Vermeer introduced a support chat feature with the launch of their new customer and dealer portal. This was a big step for the team, as support scenarios had previously only been handled through phone calls and emails. As the userbase steadily grew, the customer support team started to feel the pains of managing different contact channels. The team approached UX and asked if there was any way to streamline the process and make chats less disruptive.
I started by looking through historical chats and documenting the typical experience for users.
Several areas of improvement were highlighted during this activity. Two points that stood out in particular were the opportunity for automated processes for repetitive steps and gathering complaint data upfront to reduce agent hand-offs. I arrived at the idea of creating a basic bot to help gather initial chat details and direct traffic.
I knew I would face some headwind suggesting a bot – there was a general apprehension towards bots amongst our customers and within the company. I was often reminded that “customers don’t want a bot” and that “they would rather speak to a real human.” While I recognized these hesitancies, I also couldn’t forget about the humans that were supporting the chat and struggling to keep up. Customers didn’t not want a bot - they just didn’t want an unhelpful bot.
I sorted through the history of chats our team had assisted with thus far and identified the common patterns. This helped me define the base logic and workflow of the bot.
The end result was what I dubbed a “dumb bot” – AI agents were on the rise, so I needed to make a distinction between the two. I wanted to illustrate to our internal team that this bot had been manually created with very strict logic and was not the open-ended bot they were worried about. I pitched the bot as an additional member of our support team - one that helps automate the chat intake process. The traffic bot helps gather initial complaint data and directs traffic to the appropriate team.
