How customer data platforms tear down the data silos

April 23, 2015 | By

How Customer Data Platforms Tear Down the Data Silos

Marketers have struggled to unify all their customer data ever since they started storing that data on computers – and probably before, judging from historical pictures of giant file rooms. But the need has become more urgent in recent years as interactions are spread across more channels and each channel produces larger volumes of information.

Happily, a problem this important attracts many solutions. The most promising recent approach has been the “customer data platform” (CDP), which Raab Associates defines as a…

“marketer-controlled system that uses persistent, cross-channel customer data to support external marketing execution”.

In other words, a CDP puts marketers in control of the database building project, and creates a database that can be used by many different marketing systems.

Many marketers, lost in their forest of data silos, will question whether such a mythical solution actually exists. But CDPs are indeed real. Most take the form of decision applications like lead scoring, campaign management, or content recommendations, which assemble customer data for their own use. To qualify as a CDP, those systems must expose their data to other applications. Ideally they would do this by making the raw data itself available for external queries, but exposing only processed results like model scores or recommended treatments will also suffice.

A visual thinker might distinguish the two varieties of CDP by drawing them differently. A “pure” CDP unifies only data and connects to multiple decision systems, which in turn connect to channel systems that deliver the selected messages to customers. A decision-making CDP unifies both the data and decision levels, still connecting to multiple delivery systems. The data-plus-decision CDPs are often called “platforms”, although it’s important to note that not every system calling itself a “platform” would fit this model.

The picture raises the obvious question of whether there’s a third alternative that unifies the delivery layer as well. Such systems do exist; they’re the “integrated suites” sold by vendors like Oracle, and Adobe.

Customer Data Platforms - David Raab

So now we have an embarrassment of riches: not one but three CDP configurations to consider from, with many choices within each category. How to decide?

In practice, the configuration decision tends to be pretty simple: only large companies choose a pure CDP because smaller firms can rarely justify building the database without also acquiring an application that will use it. The integrated suites also tend to be purchased by large companies, mostly because of the cost. (That said, the suite vendors are eager to sell to mid-size firms, but those buyers typically purchase only some components, effectively treating the suite as a platform.)

This means that most buyers will end up with a platform CDP, combining data and decisions.

But that doesn’t really solve the problem, since there are many platform CDPs to choose from. Key elements to consider include:

  • Data types: what data sources will the CDP be able to assemble? In particular, can it handle unstructured data such as Web interaction logs and social media comments, in addition to standard transactions. And how hard is it to add new data sources and data types? This type of flexibility is critical to ensuring you don’t outgrow your CDP as your business evolves.
  • Identity association: remember that the whole point of the CDP is to build a unified customer view by linking data from different sources that relates to the same individual. Unless your inputs will already be coded with a customer ID that’s shared across systems, you’ll want to be sure your CDP can build those links by looking for name/address matches, stitching together devices used by the same person, and incorporating external databases that have found matches between different customer identifiers.
  • Decision types: most platform CDPs started by providing a single decision application, such as content recommendations. Many of the vendors have now expanded the scope of products as clients request a broader range of capabilities. Identify your several most important applications and look for a CDP that can support them, either with its own features or through pre-built partner integrations.
  • Openness: along with identity association, the ability to connect with external systems is a core CDP capability. Be sure your CDP can work with external delivery systems and, preferably, with external decision systems as well. “Open API” is the buzzword to listen for, but be sure to dig beneath that to understand what your system will really be able to do. Like the ability to handle many data sources, the ability to connect with other applications will let your CDP adapt to new circumstances as they arise.
  • Usability: this is a tricky term, because what makes a system usable depends on the particular users and intended applications. So evaluating usability means finding a system that the people at your company can use to meet your needs. The skill levels and resources of those people and the nature of those needs will vary greatly at different companies, and so will the features a suitably “usable” system.
computer icon

See what BlueConic can do for you

Request Demo