Explore how the evolution of Customer Data Platforms (CDPs), fueled by AI and the need for real-time engagement, is transforming the way businesses personalize customer experiences—featuring insights from industry pioneer David Raab.
It’s not every day you get to chat with the person who coined an industry-defining phrase.
That’s exactly what happened though when I sat down with David Raab, founder of the CDP Institute and the brain behind the term “Customer Data Platform” at this year’s CDP Live. Since David introduced the term in 2013, CDPs have gone through several evolutions, most recently brought on by the influx of AI and the never-ending race to meet consumers' real-time expectations.
In our conversation, David and I discussed the hotly contested debate between composable and packaged CDPs, how consumer behavior is reshaping the industry, and why data isn’t just a tool for personalization, it’s about future proofing customer engagement and loyalty long-term.
Read on for the highlights from our conversation and learn about some of the tectonic shifts happening in the customer data platform industry.
*This interview was originally conducted as part of our CDP Live Event, and has been lightly edited. You can watch the full conversation here.
Key takeaway 1: Composable vs. packaged CDPs debate is more about terminology than functionality
Robin Grochol: Recently, there’s been one conversation dominating the CDP landscape: the merits of a composable CDP vs. a packaged CDP. At Twilio Segment, we don't think it's as cut and dry as choosing one or the other. I'd love to hear your thoughts on this debate – why do you think it's such a point of interest recently?
David Raab: It’s certainly a hot topic in the industry. Composable is a technical term that means a system is comprised of components – not too shocking. But the way it's been used in the CDP industry is a little different: it tends to refer to systems where the data warehouse is built separately from the CDP.
It’s kind of how it was done before CDPs existed, an “everything old is new again” situation. Most people now refer to it as “warehouse-centric” or a “warehouse-based CDP” rather than “composable,” which is just semantics. Even though “composable” may not be the most technically accurate term to use...
The reason CDPs have evolved is that people decided they wanted a piece of packaged software to build that customer database, because most data warehouses weren't doing the job back then. And most IT departments would struggle if you asked them to do it.
But the argument being made is that data warehouses have evolved considerably since 2013. Now, it's plausible that you would have a data warehouse that could do what a CDP would do – which relates to building customer profiles.
So, collecting all the different data sources – not just structured data, which is traditionally what you're limited to in a warehouse – and merging it together to make those customer profiles available to other systems. Then, making it easy to add new data sources, which again, is a typical pain point for the traditional data warehouses.
The reality is that it depends on the company. Some have excellent data warehouses that actually are CDP-capable. Others, the data warehouse is built for analytical purposes, which is typical. But it certainly doesn't have the identity resolution needed to build customer profiles. It’s often missing a few other things as well, like connectors to marketing systems or applications in real time. Or access is another common gap in a data warehouse.
We're not saying it's impossible to construct your own CDP based on the components that you buy. But it isn’t something you should just narrowly assume: oh, yeah, we have a data warehouse, therefore we have a database that will work for us. Often the data warehouse will need substantial work, and then you get into the debate about which is more cost-effective. Is it to modify and enhance your data warehouse, or is it to go out and buy a package CDP?
Whether you build or buy a CDP, the CDP deployment should remain the same. Source.
Key takeaway 2: A 360-degree view of the customer depends on both CDPs and data warehouses
Grochol: I want to pull on this a bit more. Data warehouses are actually the second most popular destination on the Segment platform. Part of the benefit of using CDPs and data warehouses in tandem is the ability to pair real-time behavioral data with the contextual data that lives in the warehouse. Can you provide some examples of how businesses are using a CDP and a warehouse together to engage their customers?
Raab: I think that's exactly right. Most data warehouses don't pull in web data or deal with real-time information. That's why you use a system like Segment or another platform that was originally designed to do that. Anything involving real-time personalization requires real-time data, like website personalization, reacting to a dropped shopping cart, or giving your call center agents access to a complete customer profile when a customer pops up on their screen.
As far as the warehouse being a destination for the CDP, that's very intriguing. Because it wouldn't be a destination if the warehouse had all that stuff in the first place. So you have this data that the CDP collected, which the warehouse is not collecting. Typically, you're going to summarize it because warehouses are not great at handling unstructured data. And most of your web data and your real-time data is unstructured or semi-structured. If it's a JSON format, for example, then the CDP will put it into a format that's more structured and easier for the warehouse to handle within the traditional relational data model.
Key takeaway 3: Real-time personalization is a must have
Grochol: Personalization is critical across the customer journey. What capabilities do you think a business needs to make that kind of personalization possible across the customer journey?
Raab: There are a few key things. You need data, and that data needs to be unified so that all sources relating to the same customer are pulled together for a complete profile. This data must be accessible in real time for immediate use. Connectors to various systems—like call centers, websites, mobile apps, and analytical systems—are essential. You also need predictive capabilities, AI, or machine learning to build accurate predictions and recommendations. An orchestration engine is crucial to determine the right message and channel for each customer interaction, ensuring consistent and optimal treatment across their lifecycle.
Key takeaway 4: The shift to AI-driven personalization is inevitable
Grochol: Looking ahead, what do you see as next for CDPs? What's the next big innovation on the horizon, and how might that reshape the way organizations approach customer engagement in the years to come?
Raab: The next big thing is AI-driven personalization. We'll move away from campaign-based strategies to continuous, individual-focused interactions. AI can determine the next best action for each customer, making marketing more about responding to individual needs in real time rather than following predefined campaigns. This requires new metrics and significant organizational changes. The shift from marketing tools to enterprise tools means CDPs will be run by IT or data teams, which requires teaching technical teams about marketing needs. This change management is as crucial as the technology itself.
Watch the full conversation and learn more about CDPs and changing customer expectations in the full recording here.
Our annual look at how attitudes, preferences, and experiences with personalization have evolved over the past year.