Your Chatbot Is Not Failing Technically. It Is Failing Socially.

India's AI story is often told through the lens of scale. We celebrate the billions of UPI transactions processed every month, the rapid digitisation of financial services, and the growing use of artificial intelligence across banking, healthcare, telecom and retail. Increasingly, the first representative of an organisation that customers interact with is not a human employee but an AI-powered chatbot.

For businesses, the attraction is obvious. Chatbots promise lower costs, faster response times, 24-hour availability and the ability to handle millions of customer interactions simultaneously. As generative AI becomes more sophisticated, organisations are rushing to embed conversational agents into virtually every customer touchpoint.

Yet amid this enthusiasm, a critical question remains largely overlooked: Are organisations measuring chatbot success using the wrong criteria?

Most firms evaluate chatbots as technical systems. They ask whether the bot understood the query, resolved the issue, reduced waiting times or lowered service costs. These metrics matter. However, they capture only part of what customers actually experience.

Customers often evaluate service interactions through a very different lens. They ask whether they felt heard, respected, reassured and supported. In routine interactions, such as checking an account balance or tracking a delivery, this distinction may not matter greatly. But when something goes wrong, the social dimension of service suddenly becomes far more important.

This issue is becoming increasingly relevant in India. In March 2025, Reserve Bank of India Governor Sanjay Malhotra urged banks to leverage artificial intelligence to improve grievance redressal and customer service. The call was prompted by a striking reality: during FY2023-24, India's 95 scheduled commercial banks received more than 10 million customer complaints. The challenge facing organisations is no longer whether AI can be deployed at scale. The challenge is whether AI can effectively support customers during moments of frustration, uncertainty and vulnerability.

Consider a customer whose digital payment has failed, a patient attempting to reschedule an urgent medical appointment, or a telecom subscriber struggling with a prolonged network outage. In such situations, customers are not merely seeking information. They are seeking reassurance, accountability and confidence that their concerns are being taken seriously.

This is where many AI systems encounter difficulties.

Our research suggests that chatbot failures are often interpreted by customers not as technical breakdowns but as social experiences. Customers frequently judge service failures based on how the interaction made them feel rather than solely on whether the issue was eventually resolved.

As Dr. Abhisek Kuanr, Lecturer in Marketing at the University of Essex, explains:

"Many organisations view chatbot failures primarily as technology failures. Our findings suggest that customers often experience them as social failures. When consumers feel ignored, dismissed or unsupported during a service breakdown, the damage extends beyond the immediate interaction and can affect their broader relationship with the brand."

This distinction matters because customers do not separate the technology from the organisation deploying it. A chatbot may provide a technically correct answer and still leave a customer feeling frustrated. It may follow the programmed workflow perfectly and yet create the impression that nobody is listening.

Evidence from broader consumer research supports this view. A global YouGov study found that 54% of consumers believe the company using a chatbot should be held responsible when the chatbot provides incorrect information, while only 26% primarily blame the chatbot developer. In other words, customers hold brands accountable for AI failures, regardless of whether those failures originate from the technology itself.

Our study further found that customer reactions become particularly negative when service failures occur in high criticality contexts. The tolerance customers may show toward an AI assistant helping them find a restaurant reservation is very different from their expectations when dealing with healthcare appointments or other high-stakes service situations. In these contexts, customers expect not only competence but also empathy.

According to Dr. Debasis Pradhan, Professor at XLRI - Xavier School of Management:

"Many firms still approach conversational AI primarily as an efficiency tool. Customers, however, evaluate it as part of the relationship they have with the brand. During moments of service failure, communication style, empathy and responsiveness become just as important as problem resolution."

This insight has important implications for business leaders. The future of AI-enabled customer service will not be determined solely by advances in algorithms or language models. Those technologies are rapidly becoming accessible to everyone. The more significant competitive advantage may lie in designing AI systems that understand the social realities of customer interactions.

This does not mean organisations should abandon automation. Quite the opposite. AI will continue to play a critical role in scaling customer service operations. However, firms must recognise that not every customer interaction is simply an information exchange. Some interactions involve anxiety, uncertainty, frustration or urgency. In those moments, customers are looking for reassurance as much as resolution.

The next frontier of conversational AI, therefore, may not be artificial intelligence alone. It may be social intelligence.

The firms that succeed will not necessarily be those with the smartest chatbots. They will be those that recognise a simple but powerful truth: when customers experience service failures, they are not only evaluating whether the technology worked. They are evaluating whether the organisation cared.

In that sense, many chatbots are not failing because they lack intelligence. They are failing because they have not yet learned the social side of service.

References

Kuanr, A., Moharana, T. R., Yan, M., Pradhan, D., & El-Manstrly, D. (2025). When chatbots cause trouble: How agent type and conversation style shape consumer responses to service failures. European Journal of Marketing, 59(12), 2788–2839.

Reuters (2025). India's central bank governor calls on banks to adopt AI to address consumer complaints.

Bank for International Settlements speech archive: Sanjay Malhotra, RBI Governor, "Transforming grievance redress – the AI opportunity"

YouGov (2024). Consumers hold companies responsible for AI chatbot errors