The Golden Record: An Interview with David Raab, Founder of the CDP Institute

For most marketers the dream of a “single view of customer” remains as elusive as ever. Tired of waiting for IT to deliver a “golden record”, marketers are turning to commercial solutions called Customer Data Management Platforms, a trend likely to accelerate in coming years, according to the world’s leading CDP expert David Raab.

By: Stephen Shaw
Read time is 12–14 minutes

David Raab is the Founder of the CDP Institute, and a widely known expert and consultant on marketing technology and analytics.

A single source of the truth – a 360-degree view of the customer – a “unified profile” – the “golden record”. Call it what you will, it remains a utopian vision for most organizations, despite a massive investment over the past decade in building intricate, cloud-based enterprise data stacks.

Ever since the 1980s, when data-driven marketing first became practical for most companies thanks to PC-based computing, marketers have pined for a “single view of the customer”. The response by the IT gatekeepers was to build galactic data warehouses, usually on a costly Oracle or IBM relational database management system, and provide end users limited access for analytical reporting through custom built data marts. But in the early 2000s this monolithic architecture was overwhelmed by the deluge of “Big Data”. New storage solutions like “data lakes” were needed to ingest many forms of raw data from many different digital channels. But data lakes quickly became “data swamps”, filled with extraneous data and difficult to wade through. The breakthrough finally came in 2012 with the launch of the cloud-based Amazon Redshift data warehouse which was both fast and cheap. Its arrival unleashed a wave of other data technologies which modernized the practice of data management.

Yet despite all of this innovation the dream of a “Golden Record” remains on the bucket list of every marketer. Because processing large datasets and then moving sub-sets of data around to meet the needs of different users takes a colossal effort, no matter how automated the data stack may be. In fact, the technical challenges are so formidable, according to Gartner Research, that 80% of organizations are expected to abandon their attempts to create a single customer view in the next few years.

That is why an alternative solution, which first appeared around the time Redshift was released, has become popular amongst marketers: The Customer Data Platform. A CDP is simply a purpose-built platform for storing customer-level data. Unlike most other end users, marketers are not simply passive consumers of data: their job is to put the customer profile information to commercial use. A reliable, up-to-date, and comprehensive view of the customer, including all past interactions, is critical to success.

The term “customer data platform” was coined by David Raab who is a highly regarded expert and consultant on marketing technology. In 2016 he founded the Customer Data Institute whose mission is to serve as a clearing house of information about this fast-growing but fragmented (and confusing) category. The fledgling industry is now valued at $1.5 billion, according to the CDI, and grew by 20% last year.


Stephen Shaw (SS): You became a marketing technology consultant back in 1987 when the industry was in its infancy. Looking back, how would you describe the evolutionary arc of MarTech?

David Raab (DR): Well, the big turning point, of course, was the web, which ushered in a plethora of MarTech tools. We went from almost nothing to this infinite set of choices that we have today. That really was the inflection point. Next came the growth of SaaS – software as a service – which was a fundamental change in the industry. The pandemic may turn out to be another inflection point. Digital marketing is now something everybody does. MarTech used to be expensive and difficult but now it’s democratized and simplified.

SS: I certainly remember the PC based-era when desktop marketing systems first came along. So much has changed since then – except for the underlying principles of one-to-marketing. They remain the bedrock of data-driven marketing even today.

DR: Direct marketing principles go back a century at this point. I’m old enough to remember the days of direct mail when mailing lists were kept on addressograph cards! But you’re right: the fundamentals haven’t changed at all.

SS: What has changed is the fact that you’ve got to accommodate a multi-device, multi-channel customer. And that continues to be a significant challenge for marketing.

DR: There’s still quite a ways to go in a lot of organizations. Because it’s hard, right? The reality is, most businesses can get away without doing it. Even if they don’t do a lot of things we would consider to be “best practices”, the real success of a company has to do with how well you treat your customers and the quality of your products and the quality of your service and a lot of things that really aren’t under marketers’ control. And even if they don’t do the best job, as long as they don’t do a terrible job, they’re going to keep you as a customer.

SS: There are many barriers that businesses face in getting omnichannel marketing right, and data is one of them – dirty data, data fragmentation, you name it. Why has it been so hard for organizations to solve this problem?

DR: Well, the data piece of it is really hard. You can’t do a mediocre job with data and succeed. And it is really hard to pull that data together. The data matching has become way more complicated. We have many more sources. The sources are very different. Web data really doesn’t look much like point of sale data which doesn’t look much like e-commerce data. If you look at my phone number and my browser cookie and my device ID, they’ve got nothing to do with each other. It’s a whole other level of complexity to pull that data together, to get that “single view of customer” everyone loves to talk about. As long as marketers can survive reasonably well without it, they won’t make the extra effort to make it happen.

SS: Yet there is an increasing focus today on delivering a better customer experience. Organizations are realizing that quality data is essential for that to happen.

DR: Absolutely. Customers care about the quality of the experience, not the quality of the personalization. What they want is personalized service. They want returns to be easy. They want delivery to be easy. They want to push as few buttons as possible. Marketers still have a hard time wrapping their head around that.

