Gartner named customer data platforms (CDPs) one of the key technologies that will demand marketers’ attention in 2018. Michael Katz, Cofounder and CEO of mParticle, explains why CDPs are not just another acronym and how consumer brands ranging from Airbnb to NBCUniversal to Zappos are using them to optimize omnichannel customer experiences and marketing outcomes, in all the moments that matter.
Originally presented at AdExchanger Industry Preview 2018 by Michael Katz, Cofounder and CEO, mParticle.
2. THE DEATH OF
THE DMP
I’m here to talk about CDPs. I know that most of you who are not familiar with the acronym are thinking, “yeah, just another
acronym.” But hopefully, by the end of the conversation, I'll have convinced you that it's not just another acronym, but it's a
revolution. This slide actually pains me a bit. My last company, Interclick, was the first client of of Bluekai. DMPs have
been near and dear to my heart for a while. This presentation is not about the death of the DMPs so much as it is about
the change that's happening all around us, the change that's creating a new wave of data and a new need for innovative
data management solutions, all of the challenges and opportunities that come along with those changes, and ultimately
why the Customer Data Platform really is the only solution going forward.
3. WE HAVE BEEN CHASING
THE SINGLE CUSTOMER VIEW
FOR 30+ YEARS
Call it a 360-degree view of the customer, or a holistic customer view, or CRM, what I mean is the notion of understanding the
customer as completely as possible. This has been a key issue for marketers since the beginning of time. Certainly since the
beginning of database marketing. However, it has been almost nearly impossible for a number of technological reasons, as well as
a number of political reasons. This isn't about pixeling a bunch of creatives, getting the social data with the video data, and painting
the same vague picture that you've probably seen for the past 10 years. This is about truly understanding the customer,
understanding what they're doing on all the different platforms, devices, and screens that they interact with your brand on.
4. DATA PLATFORMS ARE NOTHING NEW
COMPLEXITY OF DATA MODEL
COMPLEXITYOFDATAMODEL
OFFLINE ERA WEB ERA CONNECTED ERA
MARKETING
DATABASES
Purchase data mapped to identity
Low volume
Rows & colums
Direct Mail Activation
DATA MANAGEMENT
PLATFORMS
ALL ANONYMOUS
Client-side architecture
High scale
Privacy & Governance Poor
Multi-tenant DB (Shared IDs)
Paid Ads Activation
CUSTOMER DATA
PLATFORMS
Designed around Identity
Privacy by Design
Real-time event streaming
Single-tenant (no Shared IDs)
High scale
Real-time APIs
Server-to-Server Architecture
Full Lifecycle Activation &
Integrations
20122001
Why now? Data platforms are nothing new. Data platforms have been around for 30+ years, dating back to before the commercialization of the
internet. Why now? Ultimately, because the world has changed. I'd argue that the change from web to app and the introduction of true omnichannel
experiences is as big of a change as the first move from analog to digital. Secondly, I'd also argue that CDPs are not competitive with DMPs, much
like you wouldn't use a marketing database to try to do what a DMP did for so many years. You really can't get a DMP to do what you need a CDP to
do. The interesting thing is that while a lot of people will view this as an incremental change, it's really anything but that.
5. 1900’s – MID 1980’S 1970’S – EARLY 1990’S 1990’S - PRESENT
INNOVATION IS USUALLY VIEWED AS
EXPENSIVE & UNNECESSARY
Most new products are met with a fair degree of skepticism before they get a mainstream adoption. That happens because most customers are
cynical about new technology. They also think that there’s still a lot of life left in the previous paradigm, and that the new paradigm is somewhat
unproven. If people are seeing even a little bit of success in the last wave of technology, they're going to be first to dismiss the new technology. A
great example is when we shifted from typewriters to word processors in the workforce. Word processors were adopted first by legal
professionals and people who were doing a lot of templated work. Ultimately, adoption happened gradually. People aren’t still using typewriters.
It's weird to actually see a typewriter today. That change is inevitable. If anybody has read Jeffrey Moore's book, Crossing the Chasm, you know
that there's a continuum of technology buyers and adopters. While a select few are adopting the future today, a lot of people are quick to dismiss
new technology and say “it's expensive, it's a luxury, it's not going to be the mainstream…”
6. “ THINGS DONE CHANGED”
- NOTORIOUS BIG -
But, as the late great poet, Christopher Wallace, said, “things done changed”. And it's
not just one or two things. There are a handful of significant changes that are
happening around us which are changing the need from the DMP to the CDP, which
also opens the opportunity to create that single view of the customer. Since this is
industry preview, I want to touch upon five key themes that ultimately converge upon
one theme, which is the customer.
8. 1990’s - 2013
www.
