The Death of CDPs: A Missionary Perspective on the Future of MarTech
Buyer journeys have shifted, AI is reshaping CX, marketing, and composability is the only way forward. The winners will consolidate around data + decisions, not channels.
From CDPs to Decision Intelligence: My 10+ Year Journey
Ten years ago, I co-founded Zylotech (Acquired), one of the earliest Customer Data Platforms (CDPs) pioneer in this category as an MIT spin off. Various Analysts including Gartner and Forrester recognized us early (leader corner) after CDP institute use to maintain a small list of CDPs in 2014, enterprise giants like Cisco, Google, Dell, Lowes, Palo Alto Networks, Travelers and P&G to name few who trusted us.
CDPs promised a “single view of the customer” and the ability to unlock first-party data for marketing and CX. For a while, they delivered. But today, the CDP era is ending.
What replaces it isn’t just another tool. It’s a new operating model for marketing and GTM operations in a new world where buyers journey has disrupted & Agentic AI is reshaping our lives in an unprecedented way, while your stack today has to be future proof: composable, warehouse-native, and AI-ready - with Data, Decisions, Orchestration, and Engagement as distinct layers.
The Rise and Fall of CDPs
Between 2016–2020, the CDP category exploded. By 2023:
The CDP Institute tracked 194 vendors worldwide.
The industry employed 16,800 staff and had raised $7.5 billion in funding.
Industry revenue reached $2.4 billion, with Salesforce and Adobe launching their own CDPs, acquiring multiple CDPs.
The “CDP moment” was fueled by vendor multiplication, aggressive pivots into the label, significant venture funding, and broad recognition of CDPs as a must-have MarTech component.
But by 2023–2024, cracks were undeniable: client horror stories, vendor consolidation, and the rise of warehouse-native alternatives all signaled decline. Add in the AI innovation cycle, changing buyer journey and suddenly everyone was saying: “CDPs are dead.”
Why CDPs Failed to Deliver
As a framework, the CDP concept was solid. But execution created new problems alongside the broader shifts in buyer journeys and marketing noise:
New silos – A “shadow data platform” outside enterprise data infrastructure or lets CIO say or control.
Scalability gaps – Struggled with real-time, high-volume event streams (Identity & Signals).
Rigid architectures – Too slow to adapt to fast-changing business needs.
Self-service myths – Marketers needed near-MDM skillsets & resources just to manage data.
Unfit for AI – Without knowledge engineering, most CDPs can’t support agentic AI.
Composable CDPs tried to fix this with best-of-breed stitching, Reverse ETL, or orchestration focussed use cases. But they still fell short on above reasons. The problem wasn’t just data activation. It was the lack of a true decision layer a way to operationalize intelligence, tie together orchestration and learning loops, and drive business outcomes.
The Cloud Data Warehouse Eats the CDP
A recent MarTech.org piece, “Cloud Data Warehouses Set to Disrupt the MarTech Stack,” captured what I see every day: the true enterprise foundation is now the cloud data warehouse.
Platforms like Snowflake, Databricks, Google BigQuery, Microsoft, AWS, and cloud-native MDMs like Reltio are taking over the heavy lifting of data unification, real time analytics enablement and governance.
Why?
Scale & speed – Petabytes of structured and unstructured data, real time.
Composability – Mix-and-match hubs or DIY, depending on use case.
Cost efficiency – Pay-as-you-go beats bloated CDP contracts.
Flexibility – Open APIs connect directly to MarTech tools.
Instead of duplicating data into a CDP, enterprises now consolidate around their existing data platform or warehouse. Marketing and CX teams then plug in Customer Engagement Platforms (CEPs) like Salesforce, HubSpot, Adobe, or Braze.
The Data Cloud is the system of record.
The Engagement Cloud is the system of engagement.
But here’s the problem: both layers often claims to be CDPs and to show modern features often claims orchestration capabilities and both oversimplify it.
Data platforms pitch “activation” by piping attributes into tools, but orchestration here is just plumbing. For eg: Enterprise do need an Integration platform like Workato etc. But thats not Intelligence activation.
Engagement platforms pitch “journey orchestration,” but most of it is static branching logic, not real decisioning. Although recently how Braze (offet fit), Movable Ink, Hubspot does “content decisioning” is perfect example to create relevant engagement beyond manual a/b testing in a AI first way.
