A new report from Trustpilot reveals an "AI communication gap" that leaves insurance customers frustrated and more likely to leave negative reviews. For customer support teams, the findings show that failing to provide clear paths to human agents can cost a company as much as 1.6 stars off its rating - a drop that directly affects consumer trust and business reputation.
The Trustpilot Insurance Insights Report 2026 analyzed more than 145,000 consumer voices in the US insurance market. The data shows reviews that mention AI are dramatically lower rated than those that do not. The average star rating for reviews that reference AI was 2.47, compared to a 4.35 average for reviews that didn't mention it. Taylor Cunningham, Trustpilot's vice president of U.S. marketing, said the disconnect often comes down to customers feeling trapped.
"The most recurring themes are customers who were stuck in a loop, couldn't get on the phone with a real person and were kind of passed off by AI chatbots or AI automations without being able to speak to a real person," Cunningham said.
The quantifiable cost of ignoring a complaint
The report found that when a customer felt a complaint was ignored - whether by automated systems or human oversight - the company's rating dropped by an average of 1.6 stars. Cunningham pointed out that even that seemingly small number makes a sharp difference in how consumers choose a provider. "There's a huge difference between a five-star review and a 3.4-star review. That's quite a big hit to experience for ignoring a customer complaint," she said.
Importantly, the negative impact wasn't driven by the existence of a problem. It was the failure to resolve it that damaged the review score. The report also noted that insurers that didn't offer human escalation paths after the January 2025 California wildfires faced a higher risk of permanent reputational damage. Support teams in any industry dealing with crisis moments should take notice.
Where AI falls short in customer service
The report's AI sentiment analysis, covering January 2021 through December 2025, found that consumers are open to AI when it improves efficiency. But they become sharply critical when AI stands between them and a real person. Cunningham noted that customers expect dignity, respect, and human connection - not flawless perfection. When a business fails on that front, the damage to its reviews is hard to reverse.
For people in customer support roles, this is a clear signal. AI for Customer Support can handle routine inquiries quickly, but the moment a conversation requires empathy or a complex decision, a human agent must be accessible. The data shows customers don't object to AI in principle. They object to being stuck in a system that won't hand off to a real person.
Balancing efficiency and human connection
Cunningham's advice to insurers - and by extension, any support organization - centers on three principles:
- Never let AI operate as a black box, especially in claims processing or complex service tasks.
- Always provide a clear, accessible escalation path to a human agent.
- Increase transparency when AI is making or communicating a decision.
These guidelines do more than protect star ratings. They create an advantage for businesses that handle the balance well. "AI should empower agents to resolve issues faster, know more about their customers, so they can act faster and be more personalized. But AI should not act as a barrier that keeps customers away from human support," Cunningham said. Many AI Agents & Automation tools can route queries intelligently, but the design must keep the human handoff visible and simple.
Why this matters for customer support
The Trustpilot data shows that the negative sentiment around AI in service interactions isn't about the technology itself. It's about poor implementation. Support teams that design their workflows so that automation speeds up resolution - not buries it - will see better reviews and higher retention. The report suggests that even a single ignored complaint can undo months of positive customer experience work. For support leaders, that means every chatbot script and every IVR tree should be evaluated by one question: does this make it easier or harder for a customer to reach someone who can help?
Cunningham put the opportunity plainly: "This is a great opportunity for businesses that prioritize the customer experience to really stand out." In other words, getting the human element right isn't just a cost avoidance measure - it's a competitive differentiator.
Your membership also unlocks:
