Citizens Advice, the frontline charity that helped more than 2.7 million people in England and Wales last year, has looked closely at what happens when essential service companies push customers toward chatbots instead of humans. The findings are stark. Nearly half of people who used a chatbot found it unhelpful, and many described a system that felt built to wear them down rather than solve anything.
The numbers behind the frustration are worth sitting with. Of people who struggled to reach a human, 49 percent felt stressed or frustrated, 44 percent faced delays, and 14 percent simply gave up trying to fix the problem at all. For something as basic as correcting a wrong energy bill, the charity found that a working online system takes under an hour, while going without good digital access can stretch the same task out to six weeks.
This gap matters because it hits the same people twice. Citizens Advice estimates that at least 14 million people in the UK are digitally excluded, whether from age, disability, poverty, or simply not owning a laptop or smartphone. When a company removes phone lines and staffs the front door with a chatbot, it saves money on the customers who can adapt and quietly punishes the ones who cannot.
Companies did not add chatbots by accident. They are cheaper than paying staff to answer phones, and boards have been told for years that automation is the fix for rising service costs. Klarna became the poster child for this idea in 2024, publicly cutting customer service jobs and handing the work to an AI assistant. By last year, the company was rehiring humans, a quiet admission that the savings looked better on a slide than they worked in practice.
Britain is not alone in noticing the problem. Dutch regulators told companies in October that chatbots cannot fully replace a human option, and the European Commission has flagged that consumers currently have no general legal right across the bloc to insist on speaking to a person when a chatbot fails them. That gap is exactly what Citizens Advice is now asking UK service providers to close voluntarily, before a regulator forces the issue.
For any business running a chatbot as its main customer support channel, the lesson is not that AI support is bad. It is that AI support without a working human escape hatch creates a cost that does not show up on the balance sheet: lost customers, reputational damage, and now, increasingly, regulatory attention. A chatbot that resolves a password reset is fine. A chatbot that stands between a person and their heating bill, their bank account, or a homelessness application is a business risk dressed up as a savings plan. Companies that build a fast, visible way to reach a human when the bot fails will be the ones customers stick with when the next provider switch becomes easy.