2013 - PRESENT
IT’S NO LONGER SINGLE-THREADED
First, the digital experiences has changed. It's no longer single threaded. Most people have a
minimum of six connected devices, which means that the customer journey is completely
fragmented. It also means that people are connected all the time. In the first 15 years of the
mainstream web, the only way to interact with a brand digitally was by going to their website.
That paradigm has totally changed, first starting with mobile. Now, you can interact with the
brand's mobile web property or download their mobile app, and can then extend into the living
room with OTT. I think we'd all probably agree that the next big platforms are going to be voice,
AR and VR.
9. AS THE CUSTOMER
EXPERIENCE
BECOMES MORE
DECENTRALIZED…
THE GREATER THE
NEED TO CENTRALIZE
DATA BECOMES
As this customer experience continues to become fragmented, there becomes a greater
need to centralize the data. Because of all of this device hopping, the customer journey is
not linear. The expectation from customers today is that when they start a purchase
process on one device or platform, brands know exactly where they left off when they pick
that process up on a completely different platform. Today, anything short of that is
completely intolerable.
11. IDENTITY
ANONYMITY
DIGITAL
ANALOG
There's probably no more important topic in the industry today than identity. Identity is going to dominate the headlines in 2018 more so
than any other topic. As we shift to this multi-device reality that ultimately ties to these persistent device IDs, there's a certain place for an
anonymous behavior, and that's typically at the top of the funnel. That said, the shift from anonymity to identity based experience is as big
a shift as the shift from analog to digital. Ultimately, what identity brings is stability, right? You don't have to worry about people deleting
their cookies. I know this is super tactical, but identity provides the foundation to bring stability, accountability, predictability, and
repeatability. Through identity, you can start to target people and not proxies.
12. ANONYMOUS
(NO PERSISTENT ID)
PERISHABLE ID
(COOKIE, AD ID)
PROBABILISTIC MATCH
(XDEVICE ID)
DETERMINIST MATCH
(HASHED ID)
KNOWN USER
(EMAIL, MOBILE)
VERIFIED USER
(2FA, VERIFICATION)
PSEUDOANONYMOUSID
IDENTITY IN THE CONNECTED ERA
Opt-Out
Vertical Regulation
Opt-Out
Opt-In
First Party TOS
Proxy-based targeting
Audience Analytics
Content Personalization
xDevice Ad Targeting
Multi-touch Attribution
Simple Journey Analytics
1:1 Promotions
Advance Ad Attribution
Multi-channel Marketing
Customer Support
Full Journey Analytics
Lead Scoring
Direct Sales CRM
Private Sales Access
LTV Analytics
Payments
Privileged Info Access
Contextual targeting
Identity is a bit of a convoluted
concept. It's used in so many
contexts across this industry.
People often think about
identity in the context of
matching, but there's actually
so much more to it. With the
emergence of all these
different devices, you actually
have lots of different forms of
identity which need to be
governed and managed lots of
different ways. Managing
identity along this continuum,
you have to account for the
differences in technical,
geographical, organizational,
and political contexts.
Additionally, you have to
assume that identity keys are
not static. I have multiple
email addresses. I change my
credit card info all the time. I
have multiple social handles. If
you're not accounting for this
stuff in your identity strategy,
you're not going to be able to
deliver the optimal customer
experience across all these
different platforms.
13. IDENTITY IS WHAT ENABLES EVERYTHING
UNIFIED CUSTOMER
EXPERIENCE
360 DEGREE
CUSTOMER VIEW
GDPR GOVERNANCE
& ENFORCEMENT
INTEGRATED
MARKETING
Additionally, you really have to think about identity as a framework. It's what allows you to deliver a unified customer
experience across all these different devices. It's what allows you to create a holistic view of the customer. It's what
everything gets keyed off of. It's what allows you to combine paid ads and retention, marketing strategy, so that everything
is not siloed and you can deliver that unified marketing experience. It's also what allows you to solve for GDPR, which we’ll
discuss shortly.
15. SPECIALIZATION CONTINUES IN THE SUPPLY CHAIN
The third big trend is that money and resources are being shifted to the cloud, and the budget for Saas tools is quickly following. An estimated 45%
of all corporate spending is going to be on on Saas tools. Nowhere is this growth more apparent than in the marketing tech landscape. We've all
seen the the LUMAscapes in various forms and formats, but they're dwarfed by the marketing tech landscape that ChiefMartech has been producing
over the past few years. You can't just call these ecosystems messy or fragmented, that's a little bit lazy and it's not necessarily accurate. This is
about specialization. Every time you get one of these consumer platforms coming to market with a ton of investment driving a lot of innovation, you
get a whole new crop of service providers popping up that are specialized for that platform or extensions of some of the existing tools. With so
many different systems in place, you can get to a best in breed stack. However, there are also a number of challenges. Data fragmentation, system
incompatibility, disjointed approaches, inconsistent analytics, and poor customer experience. The opportunity to integrate all these different tools, to
be the spine that connects them into a tightly knit data supply chain, moving data freely to and from each of these systems, is a massive opportunity
that has never really existed before.
