Part Seven: CDP 2.0 and Zero-Waste

This post is part of the CDP 2.0: Why Zero-Waste Is Now series. For the best experience, we recommend starting with the Introduction and reading each chapter in sequence.

As foundational CDP features are commoditized, the challenge for CDPs to differentiate themselves will revolve around how well they can address the "big challenges" in using customer data to solve marketing problems. The big challenges orient around creating meaningful business outcomes through accelerating time-to-sustained-value, and eliminating wastage from the digital value chain.

The following is our view of what the "big opportunities" for the zero-waste CDP to solve are. The CDPs which can address these will differentiate themselves from the increasing commoditization of CDP feature sets.

The most important challenge is introducing intelligence into CDPs. Given the privileged access to the various data assets of their customers, CDPs are well positioned to provide insights and AI enabled solutions to marketers. For example, one significant problem for commoditized CDPs is they present a cold start challenge, requiring users to either have a sophisticated audience activation strategy, or to effectively guess which audiences to assemble and activate, which is both expensive and inefficient. Beginning with audience discovery, marketing teams need an easy way to understand how best to get started on their path to campaign success. 

Additionally, as marketers shift right along the audience activation lifecycle, workflow improvements powered by generative AI are being developed or are in early release stages by the leading CDPs. These will greatly reduce time spent on manual tasks such as audience discovery, assembly, measurement and optimization. By incorporating natural language processing (NLP) abilities, users will express their goals in normal language, eliminating the need to translate these into CDP or database logic. For example, users will use common language to create audiences, calculated profile attributes, and predictive audiences. 

These systems will also automatically set up experiment design, ensuring that marketers will find optimal data solutions without extensive guesswork or experimental design. Underpinning this capability will be the CDP’s ability to see the full customer lifecycle. This will lead to CDPs having the ability to automatically generate marketing strategies in response to business or campaign goals, further enhancing marketing effectiveness and reducing administrative tasks.

Eventually, CDPs will offer agentic AI capabilities. These will monitor data quality and business outcomes, and offer various strategies to improve outcomes. While there is plenty of innovation required before CDPs achieve this state, roadmaps leading to this goal are being contemplated by leading CDPs.

Underpinning the zero-waste CDP is the ability to optimize CDP-specific tasks which are compute intensive, especially when considering ML/AI. While exciting, ML/AI compute tasks, especially ones that consider historical data and real-time signals contemporaneously, are prohibitively expensive. This grows by orders of magnitude as the concurrent number of queries grows.

While low latency capabilities are often unappreciated, marketers with highly interactive customer journeys and/or low consideration offerings will need these to deploy "smart" personalizations. Current architectures struggle to contain compute costs when responding to events in real time. Without the ability to address compute scaling in low latency applications, AKA real time, CDPs will be limited in how much data and how many use cases they can address in a commercially viable way.  

Identity is a well-known challenge for marketers, and will continue to be for the foreseeable future. Driven by ongoing developments in privacy, regulatory, and technical markets, the ability to deal with identity challenges will have a direct impact on the leverage a marketer can exert with their data. These challenges cover a large number of use cases we collectively refer to as addressability—meaning how much your data can be leveraged in a particular context.

The spectrum of identity capabilities varies greatly across the CDP (and martech) industry.  While most systems have a basic level of identity management, handling the multiple contexts and regulatory regimes marketers face requires a deep knowledge of the use cases and technical sophistication. Furthermore, identity systems will need to work in real time to power use cross-channel interactions, which consumers are growing to expect. For example, if a customer clicks on an offer in a marketing email, then immediately walks into a retail location and opens their retailer’s app, they would expect for the retailer to be informed of their interest. This is exceedingly difficult to enable without a low latency identity capability.

Some other applications of addressability include increasing audience match rates with marketing platforms, custom identity spines, improving conversion attribution, enriching customer profiles, and grouping capabilities.

For organizations who operate in a multi-brand and/or multi-region context, a common challenge is organizing data in ways which enable global IT teams to ensure proper governance and controls, while at the same time providing brand or regional marketing teams sufficient access and self-service. Many of the use cases involve how global data assets are deployed and shared, while at the same time ensuring governance and performance are not compromised. From a technical perspective, this is understanding how data should be segregated and aggregated across a portfolio of brands and/or regions.

At mParticle, we have been working on these challenges for several years, as our client base is primarily large B2C marketers with a portfolio of brands and international marketing groups.  Previously, data silos provided (an often inadequate) solution for this issue.  As data centralization efforts progress, the data portfolio management challenge is starting to surface. In industries where brands are frequently moved or spun off, this issue will become even more prevalent. This is an issue which CDPs, ostensibly the primary data orchestrator, are positioned to handle, provided they are designed to address this challenge.

In times of commoditization, a common differentiator will be vertical specialization. A well understood specialization will be heavily regulated industries, an area we already see some CDPs addressing. The barriers to entry are fairly high in some industries and should provide some competitive differentiation from external CDP competitors. CDW and cloud platform CDP built-in capabilities may create some commoditization pressures, especially in the financial and insurance verticals where customers have significant IT capabilities themselves.

The ML/AI dynamic combined with vertical specialization may end up being the most important, but maybe under-appreciated driver in the entire CDP space. As general ML/AI capabilities are also rapidly commoditizing, CDPs which are positioned to develop unique AI capabilities due to their access to vertical use cases, modalities and processes will have substantial advantages over competitors who do not. If a virtuous cycle of vertical use case access and ML/AI product development gains momentum, we should expect to see a clear bifurcation of the "intelligent" platforms and undifferentiated "rest of market."