Safely manage your Zendesk from the AI assistant you already use, via the Deltastring MCP. Beacon configuration platform
← Back to news

Klarna Scales Back AI Customer Service After Quality Issues

Klarna's decision to rehire human customer service staff after eliminating approximately 700 positions between 2022 and 2024 exposes a critical miscalculation in how rapidly AI can replace human judgment in customer-facing operations. The fintech's aggressive automation strategy, designed to reduce costs, instead triggered a cascade of service degradation and customer complaints that CEO Sebastian Siemiatkowski has now publicly acknowledged. What makes this reversal particularly instructive for CX teams is the disconnect between operational metrics and customer experience outcomes: Klarna achieved 25% revenue growth and expanded to one million connected merchants globally, yet its share price collapsed 56% since its September 2025 IPO, signalling that investors view service quality failures as a fundamental threat to long-term viability. This suggests that cost savings achieved through wholesale AI replacement are being rapidly eroded by churn, reputational damage, and the very real costs of rehiring and retraining staff.

The company's pivot toward a hybrid model—integrating technology more tightly with human employees rather than replacing them outright—reflects a pragmatic recognition that certain customer interactions demand contextual reasoning, empathy, and exception handling that current AI systems cannot reliably deliver at scale. For teams already operating within platforms like Zendesk or Freshdesk, Klarna's experience raises an uncomfortable question: are you measuring the right metrics? Revenue and ticket volume throughput can mask deteriorating CSAT and effort scores until reputational damage becomes visible to shareholders. Klarna's trajectory also demonstrates that the cost discipline narrative—the company still projects workforce reduction from 3,000 to 2,000 by 2030 through attrition—rings hollow when immediate service failures demand immediate human intervention.

The broader implication for CX professionals is that the binary choice between automation and human labour is a false economy. Klarna's case study validates what emerging research suggests: customers increasingly expect blended support models rather than pure AI or pure human channels. The fintech's struggle to scale its U.S. credit business efficiently whilst simultaneously fixing service issues reveals that automation decisions made without adequate pilot testing, quality benchmarking, or customer feedback loops create compounding operational debt. For support leaders evaluating AI-driven tools or considering aggressive headcount reductions, Klarna's 56% stock decline is a more honest measure of success than its revenue figures.