The Agentic CDP? Executive Summary, 1H 2026 CDP Industry Update
From the 1H 2026 CDP Industry Update, the 19th edition and the report's tenth anniversary. Back to the Industry Update
Ten years after the customer data platform became an established category, it faces a new test. AI agents can now read customer data and act on it, and every system that holds that data (the CDP, the data platforms beneath it, the marketing applications around it) is being recast as something an agent can use. Vendors, data platforms and analysts have each announced what this means for the CDP. This edition asks what the evidence says.
The short answer is that the agentic CDP is not yet a product class, and the CDP is not being displaced. What is happening is less dramatic and, for buyers, more consequential.
The claims have run ahead of the controls. More than half the vendors we track now claim at least one agentic capability. Far fewer document what a buyer most needs to know before letting an agent act on customers: what it may do, what record it keeps, and how anyone can tell whether it did the job well. Only six document all three. “Agentic” now covers everything from a chat assistant to an agent-building toolkit, and no independent mechanism yet tests what lies behind the label. The Institute intends to extend RealCDP certification to agentic capabilities. Until it does, buyers are choosing on claims.
Demand points the same way. Set against what buyers rank highest, vendor documentation is thinnest at what agents depend on: agent governance, the semantic layer, next-best-action, and privacy and consent. A company pressed to deploy agents is likely to find the market thinnest where its risk is greatest. The same comparison gives a manager a benchmark against peers: what most of the market already documents, what can wait without falling behind, and where proof should come before a launch date. The loudest agentic topics in the trade press, interoperability and assistants, run ahead of demand.
The hard part hasn’t moved. An agent is only as good as the customer data it works from, and it carries that data’s mistakes to every customer it touches. Members report the same kinds of difficulty across three survey waves. Today their leading obstacles are getting teams to work together and agreeing what the data means. Moving the data into a warehouse doesn’t solve either, and neither does adding an agent on top. That supports the CDP’s purpose, which is to resolve, define and govern customer data. It is also a caution for anyone who expects AI to make that work unnecessary.
What is changing is who does the work. Data platforms now offer customer-data functions of their own, warehouse-native vendors build on the customer’s data store, and engagement and experience platforms build profiles. The architectures differ most at the agent layer: vendors that work on the customer’s warehouse document agent context, interoperability and agent building far more often than those that keep only their own store. The CDP’s work is being divided among more suppliers, but accountability for the result hasn’t followed. None of the large data platforms yet supplies, by itself, the consent, capture and identity work an agent depends on. Buyers still assemble it, from a CDP or from several suppliers acting as one. Responsibility for a failure is getting harder to place.
The market is reorganizing, not collapsing. The number of vendors is almost unchanged, CDP vendors buy other companies as often as they are bought, and financing continues in smaller rounds. Forecasts that the category will be absorbed have yet to show it. They rest on revenue no one can separate by category, and they don’t show the CDP faring worse than other software facing similar pressure. On our reading, AI is reorganizing software that holds customer data across the board, and the CDP is one case of it: likely to evolve now, and to be transformed later along with the rest of MarTech.
What follows. For buyers, the test is simple to state and harder to meet. Judge an agentic offering by what it can show, not by what it lists, and settle who answers for each part of the work before an agent acts on a customer. For vendors, the claim on AI rests on the resolved customer data, context and governance they bring to the ways buyers already obtain AI. And for anyone weighing the category’s future, the question worth asking isn’t whether the CDP survives, but which parts of its work move, to whom, and with what accountability.
This edition provides the baseline for answering it: a sourced, dated record of what every active vendor documents, against which later editions can measure change.
Download the full report (members) · Back to the Industry Update

