As AI solutions spread, more and more companies are building chatbots into their customer service processes. These tools can be very effective when they are configured well and run under proper supervision. But as a recent case showed, these systems need to be handled with care.
The case study…
While testing the customer service chatbot of an international company, we noticed that it was the first and only customer service channel. The goal in this situation was to return a faulty product, but the courier service’s system did not accept the return code for the package. When we tried to resolve this, the chatbot could not tell where the process had stalled, and kept repeating the same information, which did not help us move forward.
At this point, the conversation went like this:
- Chatbot: „Here is your return code: XYZ123.”
- User: „The courier service’s system does not accept this code.”
- Chatbot: „According to our tracking system, the product was handed over to the courier service on 30 September 2024.”
We were able to get back to exactly the same point every time, so there was no way to solve the problem. Because there was no option to escalate, the customer service chatbot could not provide any meaningful help.
…and the lessons
Integrating AI into customer service undoubtedly holds great potential. However, it is important to monitor how chatbots perform on a regular basis, especially during the rollout period. If this is neglected, the result can be dissatisfied customers and unresolved problems, and the company may miss the chance to spot system errors and process gaps in time.
By avoiding these mistakes, you can make sure that chatbots work as an effective, customer-focused solution: they complement the existing human customer service and help and support its staff.