Summary

• Talkdesk's EMEA research finds that 30% of surveyed organisations use AI to resolve more than 40% of customer issues, while 17% report no AI-resolved issues.
• Only 35% report integrated customer data, and 22% consistently measure the business impact of AI.
• The findings are based on 60 senior respondents surveyed by NewtonX in April 2026, limiting how widely the results can be generalised.

Businesses across Europe, the Middle East, and Africa are achieving markedly different results from customer service automation, with research from Talkdesk identifying substantial gaps between organisations already resolving a large share of customer issues through AI and others that have yet to achieve automated resolution.

The company’s State of Customer Experience Automation in EMEA report, published on 3 November, found that 30% of surveyed organisations used AI to resolve more than 40% of customer issues, while 17% said the technology did not resolve any.

The findings come from an April 2026 online survey conducted by NewtonX. Its regional sample comprised 60 director-level and more senior decision-makers, or people with significant influence over relevant decisions, at organisations employing at least 250 people.

The relatively small sample is important when interpreting the results. The percentages describe the organisations surveyed rather than providing representative measurements of adoption across every industry or country within EMEA.

Automation progresses at different rates

Customer service AI can perform several distinct functions, from retrieving information and helping employees prepare responses to resolving customer requests without direct human intervention.

These applications involve different levels of operational integration. A system that answers general questions using a knowledge base may require relatively limited access to business applications, while completing a transaction could involve customer records, identity verification, authorisation, and changes to operational systems.

The proportion of enquiries handled automatically therefore depends partly on the types of requests customers make and how much responsibility the organisation permits software to exercise.

Talkdesk’s regional findings show a greater concentration of organisations at the extremes of reported resolution than in North America.

In the North American survey group, 67% of respondents said AI resolved no more than 20% of customer issues, while 8% reported resolution of more than 40%.

EMEA also contained organisations making limited use of automated resolution, but a larger proportion reported handling substantial volumes of customer issues through AI.

The comparison suggests that organisations may introduce similar technologies while integrating them into customer operations to different degrees.

However, an automated resolution rate does not independently establish service quality. Completing a larger proportion of enquiries without employees may be useful, but the outcome also depends on accuracy, customer satisfaction, repeated contact, and whether unresolved cases reach the appropriate team.

Organisations handling relatively straightforward enquiries may be able to automate a larger proportion than those dealing with complex financial, technical, or regulated services.

Data integration limits what AI can complete

Talkdesk found that 27% of EMEA respondents identified legacy technology among their largest barriers to AI adoption, compared with 12% in North America.

Established customer operations often rely on several applications covering account information, orders, payments, service history, and specialist business processes.

AI systems may be able to answer a general enquiry without accessing all those applications. However, completing a request that concerns a particular customer usually requires accurate information from the relevant systems and permission to carry out the necessary action.

For example, explaining a company’s refund policy is different from determining whether an individual purchase qualifies for a refund and initiating the corresponding payment.

The second task may involve checking the customer’s identity, retrieving transaction information, applying current commercial rules, and recording the resulting decision.

Where those activities remain distributed across disconnected applications, an AI assistant may still depend on employees to complete parts of the process manually.

Only 35% of surveyed EMEA organisations reported integrated customer data, suggesting that many respondents were pursuing automation without having consolidated the information required for more extensive operational tasks.

Data integration also involves governance considerations. Organisations need to determine which information an automated service can access, whether it is current, and what restrictions apply when data concerns sensitive or regulated activities.

Connecting systems can create new opportunities for automation, but it also requires dependable access controls and procedures for managing errors or unexpected results.

Employees take on more complicated cases

As AI handles more routine interactions, the remaining work for customer service employees can become more demanding.

Among EMEA organisations using AI agents, 59% reported that human employees were handling more complex cases. Yet only 13% said they dynamically coordinated the work of AI systems and employees through orchestrated handoffs.

The difference reflects the distinction between removing some straightforward enquiries from a queue and redesigning the workflow surrounding the cases that remain.

A customer may begin an interaction with an automated assistant before encountering a problem requiring specialist attention. The system then needs to transfer the relevant context, including what the customer requested, which information has already been checked, and whether any actions have been completed.

Where that information is not passed on reliably, employees may need to repeat questions or reconstruct the case before providing assistance.

Such interruptions can increase handling time and frustrate customers even when the automated part of the interaction operated as intended.

Coordinating the two forms of service also requires clear boundaries. Some actions may be suitable for automatic execution, while others require an employee to review the circumstances or approve a decision.

The research found that 50% of EMEA respondents identified job security among their leading concerns about people working alongside AI, while 47% cited AI accuracy.

These findings indicate that workforce questions remain closely connected to the technical challenges of deployment. Employees need suitable training and operating procedures as their responsibilities change, particularly when more complicated enquiries become a larger part of their workload.

Performance measurement remains limited

Only 22% of EMEA respondents said their organisations consistently measured the business impact of AI.

That limits the information available for assessing whether a deployment is improving operations beyond increasing the volume of automated activity.

A company may be able to count the enquiries handled by an AI system without knowing whether customers subsequently contacted the organisation again, whether an automated decision required correction, or how much the service cost to maintain.

Measuring outcomes requires information about accuracy, resolution quality, customer experience, employee workload, and the resources used to operate the technology.

These measures may also need to distinguish between different types of enquiry. An automated system resolving basic account questions could perform well against one set of indicators, while a service handling sensitive financial transactions might require additional assurance.

Talkdesk found that 67% of EMEA respondents expected AI to resolve a greater proportion of customer issues within two years.

Those expectations suggest continued interest in expanding automation, but they are forecasts from respondents rather than outcomes already achieved.

Similarly, 62% expected AI to help their organisations expand without increasing headcount, and 40% anticipated improvements in employee satisfaction as human agents moved towards more demanding work.

The eventual results will depend on how organisations manage implementation, workforce requirements, and the quality of automated interactions.

Operational controls shape further adoption

Talkdesk’s research indicates that expanding AI beyond isolated activities requires greater coordination between information sources, business applications, and employees.

An organisation may have introduced an AI agent yet still lack the infrastructure needed for that software to complete a customer request across several systems.

Integrating additional applications can increase the range of tasks available for automation, while also making testing, permissions, audit records, and exception handling more important.

The systems must remain accurate as business processes change. A revised returns policy, updated customer verification procedure, or new product can affect the information and actions an automated service requires.

Pedro Andrade, global head of AI at Talkdesk, has identified access to reliable knowledge and systems, coordination with employees, and consistent measurement as important foundations for wider automation.

The company has published a regional report and maturity assessment based on findings from its wider global research programme.

The survey included respondents from financial services, retail, technology, life sciences, and other industries, but its 60-person EMEA sample does not support precise conclusions about individual countries or every type of organisation.

The responses are also self-reported. They do not provide independently audited figures for customer satisfaction, costs, or the accuracy of automated resolutions.

Nevertheless, the findings identify concrete areas where surveyed organisations report differences between AI deployment and operational readiness. Customer data integration, reliable handoffs to employees, and consistent measurement will influence whether future automation improves service delivery as organisations extend its use.