AI dominates trade shows, boardrooms and sales conversations. It’s mentioned at nearly every event or webinar. With the potential of generative AI anbd other advanced AI models, this excitement is understandable.
Customer data platforms (CDPs) are not immune from this hype and excitement. Since a CDP can collect and consolidate customer data from many sources, AI’s role in CDPs certainly doesn’t go unnoticed. It’s become a core piece of product strategy among many leaders in the space. But does AI really need to be an integral part of a CDP?
The excitement of AI in CDPs
AI promises to bring a lot to the table for CDPs. It can enhance personalization, making customer experiences more tailored and relevant. Large language models can allow marketers to generate bespoke landing page or email copy based on the customers’ past behavior and purchases.
Then there is automation. With AI, routine tasks like data exploration, cleaning and sorting can be automated. Segmentation and queries can be completed by asking the CDP AI agent instead of building a SQL query or even using logical operators.
Finally, AI can be used to uncover insights into audience behavior that reactive analytics overlook or are just too hard to see granularly.
At a time when CDPs are facing many economic headwinds, this all sounds full of potential. What’s the catch?
Dig deeper: AI-powered features to look for in customer data platforms
CDPs: From pandemic boom to market slowdown
Before we discuss the catch, let’s step back for a moment and consider the current state of the CDP space.
CDPs have been a hot topic of discussion in martech for a long time. During the pandemic, there was a strong need to break down barriers like data silos. A lot of funding was also flowing into the space, both from customers willing to invest in technology and also from the market and venture capitalists investing in the development and growth of CDPs.
While the collapse and contraction of the CDP space hasn’t happened as some predicted, there has been some slowing across the space, both in funding levels and innovation.
It makes sense, as many early adopters of CDP were buying a solution to a single problem. They hadn’t bought a CDP before, and so they didn’t know how to select a solution for the long term. There was confusion about the best way to deliver value with a CDP, and many of those early implementations turned out less than spectacular (across multiple vendors). That’s not to say there weren’t large success stories, either.
However, in a slowing market where all decisions are scrutinized and rationalized, failures speak much louder than successes.
What does that have to do with AI? AI could be the shot in the arm that CDPs need to move again.
The problem with AI
There’s only one problem with the AI in a CDP. There are a lot of potential but very few concrete successes. And most of them aren’t very differentiated. Simply adding a customized OpenAI model on top of an existing CDP stack does little more than check the box to call yourself AI-enabled.
Recently, I was on an industry conference panel, and a fellow panelist summed up the market well. To paraphrase, “There are a lot of killer ingredients derived from AI, but not a lot of killer apps yet.”
We have the ingredients to create a brand-new dish, but right now, we’re just sprinkling them on top of things we’ve already made.
The reality is that AI works best with large amounts of data. Sometimes, what is in the CDP is enough, but often, there’s too much data missing from the CDP to really unlock the potential. Corporate data product descriptions, specs, and use cases are often missing from the CDP stack, which limits the potential for personalized content.
As advanced as the AI models are, they are not nearly perfect. AI systems can sometimes perpetuate biases in the data they’re trained on. This can lead to unfair treatment of certain customer segments, which is not only ethically wrong but can also harm your brand’s reputation.
The idea of a biased flywheel has become a topic of recent roundtable conversations I have participated in. Suppose a bias exists in a segment and AI uses that information to make decisions. In that case, it will further enhance that bias, magnify it and create a self-fulfilling prophecy of perpetuating that bias in new markets, new customers and new offerings.
Dig deeper: How to make sure your data is AI-ready
The AI opportunity
Given the rapid development, enhancement and improvement of AI from the major players, there’s no doubt about the impact AI will have. However, it can also become a major drag on a brand’s resources if every application and piece of the tech stack implements pockets of AI.
While the idea of AI in a CDP sounds like a winner on the surface, brands need to consider a broader AI strategy and focus on investing in an AI framework that consumes data from all sources, including but not limited to the CDP.
I believe CDPs should resist the allure of trying to be central use case for AI. They would be best served doing what CDPs were born to do: making customer data available to whatever system needs that data to be successful.
Dig deeper: How to assess your organization’s AI readiness with the 5P framework
Contributing authors are invited to create content for MarTech and are chosen for their expertise and contribution to the martech community. Our contributors work under the oversight of the editorial staff and contributions are checked for quality and relevance to our readers. The opinions they express are their own.