17. GROWTH IN WEB APIS SINCE 2005
The fourth big point boils down to customer experience. The ongoing battle between the CMO and the CIO for budget control and influence
still comes down to customer experience, whether that's pre-purchase or post-purchase. Ultimately, you want to turn a casual user into a loyal
user and turn that loyal user into an enthusiast and an advocate. The digital transformation that's happening across many brands and
vendors in this room today is being built on the back of the API economy. A lot of the ad tech tactics like cookie syncs were never built to be
sustainable. At best, they're short term temporary hacks. At worse, they enabled fraud and security problems.
19. DATA IN MOTION VS. DATA AT REST
DATA
WAREHOUSING
1990’s
HADOOP &
MAP REDUCE
2008
STREAMING
DATA
2015
Drop relational assumption
Programmability
Open Source
Batch → Real-time
Daily → Continuous
In a world where everybody's connected all the time, the technology required to help brands keep pace with their customers has to be fast. It
can't be based on a batch system that takes hours or days to let data trickle in from the experience layer, whether that's the site or the app,
down into a data warehouse and allow that to be your single source of truth. The shift from data warehousing to Hadoop to now streaming
data with vendors and solutions like Kafka, Spark, and RabbitMQ ultimately allows you to create these real-time transmission of data back and
forth to accumulate a real-time view of the customer. This makes it possible to keep up with what they're doing regardless of the device.
20. MOST DATA ARCHITECTURE IS A MESSEXTERNALINTERNAL
API LAYER
RAW STORAGE
DATA
WAREHOUSE
HADOOP
CR
M
B
I
ETL JOB ETL JOB
This is what most companies’ data stacks end up looking like. This was fine,
historically, but it actually doesn't work anymore. Data is generated in real-time,
and the expectation is that if we're talking about customer experience, you know
about me as the user up to that second or minute at the latest.
21. ANALYTICS
ATTRIBUTIO
N
USER FEEDBACK
PAID
ADS
EMAIL & PUSH
CONTENT
EVENTS AUDIENCE
CREATE A REAL-TIME SOURCE OF TRUTH
That brings us to where we are today. Creating a real-time source of truth if the experience is based on two-day old data or week-old data is not
going to work. The technology that enables this wasn't around five years ago. A lot of this technology is new, and it's being commercialized by
companies like mParticle. But when you think about what the optimal data solution is, it really needs to have three pillars.
1. It needs to be open, connected to lots of different APIs, and able to write to those APIs as well as read from those APIs.
2. It needs to be extensible, able to get data back from those systems and accumulate data to a central profile that can then be shared out.
3. It needs to be real-time. Ultimately, the moments that matter don't last hours or days. Being there in real-time is the difference between
winning and losing.
22. WHY IT MATTERS
THE RIGHT , TO , .AD THE RIGHT USER AT THE RIGHT TIME
THE RIGHT , TO , .DATA THE RIGHT SYSTEM IN REAL-TIME
THE RIGHT , TO , .IDENTITY ANY SYSTEM EVERY TIME
For as long as I can remember, we've all been talking about delivering the right ad to the right user at the right time. Although that is
as cliché as it gets, how it gets done is by having the right data to the right system in real-time. The fact is that you can't expect to do
that if the right system isn't mapping the right identity across any different system all the time.
24. NEW MARKET REALITIES FOR THE CONNECTED ERA
Digital Fragmentation
(Web, Mobile, OTT, Voice etc)
SaaS Explosion
(Ads, Analytics, Marketing, CX etc)
Emergence of Identity
(Email, Mobile IDs, Device IDs)
RISE OF APIs ⎯ Data fragmentation
⎯ Systems incompatibility
⎯ Disjointed Approaches
⎯ Inconsistent Analytics
⎯ Poor Customer Experience
A CDP is really the only way that you can address GDPR at scale. It can't be a system built on cookies that are unstable, because if somebody
expresses their right to be forgotten and delete their cookies, they may come back two days later and notice that you're targeting them. These are
$20 million fines for every single instance. The stakes are super, super high. It cannot be a system in which there are shared identity spaces across
customers. A core tenant of any CDP, including mParticle, is that all customer data has to be logically or physically segregated from all other
customers’ customer data. It has to be a real-time system that is able to propagate those changes out and send delete commands out to the network
of partner integrations that you have cascading that command down to every system. It can't be isolated in one or two tools.