Who Should Own CX AI? Why the Answer is More Complicated Than it Looks

AI is exposing a problem that has existed in customer experience for years: no single department actually owns the customer journey. Marketing owns acquisition, customer service owns support, and IT owns the technology. AI cuts across all three at once, making organisational boundaries far more visible than they used to be.

That visibility is forcing a question into the open that many organisations had quietly avoided. A company decides to invest in AI for customer experience, and almost immediately the arguments start. Should marketing lead, given its grip on personalisation and customer data? Should customer service take charge, since most AI pilots launch in the contact centre? Or does it belong with IT, which carries the infrastructure, security and compliance burden? The biggest risk organisations face is not picking the wrong owner. It is allowing CX AI to become everyone's responsibility and, in practice, nobody's accountability.

The instinct to ask "who owns this" comes from how most enterprises have always run technology projects whereby you assign it to a department, give that department a budget, and hold it accountable. Customer experience has never fitted that model cleanly, since it spans marketing's acquisition and engagement work, customer service's support and retention remit, IT's infrastructure and governance responsibilities, and the data teams sitting underneath all of it. AI now touches every one of these areas simultaneously, which is why the question of ownership has become so much louder than it was even two years ago.

Each function has a legitimate claim, and each comes with a genuine blind spot. Marketing teams already manage personalisation, customer data and journey optimisation, so AI initiatives often emerge naturally from that territory, bringing strong customer insight and an experience-led mindset. The risk is that marketing-led AI tilts towards acquisition rather than the full journey, with limited visibility into what happens once a customer reaches the service function.

Customer service, meanwhile, frequently controls the largest AI budgets within CX, since most deployments begin in the contact centre. In a 2025 Gartner survey of 321 customer service and support leaders, satisfaction, efficiency and stronger self-service performance emerged as the priorities leaders were most focused on heading into 2026, with AI increasingly positioned as the route to resolving more issues on first contact and easing the burden on customers navigating support.

Service-led ownership delivers direct visibility into customer problems and faster routes to demonstrable ROI, but it risks a narrower focus on service interactions rather than the entire journey. As Kim Hedlin, Director of Research in Gartner's Customer Service and Support practice, explained: "Service organizations are entering a period where AI and human expertise must work in tandem." Leaders, in her view, are not simply rolling out AI tools but rethinking how service operates so that technology and human judgement reinforce each other.

IT, for its part, carries the infrastructure, security, governance and integration requirements that AI introduces, along with the technical expertise to manage them. But IT-led ownership tends towards technology-first thinking and slower experimentation, with more distance from frontline customer needs than either marketing or service.

None of these functions has the complete picture. Organisations that hand CX AI entirely to one department often end up building disconnected silos rather than a coherent strategy, a pattern that explains why so many otherwise well-funded initiatives struggle to deliver.

A more useful question than "who owns this" is "who is accountable for it". Many teams can and should contribute to CX AI. For outcomes to be tracked and problems escalated, however, one leader needs to remain answerable for results, even while execution stays distributed across functions.

This is not unique to CX. McKinsey surveyed roughly 500 organisations between December 2025 and January 2026 for its latest AI Trust Maturity research, and the pattern it found was stark. Businesses that had named someone specifically responsible for AI governance, whether through a dedicated role or an internal audit and ethics function, scored an average of 2.6 on McKinsey's maturity scale, well ahead of the 1.8 average among those that hadn't. McKinsey's Rich Isenberg has described the model many organisations are converging on as “centralized governance with federated execution”, in which central teams manage shared platforms and guardrails while individual business units decide their own use cases and build their own capability on top.

That gap is widening further as AI moves from assistive tool to autonomous decision-maker. A recent CCW Europe Digital report found that fewer than one in four enterprises have a fully centralised system governing AI in customer experience, with most still relying on partial or fully decentralised models that the report warns undermine consistency, transparency and accountability.

Increasingly, ownership is moving above individual functions altogether. Rather than sitting permanently with marketing, service or IT, AI programmes are landing with chief digital officers, chief transformation officers, COOs or dedicated enterprise AI offices, whose job is to coordinate AI investment and governance across multiple business units at once. This shift reflects the same logic McKinsey's governance research points to: accountability works best when it sits with a role built to look across the whole organisation, rather than one accountable for a single slice of it. It also raises another question of how CMOs and CX leaders should divide responsibility once AI sits above both of their functions.

Vendor platforms are reinforcing the same pattern. Salesforce has built Agentforce so that it sits underneath, rather than inside, any single cloud product, with the same agents drawing on sales, service, marketing and commerce data to bring automation into workflows across the wider Customer 360 system instead of staying confined to one department's view of the customer.

In practice, that means an agent handling a service query can pull in sales history or marketing engagement data without a human stitching the systems together, since the platform is designed to give agents the fuller customer picture and let them act inside whatever tool an employee already happens to be using. When the underlying technology is built to span functions by design, an organisational structure that assigns ownership to a single function starts to look increasingly out of step with how the tools actually work.

A practical structure is taking shape across organisations furthest along this path. The CX function defines customer outcomes, experience priorities and success metrics. IT enables the underlying platform through infrastructure, security, governance and integration. Marketing, service and other business functions identify opportunities and execute specific initiatives.

Governance is shared across CX, IT, legal, compliance and data teams, often through a cross-functional steering group responsible for prioritisation, funding and performance oversight, sometimes paired with a centre of excellence that sets standards while letting business units deploy locally.

Success, in the organisations doing this well, is measured against customer satisfaction, retention, revenue and efficiency, rather than departmental targets that quietly reward teams for optimising their own slice of the journey. For organisations putting this structure into practice for the first time, it helps to follow a structured approach to building a CX AI roadmap rather than improvising governance as projects scale.

The mistakes that derail this are familiar, such as treating CX AI as purely an IT project, leaving ownership undefined, letting departments build disconnected AI silos, ignoring governance as systems gain more autonomy, or measuring departmental success when customers experience the organisation as a whole.

Marketing, customer service and IT all have legitimate claims on CX AI, but none of them can succeed with it alone. Perhaps the question of ownership is becoming a little outdated, however. After all, the organisations getting the most value from CX AI are not deciding whether marketing, service or IT owns it. Instead, they are redesigning how those functions work together. In that sense, AI is becoming less another technology project and more a test of organisational design.