Stop automating pain: why AI won't fix your customer service issues.

The story of the past three years in customer experience design has been artificial intelligence. No other subject comes close. The book I published earlier in 2025 featured AI in almost every chapter. The message in 2023 was that large customer service departments can be almost entirely replaced with AI. But as one company after another struggled to scale pilots up to the enterprise, it soon became clear that AI is powerful, but it needs time, expertise, and effort. This isn’t an out-of-the-box solution for anyone.

Customer service teams are still large. In many cases, this is true even where they have adopted new technologies and automated sections of the customer journey. Operational efficiency does not look the way the technology suppliers promised.

Companies that have adopted AI in their CX processes have not closed all their contact centers. In many cases, customer engagement volume has increased. More complex questions, more channels, and more proactive engagement from brands have pushed the number of customer interactions up. None of this is what the AI companies were telling us in 2023. The transformation was going to be almost immediate - and simple.

Our unconscious choices around CX could be the main blocker. Executives are rapidly deploying AI with the hope that efficiency will be improved, yet the problems are almost certainly elsewhere in the business, or in the process of how the company operates. Think about customer call volume. Instead of buying more and more technology to try automating customer interactions, we should really be asking why all these customers need to contact us in the first place. 

Taking this approach will naturally shift the focus away from the customer service team. If the customer problems have a theme or recurring pattern then that is where you need to focus. There may be unresolved issues around quality or late deliveries that need to be addressed. You are not staffing to meet customer demand, you should be able to eliminate the demand itself.

Anyone with a background in consulting will recognize this approach as root-cause analysis. By taking a systematic approach to how a business is operating, you can determine where problems or mistakes are being introduced.

Let’s go further with this CX example. Why are your customers calling so often? By exploring the underlying reasons for the calls, we can trigger changes that should reduce the number of times customers need to contact.

Some ideas for where to explore might include:

• Why are customers ignoring or bypassing the self-service option that is already available? Many customers hate using a self-service system, but if the majority of your customers insist on speaking to an agent, does that mean your self-service system could be improved?
• Which product features, policies, or billing issues generate the highest volume of questions and complaints? You may find that a small number of issues are generating the majority of customer questions.
• Is your pricing structure, billing statements, or subscription tier matrix so complex that customers often need to talk to someone just to figure out what they are paying for?
• How do small internal changes - such as a billing date change - lead to high volumes of calls where every customer asks ‘Why was I charged on this date?’ Potentially, with the need for chargebacks too.
• Do your agents need to switch between multiple systems to get a single view of the customer? An agent should have a complete overview, rather than needing to login to the CRM, inventory, and billing databases all on different screens. Not only does this introduce errors (which leads to more complaints), but also the agent is forced to work slowly and AHT is needlessly increased.
• Do we have any company policies that are significantly different to our competition? For example, refund terms, warranty claim terms, delivery windows… are these causing a constant stream of customer complaints because their expectation differed from what was in our small print?
• Is your marketing and sales team making aggressive promises that the company cannot deliver on? If they are promising services or features that don’t exist now, “but are coming next week,” just to close a sale, then your sales team is actively driving up the number of customer complaints.

Framed like this, these ideas don’t seem like rocket science. There is no hidden wisdom. This is basic common sense, but it is illustrative of what has been understood for decades in manufacturing. A small problem early in the supply chain will cause a much bigger problem later on.

Car recalls are a very clear example of my point. In 2024, there were almost 28 million vehicles recalled by their manufacturer in the USA alone. Imagine the efficiency that could be generated if you identified more problems before the cars were delivered to customers. This is our problem when designing customer service processes. We often design how best to deliver a great customer experience rather than believing that the best customer experience is when the customer does not need to call a brand for help.

As we go into 2026, it is worth repeating this message about unconscious choices. Stop and think about what is driving your actions. Are there changes that can be applied inside your business that would quickly reduce the need for customers to seek help? If so, address these issues before running any more technology pilots. Fix the underlying problem, not the symptom.