A customer data platform contract usually runs three to five years. Renewals are sticky, migrations are painful, and the architecture you pick quietly shapes every campaign, journey and report your team can build on top of it. So the most important question in front of a CDP buyer right now is not which vendor scores highest on a feature checklist. It is which of two roads the whole category is walking down, because they lead to genuinely different places.
Gartner gave the two roads names in its 2026 Magic Quadrant for Customer Data Platforms, published in early 2026. The report says the market is bifurcating into platformization and agentification. Those are not marketing words a vendor coined. They describe two real and incompatible answers to the question of what a CDP should be. Understand the fork, and a confusing market suddenly has a shape. Miss it, and you can buy a perfectly good product that is built for the wrong future.
Where the fork came from
For most of its life the CDP had one job description. It collected first-party customer data from every channel, resolved identity into a single persistent profile, and let a marketer build segments and push them to downstream tools. The marketer was the user. The CDP was a place you logged into.
Two pressures pulled that simple picture apart.
The first came from below. Cloud data warehouses, Snowflake, Google BigQuery, Databricks and Amazon Redshift, became the place where enterprises actually keep customer data. Once the warehouse holds the real source of truth, copying all of it into a separate CDP starts to look wasteful. The composable, or warehouse-native, CDP grew out of that observation: leave the data where it lives, and activate it in place. The CDP shrinks from a data store into a thin layer of marketing logic on top of the warehouse.
The second pressure came from above. AI agents arrived, and they need customer context to do anything useful. An agent deciding what to offer a customer has to read that customer's profile, history and consent state first. Suddenly the CDP has a second possible job: not a dashboard for a human, but a context service for a machine.
Put those two pressures together and the category had to choose. It could become bigger, the deep foundation of a sprawling application suite. Or it could become smaller, a lean context engine that agents query. Gartner's 2026 report is the moment an analyst said out loud that both are happening at once, and that vendors are visibly sorting themselves into one camp or the other.
Platformization: the CDP as foundation
Platformization treats the CDP as the bottom floor of a large, integrated building. The data layer is no longer a product you think about on its own. It is the foundation that a stack of marketing, sales, service and commerce applications all stand on, and those applications are built and run natively on top of it by the same vendor.
This is the Adobe, Oracle and Salesforce path. Salesforce renamed its CDP from Data Cloud to Data 360 and made it the data foundation for Agentforce, its agent platform. Gartner placed Salesforce furthest to the right in the 2026 quadrant on both axes, and in its fiscal third quarter Salesforce reported that Agentforce and Data 360 together reached close to 1.4 billion dollars in annual recurring revenue, up 114 percent year over year. Adobe retired the Experience Cloud brand entirely and replaced it with Adobe CX Enterprise at its Summit conference in April 2026, an agentic system with the Real-Time CDP underneath it. Oracle, having shut its advertising business, was also named a Leader, with Unity as its CDP foundation.
The appeal of this path is real. When the data layer and the applications come from one vendor, the seams that usually cause pain disappear. Identity rules are enforced once. A consent change ripples through every connected system automatically. Decisioning can reach across marketing, service and commerce because they all read the same profile. Gartner specifically notes that regulated industries, banking, insurance, healthcare and pharma, gain from this, because coordinated execution with enforced consent is exactly what a compliance team wants. If your organization runs mostly on one of these suites already, platformization is the path of least resistance and genuine coherence.
The cost is the obvious one. A foundation is hard to pull out from under a building. Choosing a suite means the data layer, the applications and the agent layer all come from a single company, and switching any one piece later means confronting all of it. Analysis of the Salesforce position is blunt that clients see the best results when Data 360 is used alongside other Salesforce products, which is another way of saying the value and the dependency grow together. Credit-based pricing on these platforms is also unpredictable, and some buyers report burning through their allowance without clear results. Platformization buys coherence and pays for it in flexibility.
Agentification: the CDP as context for agents
Agentification points the other way. Instead of the CDP growing into a foundation, it shrinks into a specialist. Its job is to unify customer context and expose decisioning logic cleanly, and then get out of the way. Execution moves to AI agents that pursue goals, an agent that continuously optimizes a journey rather than a marketer who builds a fixed flowchart and walks away.
The mental model behind this path is a flatter stack. Instead of a warehouse feeding a CDP feeding fifty specialized point applications, the stack becomes warehouse plus a lean CDP plus agents. A capable agent sitting on a good context layer can do what many narrow tools used to do separately. The CDP stops being a destination and becomes plumbing: minimal, warehouse-native, and built to be queried by software rather than clicked through by a person.
Hightouch is the clearest example, and its trajectory tells the story. It started as a reverse ETL tool, repositioned as a composable CDP, and now markets itself as an agentic marketing platform. Gartner named it a Leader in the 2026 quadrant on its very first appearance, a striking result for a warehouse-native challenger. In April 2026 Hightouch raised a 150 million dollar Series D led by Goldman Sachs Alternatives and Bain Capital Ventures at a 2.75 billion dollar valuation, on the back of company growth above 100 percent in each of the prior two years. Investors are funding the agentification thesis directly.
