Agentic AI Has Been Deployed. The High-Value Customer Journeys Are Still Manual.

More than half of global enterprises have deployed agentic AI. According to Infobip’s 2026 Customer Experience Maturity Report, that adoption rate reaches 53 percent globally. Yet the customer journeys with the highest revenue potential: loyalty management, complex onboarding, returns processing, and high-touch B2B interactions, remain stubbornly manual. The deployment is real. The deployment at the right level of the revenue stack is not.

Where Agentic AI Has Actually Landed

The 2026 CX Maturity Report surveyed enterprises across global markets. Adoption in Europe sits at 44 percent, below the 53 percent global average. But the more revealing number is not the adoption rate overall. It is where within organizations those deployments are concentrated.

Feedback collection leads adoption at 56 percent, followed by reminders and notifications at 52 percent and authentication at 45 percent. These are the low-complexity end of the customer engagement spectrum. They require little real-time data access, no multi-system reasoning, and virtually no human-judgment handoff. They are also, by definition, not the journeys that move revenue at scale.

By contrast, loyalty management has reached only 35 percent adoption. Delivery management sits at 28 percent. Customer onboarding, which is among the highest-value touchpoints in any subscription or B2B model, is at 26 percent. Product returns and refunds, which carry significant retention implications when handled poorly, sit at 15 percent.

The pattern is consistent: agentic AI has been deployed to handle the routine, the predictable, and the low-stakes. The complex, the high-context, and the high-value remain in manual or semi-manual workflows.

Why Data Architecture Is the Blocking Layer

Infobip’s research identifies the structural cause: fragmented data systems. Agentic AI agents require real-time access to centralized, current data to reason across a multi-channel customer journey. When that data lives in disconnected CRM systems, separate marketing platforms, siloed messaging tools, and unlinked order management records, the agent cannot build the context it needs to act appropriately on anything more complex than a single-variable trigger.

Only 27 percent of brands in the study report using a communications orchestration platform. Only 58 percent report fully synchronized communication channels. For the 42 percent whose channels are not synchronized, the agentic AI can process a single-channel interaction, but it cannot orchestrate a journey. The agent is powerful in isolation, but it has no visibility into what happened before or what should happen next across channels.

“CX Maturity will be the key differentiator between brands prepared to launch AI-powered journeys and ones that will struggle with scaling adoption,” said Ante Pamuković, CRO at Infobip.

What This Means for Revenue Operations Leaders

This finding arrives against a background of genuine revenue operations transformation. The current wave of agentic selling tools targeting specific revenue workflows has demonstrated that automation can work at the task level. The 2026 CX Maturity Report is a caution that task-level automation is not journey-level automation, and journey-level automation is where the revenue impact actually lives.

The sequence that produces revenue in a B2B context: prospect to qualify to onboard to expand to retain to recover, requires reasoning across time, context, and data sources. A 15 percent adoption rate for returns and refunds processing suggests that most agentic deployments have not yet reached the parts of the journey where the economic stakes are highest.

For revenue operations leaders, the strategic implication is that the AI deployment question and the data infrastructure question are the same question. Organizations that built clean, centralized data architecture before deploying agents will be able to push those agents up the complexity and value chain. Organizations that deployed agents first will find the ceiling relatively quickly, and the answer will be a data architecture project, not a new AI vendor.

The Untapped Revenue in the Gap

Loyalty management at 35 percent adoption is also, read differently, a 65 percent opportunity for competitive differentiation. Customer onboarding at 26 percent is a 74 percent untapped surface. The organizations that close the gap between simple-trigger agentic deployment and full-journey agentic orchestration will be operating on a materially different cost structure and customer retention curve than those that do not.

The data points toward one near-term solution path. Organizations that have already synchronized their communication channels, the 58 percent that can measure across touchpoints, are better positioned to push agentic AI into complex journeys. Synchronization creates the data context the agent needs. For revenue operations teams evaluating their next infrastructure investment, communications synchronization may deliver more agentic capability than a new AI vendor contract.

What Revenue Leaders Should Do Now

Three actions follow from the Infobip research for revenue operations and sales technology leaders.

First, audit where agentic AI is actually deployed against the revenue impact map. If the deployment is concentrated in notifications, reminders, and authentication, the question to ask is not whether AI is working but whether it is working on the right journeys.

Second, treat data synchronization as a prerequisite for the next phase of agentic deployment. The research is explicit that fragmented data is the blocking constraint. Investments in communications orchestration and CRM data quality will unlock more agentic capability than additional AI tooling on fragmented foundations.

Third, rebuild the agentic deployment roadmap around journey value rather than technical ease. The 15 percent adoption rate for high-value returns and recovery workflows is an organizational signal about priorities, not a technical one. The technology is capable of handling those journeys. The question is whether the data is ready to support it.

Source: Infobip