SS: You coined the term “customer data platform” back in 2013 in a blog you wrote and then three years later started up the Customer Data Institute. What inspired you to do so?

DR: There was just this huge market need that wasn’t being served by the big vendors – Salesforce, Adobe, IBM, Oracle – the people who probably should have done it. They just didn’t quite grasp that dedicated customer data management software was different from what they were already doing. They kept telling their customers, “We’ve given you a tool for that. Just throw all your data into your CRM system or your marketing automation system.” Or some said, “Just pull it all together dynamically in real-time with some sort of a shared ID.” Well, that doesn’t work. And it took those big vendors a long time to figure out it didn’t work. It took marketers way less time. And in that time, all these other vendors came along and built systems that actually did it the right way. The Customer Data Platform Institute got started when a couple of those vendors approached me. I’d been writing about the CDP space, as you say, since 2013 and they said, “Let’s do something to promote the category.” So I came up with this idea of a vendor-neutral resource. And talk about the right idea at the right time – easiest sale I ever made. Every vendor I asked said, “Sure, sign me up.” And the Institute has just taken off since then because there’s such a need.

SS: What was the genesis of the category? Where did the vendors come from?

DR: There were a couple of different classes of vendors. Some of them came out of marketing automation or campaign management or tag management. Some of them came out of the B2B lead scoring business. And what they had in common was they had to assemble data from multiple sources in order to do what they wanted to do, whether that was lead scoring or campaign management or programmatic ad serving. Whatever it was, they needed data from more than one source. When I defined the term CDP, I realized it’s not really the application that matters – it’s the database. Because once you have the data assembled, you can use it for whatever application you want.

SS: The category has grown dramatically since you started the Institute.

DR: About 50 vendors are actually sponsors of the Institute but there’s about 150 companies that we class as CDPs. The biggest problem with CDP is it’s so confusing. A CDP is really a set of functions. The bulk of those 150 companies have a whole bunch of other functions in addition to the CDP functionality. So long as your software has that set of functions, we’ll count you as a CDP no matter what else the software does, as long as you share that data with other systems. CDP functionality is mainly creating unified customer profiles. That’s what makes you a CDP. If you just assemble the data for your own use and don’t share it, you’re not a CDP. A CDP has to share its data.

SS: CDPs are viewed as synonymous with marketing databases. Is that because marketers have been the main sponsors or purchasers of these systems?

DR: That’s right. Most of the early adopters were in marketing, particularly in retail and in media, both industries that have a lot of transactions per customer with a relatively low ticket and a short buying cycle. So you need segmentation and personalization which is the main thing you do with the CDP. The value of better data is immediately apparent in the quality of the segmentation.

SS: The year 2016 seemed to be an inflection point when the category really came into its own. Do you have any idea why the category suddenly took off at that point?

DR: You know, I’ve never been able to figure it out, as much as I’d like to take credit. The need was just more and more pressing, I guess, and people figured out they weren’t going to get help from their normal vendors. I wish I could point to something specific. And believe me, I’ve scratched my head many times over it.

SS: Just for clarification, can you share your definition of a CDP and explain how it’s different from say a DMP1 or a Master Data Management solution2?

DR: Sure. The official CDP Institute definition is “packaged software that builds a persistent unified customer database accessible to other systems”. Unified, meaning it brings in data from all sources. Persistent, meaning it stores it someplace. Database, meaning it’s organized around customers. A CDP offers a customer view of the data, not a product view or a geographic view. Accessible to other systems means it shares the data. It doesn’t just use it for its own purposes. A DMP, on the other hand, is a much narrower kind of product. It doesn’t actually store all the data. It’s built really just to identify audiences primarily for advertising. It has to work with huge masses of data. And the way you do that is you don’t store every transaction: you just have tags that say, “This guy’s a shoe buyer,” or “This guy’s bought in the last 90 days”. And because it’s mostly built around cookies, it doesn’t really store the data for that long. It throws it away after 30, 60, or 90 days because that’s how long cookies last. You don’t have that long-term persistence which is very important because you really want to know how your customer profile changes over time. So you have to store that data someplace. CRM and marketing automation email systems work primarily with their own data. They’re not very good at importing other data sources. And MDM systems are primarily used to set up master identifiers that link different records together. It can be used for other things that have IDs like products.

SS: Is the CDP a client database that draws data out of the data lake, out of the EDW2, out of the CRM system, and just simply unifies it, or is it really a means of adding value to the data and feeding that back out to other systems for activation?

DR: Data lakes are just giant pools of data used typically by a data scientist – data warehouses are limited in what they do. They don’t have the full scope of a CDP. The unique thing about the CDP is that it will pull in all the data from those sources and make it accessible as usable customer profiles. Now, some CDPs will have very extensive analytical capabilities – some will even have campaign management, and personalization, and recommendation engines, and some will do customer journey orchestration. The customer journey orchestration software will decide the right message and send recommendations to a call center agent or to a website while the customer is interacting. But the value of the CDP itself is constructing a properly formatted, accurate, clean profile and making it shareable with the systems of engagement. It’s not just for marketers anymore. Other departments need unified data to offer a better experience. So the CDP becomes a corporate level resource, not just a marketing resource.