The appeal here is flexibility and a smaller commitment. The data stays in your warehouse, which you already control. The CDP layer is thinner and, in principle, easier to swap. You are not buying a decade-long marriage to one vendor's entire application portfolio. The cost is different but just as real. Agentification does not remove work, it relocates it. The autonomy depends on clean data, clear ownership of definitions, and a team that can debug an agent when it acts on the wrong people. As one analysis of the split put it plainly, the work does not disappear, it moves. An agent with no shared definition of "active customer" will confidently act on the wrong segment, and someone has to catch that.
Why both are happening at once
It would be tidier if the market were simply moving from one model to the next, the way it moved from the DMP to the CDP. It is not. Platformization and agentification are advancing in parallel because they serve different buyers solving different problems.
A large regulated enterprise that already runs Salesforce or Adobe across its commercial functions has every reason to deepen into the suite. The integration is mostly done, the compliance story is strong, and the marginal cost of adding the CDP foundation is low. A digitally mature team with real data engineering talent and a warehouse it trusts has every reason to go composable and agentic, because it can absorb the work that path relocates and it values not being locked in.
What the fork punishes is the middle. The 2026 Magic Quadrant analysis is sharp on this point: there is not much stable ground between the two models. Vendors that are partially composable but still suite-driven end up expensive and directionally unclear, and several of them, including ActionIQ, Redpoint Global, mParticle and Zeta Global, dropped off the quadrant entirely under Gartner's updated inclusion criteria for CDP customers and contract value. The same squeeze applies to buyers. A half-committed strategy gets the lock-in of a suite without the coherence, or the engineering burden of composable without the flexibility. The fork rewards a clear choice.
How a buyer should weigh the fork
The decision is not about which path is better in the abstract. It is about which one matches your organization. A few questions cut to the center of it.
Who do you want to control the agent layer? This is the question buyers most often skip. On the platformization path, the agents that act on your customers come from the same vendor as the data, and that vendor accumulates a model of your customers more complete than anyone on your team holds. On the agentification path, you keep more control over which agents touch the profile and how, but you also own the governance, the guardrails and the failure modes. Decide deliberately whether you want that control or want to delegate it.
How deep is your existing suite commitment? If your marketing, service and commerce already run on one large platform, platformization is coherent rather than reckless, and fighting that gravity to assemble a composable stack may cost more than it saves. If you are not already committed, treat a suite foundation as a serious long-term decision, not a default.
Do you have the team? Agentification assumes data engineers, clear ownership of metric definitions, and people who can debug an autonomous system. If that capability does not exist and is not being hired, the lean path will quietly stall, not fail loudly. Be honest about the team you actually have, not the one on the org chart you would like.
What is the real cost to value? Look past the headline license. Platformization carries unpredictable credit consumption and switching costs that compound over years. Agentification carries warehouse compute, integration work, and the operational cost of governing agents. Both have a true cost well above the sticker.
A practical note for mid-market teams. Some analysis argues neither path fits the mid-market cleanly, because platformization assumes scale most mid-market teams lack and agentification assumes engineering depth they also lack. A growing number are reaching instead for customer engagement platforms that deliver visible results without forcing an organizational redesign first. That is a legitimate third answer: recognizing that the fork as Gartner frames it is an enterprise fork, and that a smaller team may be better served staying out of it for now.
What this means for the next five years
These are predictions, and they should be read as predictions. Gartner expects 40 percent of enterprise applications to embed task-specific AI agents by the end of 2026, up from under 5 percent in 2025. Forrester's Joe Stanhope has argued the future of the CDP depends on AI, and warns, in his own framing, that without sustained innovation the CDP risks sliding from a standalone category into an embedded feature. Both analyst houses agree the agentic shift is real. Neither can tell you exactly how fast it lands in your industry.
What is solid is the shape of the choice. The CDP is being absorbed in two directions at once: down into the warehouse and out into application suites, or up into an agent layer that treats it as context. Platformization offers coherence and pays in lock-in. Agentification offers flexibility and pays in operational burden. The buyers who do well over the next five years will be the ones who picked a side on purpose, sized the real cost of that side, and decided consciously who they wanted holding the agent layer that increasingly acts on their customers. The buyers who struggle will be the ones who bought a CDP the way they always have, on a feature comparison, and only later noticed the category had forked underneath them.
Council summary
This post argues that the 2026 Gartner Magic Quadrant for Customer Data Platforms named a real split, platformization versus agentification, and that a CDP buyer's main job is to pick a side on purpose rather than shop a feature list. The council verified the framing and the dropped vendors (ActionIQ, Redpoint Global, mParticle and Zeta Global) against Gartner coverage, confirmed the Hightouch 150 million dollar Series D and 2.75 billion dollar valuation against Business Wire and Goldman Sachs, and checked the Adobe CX Enterprise rebrand. Two claims were corrected: the post no longer calls Salesforce the sole Leader, since Gartner named four Leaders, and the Salesforce growth figure now reflects the actual reported metric, 114 percent year over year in combined Agentforce and Data 360 ARR, rather than a customer count. An unverifiable claim that Hightouch ranked highest on ability to execute was cut. The takeaway holds: the buyers who do well will size the true cost of their chosen path and decide deliberately who controls the agent layer acting on their customers.
Comments