marketing
technology and infrastructure
▪ All data harvested for insights and
optimization
▪ Customer database managed in-
house
▪ Aggregated campaign reporting
▪ Data and technology managed by
agency
▪ Vendor contracts owned by agency
MROI
investment
optimization
▪ Fully developed digital marketing datamart
▪ Advanced multi-touch and cross channel
attribution
▪ Digital channel optimization
▪ MMM top-down media planning
▪ Reliance on last click attributions
▪ 60-day post campaign reporting
from agency
Evolve
operating
model
▪ Integrated segment programs
▪ Extensive journey management
▪ Rigorous test and learn
▪ Digital “always on” component
▪ Annual planning calendar
▪ Siloed groups (e.g., brand building vs
acquisition, direct marketing vs. E-
commerce)
Drive customer
value
▪ Real-time offers with customer context
▪ Targeted cross-sell programs
▪ Limited pool of offers for existing
customers
Drive customer
acquisition
▪ Personalized campaigns for specific high-
value segments
▪ Integrated campaigns with consistent
look and feel
▪ Campaigns for specific channels
13. 13
McKinsey & Company
… resulting in significant value as the right digital marketing infrastructure and strategy
are implemented
20%
in sales productivity
through personalized
targeting
25%
Reduction in attrition rates through
a multivariate predictive model
160%
in cross-selling success
after life stage segmentation
of customers
8-10
NPS points for large bank
through optimization of
digital campaigns
35%
Expected increase in annual revenue through
digital marketing revamp of organization
200%
in online conversions after 3
months of SEM optimization
66%
cost per customer acquisition
through SEM optimization
B
SOURCE: Impact values originated from past client engagements
14. 14
McKinsey & Company
DMP helps drive value across three key benefit levers
Benefit levers Key drivers of value
Increase media
spend efficiency
▪ Focus buys only on the most effective targets, reducing wasteful spend
▪ Identify the channels and placements that drive engagement
▪ Organize and track spend efficiency and effectiveness by segment
Increase
customer
acquisition
▪ Leverage 1st party data and behavioral segmentation to created addressable profiles of high value
segments. Develop look-alike audiences for prospecting
▪ Improve targeting of discrete audiences and high value prospects across channels/
devices at scale
▪ Effectively retarget website abandoners based on unique id and targeting across platforms
Improve
conversion and
cross sell via
personalization
▪ Customize the customer experience for different customer segments (e.g., personalize landing
pages)
▪ Enable consistent experiences and offers across channels
16. 16
McKinsey & Company
DMPs fall into three categories: Hybrid, Stand Alone, and Integrated Marketing Platform LIST OF VENDORS
NOT COMPREHENSIVE
Hybrid –
Demand Side
Platforms (DSP)
which also offer a
Data Management
Platform (DMP)
capabilities
1
2
Stand Alone –
independent data
management solution
which is not tied to
any particular DSP
3
Integrated Marketing
Platform –
DMP as a component
on a broader digital
marketing cloud
Integrated Marketing Cloud
Stand Alone
Media extender Personalization enabler
▪ Quick start
▪ Limited Implementation
▪ In-flexible taxonomies
▪ Limited tools
▪ Limited transparency
(Blackbox)
▪ Integrates with a wide variety
of 3rd party data sources
▪ Connects to a broad
selection of activation
platforms (e.g., all major
DSPs)
▪ Integrates more seamlessly in the
marketing stacks
▪ Large scale of data partnerships
▪ High audience reach
▪ High cookie matching rates
▪ Multi-channel personalization e.g.,
web, mobile, call center, retail
▪ Cross platform insights
▪ Leverages CRM, CLV
▪ Rules-based controls
▪ Advanced reporting
Adobe
Audience
Manager
Hybrid (DMP/DSP)
Reason to choose:
Simplicity with no
integration needed
Reason to choose:
Connectivity with a broad
set of tools, data sources,
and activation platforms
Reason to choose: Best
integration with the rest of the
vendor’s solutions
17. 17
McKinsey & Company
Key strengths of some of the leading vendors in the market include the integration to their
marketing clouds
SOURCE: Forrester Wave Report (2017 Q2)
Vendor Strengths/ Differentiators
Krux (acquired by
Salesforce and
renamed to
Salesforce DMP)
▪ Top Forrester rating in market presence and strategy
▪ Integration with SalesForce marketing cloud
▪ Ingests raw data and applies machine learning to
allow insights to arise, eliminating the inherent
biases that algorithms entail
BlueKai (acquired
by Oracle)
▪ Integration with Oracle Marketing Data platform and
data cloud
Adobe Audience
Manager
▪ Top Forrester rating in current offering
▪ Integration with Adobe Marketing Cloud
(incl. Adobe Analytics)
Neustar
▪ Strong identity capabilities, with access to offline
personally identifiable information (PII) derived from
over 200 sources, amassed over the past 20 years
▪ Integration with Neustar’s multi-touch attribution
platform (using same tags)
Good to know:
For a marketer with ~$100M
per year in advertising budget:
▪ The DMP license cost can
vary from a few hundred
thousand dollars to over a
million per year – the pricing
formula’s depend on data
size, number of sources,
number of destinations,
optional modules, third-party
data etc.
▪ It takes 3-6 months to
activate a fully functional
DMP – depending on number
of data sources, complexity
of integration etc.
18. 18
McKinsey & Company
There are important questions to ask when evaluating DMP vendors
1 What is the match rate of your cookie cloud against my specific customer base? Do you support running a test using my data
set for evaluating the match rate results?
5 Do you support my owned-channel technologies? Do you have built-in connectors to my specific personalization engines? Email
platforms? Marketing automation platforms?
