PeopleImages.com – Shutterstock.com
Customer service is considered one of the areas in which artificial intelligence can be used particularly effectively. AI chatbots that respond to customer inquiries around the clock or AI assistants that support service employees with routine tasks promise better response times, higher customer satisfaction and noticeably more efficiency.
However, the initial euphoria about AI in customer service is increasingly giving way to a more realistic look at its actual use and added value, reports Roland Berger. Success depends less on technology. It is crucial to change business models, processes and structures and to ensure data quality.
Reality Check nach dem KI-Hype
As part of the current study “When the hype fades, reality hits” (access for data), the management consultancy surveyed a total of 550 customer service managers in ten countries. A clear turning point emerged: the experimental phase is over, what counts now is reliable implementation. Companies that already use AI productively in customer service realize measurable advantages in quality, speed and costs.
According to the survey, AI in customer service has even exceeded the expectations of current users: More than half of the companies with AI-supported customer service report a significant or even very significant positive impact on their service processes.
The effect is particularly evident in the operational key figures. On average, AI shortens response times by 19 percent and increases process efficiency by 11.5 percent. At the same time, operating costs fall by an average of 11.7 percent.
Customer feedback on AI-based interactions is also more positive than many expected. The study found that 80 percent of customers react positively or very positively to AI-based bots. According to Roland Berger, this confirms that AI is no longer just a backend optimization tool, but is quickly becoming a central part of customer interactions.
Scaling fails due to the requirements
Despite the progress, scaling AI in customer service remains the biggest construction site. Many companies currently only have a medium level of maturity in key requirements such as data availability, data quality, integration or governance. Even among AI users, legacy systems, unclear responsibilities and fragmented process landscapes are slowing down the broader rollout.
According to Roland Berger, the central question is no longer whether AI works in customer service, but rather whether companies are consistently modernizing their business model enough to increase the benefits in the long term. They would also have to make the transition from isolated use cases to a new way of handling customer service.
This realization has apparently also reached most companies. You now have a more realistic assessment of what AI actually means in customer service: While around 95 percent of those surveyed in last year’s study said they used AI in customer service, this year the proportion fell to 54 percent.
However, AI is still as relevant to customer service as ever. According to the survey, most companies that are not yet using AI are planning to introduce it in the next twelve months or are very likely to do so. Their expectations focus on greater process efficiency and lower operating costs, while also striving to improve customer satisfaction. However, another finding shows that they are slightly less enthusiastic about what AI can actually achieve than companies that are already using it.
“The market is entering a new phase, away from pilot projects with a signaling effect and towards scalable applications with clear operational responsibility,” explains Simone Schatto, Director at Roland Berger. “This is exactly what will determine in the coming years who will simultaneously improve efficiency and customer experience in customer service. At the same time, the pressure will increase on those companies that are currently foregoing the holistic use of AI in customer service.”
