Gartner released a new survey on September 2, 2026, finding that only 27% of customers would be willing to try a chatbot again after a negative experience. The research firm described this as a leaky bucket problem: one unsuccessful chatbot interaction is enough to permanently drive a customer away from AI-powered support channels.
The finding comes as companies worldwide accelerate AI deployments in customer service, with Gartner's own previous research showing 91% of customer service leaders face executive pressure to implement AI in 2026. However, the new data suggests that poor chatbot quality could undermine the very investments companies are making.
The survey adds to Gartner's growing body of research showing tension between corporate AI ambitions and customer expectations. Previous Gartner surveys found that customers are three times more likely to use third-party GenAI tools than company-provided chatbots, and 87% of customers say companies using GenAI for customer service must provide access to a human agent.
The leaky bucket dynamic means that every negative chatbot experience doesn't just fail to convert that specific customer — it actively damages the company's broader AI strategy by eroding trust. For enterprises investing millions in AI customer service platforms, the 27% retry rate represents a critical quality threshold that most deployments may be failing to meet.
