statue representation of the Kelpies mythical water horse
PHOTO: Crawford Jolly

The only topic in marketing technology discussed more often than customer data platforms (CDPs) is how crowded and confusing the CDP space is. An excellent overview of the confusion, especially as caused by the recent Adobe and Salesforce announcements, sums it up well.

The hype around CDPs is the direct result of brands recognizing the competitive advantage of tapping into first-party data, while also recognizing that their existing technology infrastructure and processes are ill-suited to take full advantage of the opportunity. The CDP, which productized critical components of existing technology into a single solution that makes the whole of the tech stack greater than the sum of its parts, is the missing piece in many cases. But it doesn’t do everything — and it also does some things that many marketers don’t realize.

Let’s clear up some of the misconceptions, uncover the hidden benefits, and reaffirm the unique value a CDP provides.

What a CDP Does Not Do

Everyone loves silver bullets and panaceas. Related: everyone is disappointed. 

Forrester’s gleeful pronouncement (paywall) that CDPs over-promise and under-deliver for B2C marketers buried the lede, which is that a system acquired to try and solve every single issue of data hygiene, measurement/reporting, journey orchestration, master data management, personalization and world peace was probably going to fail. Thus, a few things that CDPs do not do:

  • Attribution modeling at a 1:1 level: Multi-touch attribution (MTA) is a great solution for marketers who struggle to understand the true value of their marketing and advertising touchpoints — which is pretty much everyone. But MTA is not designed to understand the individual customer in a way that a CDP is. In fact, those tools will never be able to do that because, legally, you cannot marry the anonymous touchpoint data to the PII data in the same system. So while the data in a CDP can be an input to MTA (once it is de-identified), one could never replace the other. They serve fundamentally different purposes.
  • Eliminate the need for IT: Yes, the original definition of a CDP explicitly identified marketers as the buyers and users of the platform. Despite a broader “business” application, the core purpose of a CDP remains to enable marketers and business users to access and activate customer data. But that never meant IT should be bypassed or removed from the process completely. CDPs provide data access that requires less reliance on IT to stand up and use the platform day-to-day. CDPs ought to be marketing owned-and-operated, but IT approved.
  • Replace all your databases: Achieving the “single view of the customer,” as a desired outcome, is not the same as consolidating all databases into one. This is due to the fact that the architectural design of a database is closely aligned with its purpose. For example, an ERP (enterprise resource planning) database looks very little like a Hadoop data lake designed for running business intelligence reports, which looks almost nothing like an Informatica MDM (master data management) database. A CDP can indeed replace some of your databases, especially the legacy ones you pay an arm and a leg to marketing services providers to maintain for you, but it’s unlikely the CDP is the last and only database you have.

Related Article: Clearing Up CDP Misconceptions

What a CDP Does That Might Be a Surprise

Now that we’ve gotten the Debbie Downer part of the article out of the way, let’s talk about the really cool stuff a CDP brings to bear that you might not have been aware of:

  • Predictions and decisioning: If you think that CDPs lack data science chops, you’re not alone. Yes, the extent and method will vary, which is why buyers should specify needs before evaluation. Ask to see the offering and its documentation. But CDPs definitely flex data science muscles in a lot of very cool ways, from 1:1 content and product recommendations to look-alike and suggested segments.
  • Reduced reliance on external vendors and providers: Unifying customer data at an individual level; designing a schema that supports structured and unstructured data; connecting a multitude of sources for both ingestion and export of that data; supporting the infrastructure — these represent a multi-billion-dollar industry over the last two decades or more. The big consulting firms, marketing service providers, agencies and vendors’ own services teams all bank (literally) on brands outsourcing the types of tools and processes that a CDP consolidates, simplifies and makes cost-effective. Services firms can still bring creativity and ideas around segmentation, algorithms and testing — which is far more compelling and value-driving than just building integrations.
  • Improved and expanded means to activate/access the unified data: We’ve all been there: you open up an exciting new toy or piece of technology and realize the manufacturer doesn’t include batteries or some other critical part in order to take advantage. Marketing technology has an equivalent: vendors that claim to be inter-operable and have expansive lists of other tools they “integrate” with, but really, all they offer are APIs or SFTP locations. Mega-disappointing and frustrating. CDPs (real ones) remedy this unfortunate situation by providing a multitude of integration options and are agnostic about where data comes from or goes to. Never have a sad birthday again.

Related Article: How Machine Learning Is Transforming the Way Marketers Engage With Customers

What All CDPs Should Do

And if they don’t, they’re going to disappoint. If you don’t care about these requirements, a CDP might not be what you need (which is a great thing to figure out before you buy!).

For a CDP to materially advance your organization toward an actionable, 360 degree customer view, it must:

  • Create a profile that spans all stages and types of identity: The concept of a person’s “identity” is fluid and ever-changing, but inextricably linked to a single person. Importantly, companies do not own or control identities. Rather, brands want to understand all of the dimensions of identity that relate to an individual in order to market to them more effectively. A profile in the CDP exists for the range of stages that make up the typical customer life cycle, from anonymous discovery to authenticated customer. This is critical in order to use the data to activate marketing actions across the full journey, rather than partially or fragmented by phase.
  • Support unlimited, real-time segmentation out-of-the-box, for all data types: This particular requirement is treated as a nice-to-have by many CDP market observers. In reality, it represents a major gate between a small group of pure-play, built-for-purpose CDPs, and the remaining crowd of aspiring or poser-CDPs like marketing clouds and DMPs. Not only is segmentation essential to bridge unified data and the activation of that data, the capability in and of itself reveals multitudes about the underlying data architecture: whether the data is actually stored in a single profile; how many attributes can be used to define a segment; if different segments can be shared with different integrations; how often membership in a segment is update and by what method. Segmentation matters.
  • Put power in marketers’ hands with a UI: I am a big proponent of marketers learning new technical skills in the context of the digitization of their jobs; however, in-depth fluency in APIs or writing SQL queries just feels outmoded. If business users are going to work in the CDP, the CDP needs an interface that simplifies the complexity of building segments and makes accessible configuring connections with external systems. All of the CDP’s core processes should be controlled in the UI. Full stop.

Last point: what a CDP does and doesn’t do should really be determined by your use cases and organization’s unique needs. But I hope this list can act as a starting point.

Related Article: Marketing's Chicken or Egg Silo Problem