demystifying the CDP

Forget about the marketing technologies you currently use or plan to use for a moment and just think about your customer experience. At the highest level, your customer experience is a factor of all your different data sources (that you use to understand your customers) and each of your channels (that you use to interact with customers).

The customer data platform (CDP) has emerged as the technology that sits between your data sources and your engagement channels. Its purpose is to take in and make sense out of all your data and then make decisions about how to respond. Evergage’s CEO and co-founder, Karl Wirth, recently presented a webinar called Demystifying the Customer Data Platform to provide a 30-minute primer on CDPs.

He described the five main actions a CDP needs to be able to take in order to turn data into effective engagement: integrate, understand, decide, engage and analyze.

demystifying the CDP

In this blog post, I’ll describe the basics of each of those actions, but be sure to check out the full webinar replay for all of Karl’s takeaways.


Good data is at the core of any good customer experience — especially a personalized one. You can’t deliver a personalized experience if you don’t know anything about the person. But of course, you already know plenty about your customers. That information just lives in many different places across your business so it’s not always accessible to the systems that need it.

A CDP needs to be able to integrate all of this disparate data together in one place. That means relying on things like:

  • JavaScript tags and SDKs to collect behavioral data from your website and apps
  • ETL and APIs to incorporate your call center interactions, offline purchases and data from other sources
  • Out-of-the-box connectors to bring in data from common technologies such as ESPs, MAPs and CRMs
  • Custom connectors that you can tweak to fit your own needs, because out-of-the-box integrations may not always account for your company’s unique circumstances

The way the data is brought into the CDP will vary based on the source, but ultimately the CDP needs to be flexible enough to work with all the systems your company uses.


Bringing the data together is just the first data step. Your data doesn’t mean anything to you if you can’t discern what it says about a person’s needs and preferences. Thus, your CDP needs to be able to interpret data to understand a person.

A CDP brings data together into a unified customer profile for each person (what Karl calls “the beating heart” of the CDP) and stitches and merges profiles together to create one clear picture of every single individual.

Then, it must leverage machine learning to analyze what that data says about a person’s current affinities and preferences (with affinity modeling) and future plans (with predictive scoring).


Once you understand a person, you can rely on your CDP to decide which experience is the most relevant to her.

This decision is typically made through rules/triggers or machine-learning algorithms.

With rules or triggers (such as triggered emails), you manually tell the CDP which experience to select once specific criteria are met or when individuals fall into specific segments. For example, you could tell the system to send a push notification to customers within a certain geolocation. In this case, the CDP makes the decision you tell it to make.

With machine-learning algorithms, you rely on the machine to select the best products, content, offers, promotions, brands, categories, etc. to display a person in any channel based on everything you know about her. For instance, if you have several different promotions to feature on your homepage, an algorithm can determine which is the best for each person. In this case, the CDP makes its own decision.

Check out our ungated eBook, The Marketer’s Guide to Machine-Learning vs. Rule-Based Personalization for more insight on decision making.


After the decision is made, the experience needs to be delivered to engage the customer in your channels — either through the CDP taking the action itself or passing off the information to another system to take that action.

Of course, the CDP won’t just take action on its own without some kind of direction from the marketer. The ability to build, launch and test campaigns is critical for effective engagement. The CDP must give you the ability to:

  • Pass segments or 1-to-1 context to other systems
  • Manage your campaigns driven by rules and machine-learning algorithms
  • Test your campaigns and measure the results to continue to optimize


Finally, beyond looking at the results of your campaigns, you also need to analyze your data to draw out insights about your customer base or business operations.

Your CDP should give you the ability to slice and dice all of your data in any way you wish. That means allowing you to analyze differences in your segments, leverage business intelligence tools and set up your own data science models. It should also deliver predictive alerts to help you identify any potential issues with your business.

Final Thoughts

Our advice for businesses researching CDPs is to avoid getting hung up on the terminology and the category and just look for technology that meets your needs. You need a solution that can synthesize all of your data sources into a single picture of each customer, and then act on that information to deliver a relevant and engaging experience across channels — plus analyze that information to draw out insights about your customers.

You want to find a solution that can integrate, understand, decide, engage and analyze — whether it’s called a CDP or not.