But neither solves the harder problem: decisioning upstream at audience or customer data/signals level, with orchestration as an outcome of intelligence rather than a patchwork of triggers. i.e Audience decisioning (who/when/where), Pricing & Promo etc
The Missing Link: Modern Decisioning Layer
Built out Cloud Data Warehouses are excellent at storing and governing data. Engagement tools are excellent at delivering campaigns & experiences. But neither answers the real marketing question:
“Given everything we know about this customer right now, what should we do next?”
Composable CDPs tried to bridge the gap with Reverse ETL or Activation, but created new problems:
Burning warehouse compute with mindless SQL-driven activation.
Pushing integration complexity onto marketers and ops teams.
Missing the core Audience Intelligence piece (Signals Vs noise based attributes)
What’s needed isn’t just data activation - it’s intelligence activation. That’s the role of the Decisioning Layer.
The Stack Flip: Channels Don’t Drive Decisions
Most MarTech stacks today are still channel-first. MAPs and engagement tools own the decision logic by default. That’s backwards & will fail at multiple levels.
In the modern model, decisioning moves upstream.
Decisioning owns the signal, logic, prioritization, and customer intent.
Channels are just delivery trucks. They execute, but they don’t decide what’s inside the box. (we still havent seen Adtech + Martech) coming together.
Research has proven that first party data platforms or foundation is critical for even Adtech channel success, instead of point based/channel approach
Decision Intelligence: The Brain of the Stack
At iCustomer, we call this the Decisioning Layer - an always-on AI brain that sits between your existing data cloud and your customer engagement platforms.
It spans three functions:
Identity & Signals – Unify customer identity and signals, enrich beyond first-party data, and make it actionable.
Decisioning & Analytics – Combine policies, predictive models, experiments, and constraints to choose the next-best action.
Orchestration – Move beyond Reverse ETL into agentic workflows that execute across channels and capture feedback in a self-learning loop.
Most enterprises already have predictive models sitting idle. What they lack is contextual data, real-time inference, and a way to operationalize decisions. Decision intelligence makes this possible.
From Reverse ETL to Agentic Workflows
Reverse ETL was a necessary bridge. But it stops at pushing lists or attributes back into tools. Decision intelligence powers something bigger: agentic workflows — semi-autonomous processes that act on behalf of marketers with guardrails.
Agentic workflows:
Free teams from endless list pulls and manual segmentation.
Adapt campaigns in real time as conditions change.
Unlock “inference as a service” so marketers don’t need to be data scientists.
Ensure every channel is powered by one consistent, intelligent brain.
This is the move from automation → augmentation: human creativity paired with AI-driven decisioning.
What This Means for CMOs, CROs, and CIOs
This isn’t an IT architecture debate. It’s a growth debate. The impact shows up directly in your metrics:
Lower CAC – Better targeting, less wasted spend.
Faster pipeline velocity – Sales gets better-timed leads.
Improved retention – CX teams act before churn happens.
Better ROAS – Media optimized in real time.
Your data team runs the warehouse.
Your engagement cloud executes campaigns.
Your decision intelligence layer connects the two- turning data exhaust into outcomes.
The New Equation
The winning formula is not to call it All in One CDP or call composable CDPs as “Cloud Data Platform + CEP/MAP.” It’s:
Reliable Data Layer + Decision Augmentation + Agentic Orchestration + Content Creativity = ROI
Enterprises that embrace this cloud-centric, intelligence-driven stack will outpace those still patching CDPs or multiple CDPs. Decision intelligence is the missing centerpiece which rest on Identity & Signals + Analytics + Decisions + Orchestration + Learning Loop.
Final Word
The traditional CDP vendor era is over. Enterprises have already embraced the CDP framework but in a composable-first world:
The Cloud Data Warehouse is the foundation.
The Decisioning Layer is the brain.
Orchestration is the muscle.
The Engagement Cloud is the brand experience.
The question is no longer whether to build or buy a CDP. It’s whether your organization is ready to operationalize decision intelligence and outcome-focused orchestration in a data-rich, decision-poor world.
That’s the future we’re building at iCustomer. Are you ready?
Thanks for reading Data to Decision Activation! Subscribe for free to receive new posts and support my work.