SS: I know that data scientists and analysts today often spend inordinate time and effort just fetching data out of the data lake or data warehouse.

DR: That’s right. An enterprise CDP would make their life way simpler, because otherwise they just end up doing a lot of repetitive work.

SS: Gartner has recently said companies are about to abandon their pursuit of a 360-degree view of the customer, believing it to be unobtainable. I imagine you disagree with that statement.

DR: I’ve actually never liked the term 360-degree view. On some level, of course it’s utopian. There’s certain data that just makes sense to look up when you need it. But you have to use judgment about what does and doesn’t make sense. And I think there’s a lot of data that does need to be stored in a separate system that does what a CDP does. I don’t care if you call it a CDP or not. You have to make intelligent judgments about what does and doesn’t make sense to store.

SS: Has the CDP become critical to comply with privacy legislation?

DR: A lot of the things you have to do to comply with privacy regulations are the things you have to do to build a CDP. The first step in complying with privacy is discovering where all of your customer data sits in your systems. Well, guess what, that’s what you do when you build a CDP. If someone has a subject access request, well, guess what? That’s exactly what a CDP makes possible. If you want to store consent someplace central, well, what better place than a CDP? If you want to put controls or governance on the data, what better place than the CDP? So the CDP gives you a lot of the specific things you need to comply with privacy regulations. In that sense, it is a very, very useful technology and we thank all the privacy regulators in Europe for their help.

SS: Data is crucial to the customer experience, and enterprises are starting to realize that’s the case. Does IT now muscle their way back into the picture? And if IT gets more involved, are they going to favor building over buying?

DR: Well, you’re exactly right that IT is getting more involved because there’s this realization that the organization as a whole needs unified data for a variety of purposes, not just in marketing. And then, being IT people, they say, “Hey, we have a data warehouse. We have a data lake. We have an MDM system for that matter. What’s the big deal? We can build this thing. Why should we go out and pay somebody to do it?”. What they don’t really understand is it’s not just another data store. Managing that customer data is actually pretty tricky. Doing the data unification is pretty tricky. Having all the connectors to all your systems is a lot of work. Maybe you can get the 1.0 version from IT which meets 80% of the initial requirement, but then people need enhancements and at that point you’ve got the C team doing the maintenance on it and maybe they can tie their shoelaces. So you have this technical debt that just keeps getting worse and worse.

SS: The other potential trend is the verticalization of the category, where you have financial-specific data models, retail models, etc. Is that a likely direction the industry will go in?

DR: Oh, yeah, definitely. We’ve seen that for a couple of years now. In every industry, actually, we see specialist CDPs cropping up. They offer the functions that bankers need – that retailers need – that telcos need. There are now, for example, banking CDPs which have a pre-built data model. We see travel CDPs that can talk to ticketing systems.

SS: In terms of the future of the CDP industry, there’s some question as to whether it remains an independent, separate category. Do you see the CDP category being subsumed by the marketing automation guys?

DR: Salesforce and Adobe have finally built CDPs that meet our definition. Persistence was the thing that took them a while to accept. So they all have CDPs baked into their larger platforms, which makes total sense. And, again, from a user’s perspective, I really don’t care if it’s baked into something or not. It gives the users what they need it to do.

SS: You once came up with Raab’s Law, which says that marketing suites will always trump best of breed systems. Is that still the case today in this era of platform ecosystems like Salesforce and HubSpot?

DR: Actually, you’re the first person that’s ever mentioned Raab’s Law to me. People will buy separate applications and attach them to a core platform if they can’t get what they want in the suite. But integration is never easy. We just moved from an all-in-one tool that had our website and our email on the same platform to having WordPress over here and MailChimp over there and Zapier in between. And it has taken months to make what is the absolute, simplest integration in the universe. Like it’s supposed to be click, click, click, it’s done. Well, guess what? It’s not. And our developers are tearing their hair out to do a very, very simple integration. Salesforce has done a great job with the app store that they have, right? But once you get past the really simple stuff, there’s always technical work involved that will never ever go away. And that’s always going to make it easier to just buy one system to do it all.

SS: Does the term MarTech ultimately becomes obsolete? This isn’t about marketing anymore. This is about one customer, one relationship across multiple channels.

DR: Yeah. It’s not just marketing. For a while, we talked about MadTech, which is marketing tech plus ad tech. Some people actually use that term, but it really is CXTech in a broader sense. And we’ll get there. Because the boundaries are disappearing. Marketers think they’re in charge of customer experience, but nobody else thinks they are.

  1. A Data Management Platform DMP is a software platform used to identify audience segments in digital advertising.
  2. A Master Data Management (MDM) solution is an enterprise software product that supports the identification, linking and synchronization of master data across different systems.
  3. An Enterprise Data Warehouse (EDW) is a central repository of integrated data from one or more disparate sources and is mainly used for creating analytical reports.

Stephen Shaw is the Chief Strategy Officer of Kenna, a marketing solutions provider specializing in delivering a more unified customer experience. He is also the host of a monthly podcast called Customer First Thinking. Stephen can be reached via e-mail at sshaw@kenna.ca.