4 Do you support all my paid-channel technologies? Do you have built-in connectors to the DSPs I use? My specific search and
social advertising platforms?
Does your 3rd party data mart include all the providers relevant for my company? What is the pricing structure for those data
providers? Do you apply a mark-up?
3
How do you support on-boarding of first-party data? Do you support the standard LiveRamp, Neustar services? Do you have any
additional device or identity graph services built-in?
2
How extensive is your built-in reporting? Would you demonstrate a few standard reporting use case?
6
How user-friendly is your user interface? Would you support having our users run a few standard use cases to experience the user
interface?
7
Do you provide a raw data feed? How complete and detailed is that feed? Can I get raw data for my data lake? At the user-ID-
level? With the custom audience segmentation attached? With the device stitching data included?
8
How much does your DMP cost? What is the pricing formula in terms of fixed annual fee, rate for data, fee per connector, fee per
server call, implementation fee, etc.? What is the cost for my specific situation? What do data additions cost later on?
9
What will the implementation effort be for my situation? Do you charge separately for that? How many resources do you provide
and for how long? Will I need a third party implementer and/or internal IT resources, and if so, how many people and for how long?
10
19. 19
McKinsey & Company
A series of challenges and limitations should be taken into consideration when setting up a DMP
Data is
anonymous
DMP has
incremental
costs
Match rates
vary
Challenge/ limitation
▪ While data anonymity is a blessing because it
allows marketers to target anonymous user
groups on the internet while steering clear of
privacy laws, it also means that marketers cannot
store personally identifiable information (PII) or work
with individual user records
▪ In addition to establishing a DMP, own an
internal Marketing data platform (MDP) that
allows marketers to store and work with PII data
▪ Establishing a DMP and purchasing third-party
data have associated costs and yield ROI trade-
offs. Hence, a marketer must evaluate whether the
increased targeting precision can lead to conversion
and revenue benefits exceeding the costs of setting
up and operating a DMP
▪ Conduct a DMP pilot to test use cases and
quantify value impact
▪ Closely monitor ROI and set thresholds for data
acquisition
▪ Internet data is messy with imperfect coverage
and quality. With DMPs, these imperfections
impact the match rates (between website visitors
and the DMP pool, the DMP pool and the advertising
platform pools, and uploaded customers and the
DMP pool)
▪ Match rates are influenced by many factors and can
vary greatly (between 10%-90%)
▪ Before choosing a DMP vendor, conduct tests
across multiple DMPs (using the marketer’s own
specific datasets) to ensure selecting a DMP
vendor with match rates high enough to warrant
the investment
▪ Conduct tests with different 3rd party data providers
for the chosen DMP to optimize match rates by
refining the 3rd party data/ data provider selection
Potential way to address
20. 20
McKinsey & Company
We have identified 4 common misconceptions to avoid in DMP setup and operations
Misconceptions
It’s all about
technology
While the technology is important, companies must ensure that the right talent and
processes are in place to maximize value capture through the DMP. Technical training is essential
to ensure that non-technical operators are able to understand the functionalities of the DMP and
best use this tool in their campaigns
There is a “one-size-
fits-all” solution
The DMP design depends on the marketer’s unique needs and must enable the use cases
relevant to the marketer’s business priorities. For instance, a marketer with a particular need in
using international marketing channels to target their customers/prospects abroad needs to ensure
that its DMP solution can reach those channels through its DSPs and also does not violate local
privacy regulations
This is a “one-time”
effort
Marketers should continuously evaluate the adequacy of their DMP, regularly updating it with
new 2nd/3rd party data sources, exploring new analytics/reporting functions, and ensuring that the
DMP interfaces smoothly with the rest of the martech stack
The DMP alone can
unlock the value of
data-driven
marketing
A DMP is only one part of the martech stack and has specific functions mostly addressing data
activation. Marketers should think about use cases when multiple components of the martech stack
work together (see next page for an example) and prioritize their marketing investments around the
priority use cases
Description
21. 21
McKinsey & Company
Illustrative example: A DMP is much more powerful when combined with a Customer Data Platform
Campaign
Company A wants to drive sales of its new mobile phone models (two models available: basic and premium)
To: 2) Using a DMP AND a MDP
To: 1) Using a DMP
From: Using DSPs only (without a DMP)
Concept
▪ More advanced segments get fed into the
DMP for activation
▪ DMP segments consistent across channels
▪ Integration of first party and third party data
to enhance segmentation
▪ Launches digital marketing campaign through
different DSPs, DoubleClick (by Google) and
MediaMath
How it works
▪ DMP collects past customer purchase data
from the MDP
▪ The DMP can now segment audiences to only
show ads to customers who have not made a
mobile phone purchase at Company A in the
past month
▪ A DMP creates a universal audience
segmentation for the campaign and activates
it consistently across multiple DSPs
▪ Now, the customer sees the same ad served
by all DSPs
▪ Audience segment options (e.g., location, etc.)
is defined by individual DSPs and not
necessarily consistent across different DSPs
(e.g., DoubleClick and MediaMath do not have
the same options for customizing the audience
segment)
▪ When the campaign is launched through both
Google and MediaMath, the same customer
can see inconsistent ads/ messages due to
these differences
▪ For instance, Google might have a segment
option “looking for premium phone,” which
Company A selects as the audience for the
premium phone offer. However, MediaMath
does not offer the same option, and a customer
who is targeted with the premium offer on
Google sees the basic offer on display ads
powered by Mediamath
Customer
experience
▪ Sees messages that are consistent across
channels and more relevant to the customer
▪ Sees consistent messages across channels
▪ Sees different messages in different channels
DMP helps ensure that the right ads reach the right people at the right time