Only 23% of Firms See High Returns from AI in Customer Service

A new industry study finds that only 23% of companies report high returns from artificial intelligence deployments in customer service, according to reporting from ET BrandEquity. The finding suggests that despite widespread adoption of AI tools across service operations, most organizations are not yet seeing the performance gains they expected.

The research points to a gap between how broadly AI has been rolled out in contact centers and support functions and how much measurable value that investment is actually producing. The majority of firms surveyed fall short of high-return outcomes, indicating that implementation alone is not translating into the efficiency, satisfaction or cost benefits many had anticipated.

The study lands at a moment when AI in customer service has moved from experimentation to near-default status across industries, yet proof of return on that investment remains elusive for most. For leaders overseeing AI and digital transformation budgets, the figure is a signal to shift focus from deployment speed to deployment quality — measuring outcomes like resolution accuracy, customer effort and employee adoption rather than just tool rollout.

It also reinforces a pattern seen across other technology cycles: early adoption outpaces value realization until organizations redesign processes, retrain staff and rethink service journeys around the new capability, rather than simply layering AI onto existing workflows.

— 23% of firms report high returns from AI in customer service, according to the study.

A 23% success rate is not an AI problem — it is a service-design problem wearing an AI label. Most organizations are bolting automation onto broken or unexamined journeys and then wondering why the returns don't show up.

Technology never fixes a weak service model; it just makes the weakness faster and more visible. The firms stuck below that 23% line are almost certainly measuring AI by cost savings and ticket deflection, not by whether the customer actually got a better outcome. Before adding another model or vendor, operators should map the end-to-end journey AI is touching, define what "good" looks like from the customer's side, and redesign the handoffs between bot and human — because return on AI is really return on redesign.