Back in 2019, we were designing the first iteration of an eCommerce solution based on our core platform for a well-known automotive brand.
One of the most exciting features we added to the prototype was a chatbot capable of helping during all the tasks needed to find and quote a vehicle. The most challenging part was training the software; the engineers did an outstanding job connecting the bot with our services.
We performed several usability testing on the prototype, and it was immediately clear that users didn't interact with the bot, preferring a more traditional way of discovering and searching for a new vehicle. When we asked them to do the same tasks with the bot, they were faster and more efficient, yet the majority reported that they still preferred the traditional journey.
(Adjusted to be a bit more visually pleasant)
In a nutshell, they liked the efficiency but felt it was a too much-guided process. We came to the conclusion that, while the tool satisfied the behavioural level, it frustrated the reflective level.
We started exploring another option: using the "Chatbot" shape for linear tasks that are generally less engaging but emotionally dense, like the reservation journey or the trade-in journey. Currently, we implemented only some of the solutions available within the concept, but we tested it as a prototype in multiple scenarios and multiple journeys.
In general, there is a higher success rate and a lower "form fatigue".This solution is not the preferred choice of "expert users" who prefer a highly dense form and the ability to navigate it with the keyboard. It's less than a quarter of our users, however, it's important to give the option to perform the task with a traditional UI.
One of the prototype we tested
Some of the animations/microinteraction are not spot on
The most important bit, the one that makes the difference, is the shape of the content, the use of a language the user is familiar with, and sentences that do not need to be processed because their "shapes" are easily recognisable.
It also allows presenting the information when needed, avoiding the possibility of information overload and making it possible to offer onboarding within the journey without making the interface too busy.Last but not least, it allows the user to concentrate on a single micro-task, reducing errors.
There are cons. Some users felt the process to be slower than the traditional one, and some struggled with the positioning (they are used to having chats in more confined spaces, and it worked better on mobile than on desktop).
It's important to note that most of those tasks will be performed by the users once every 2/3 years, and most users are not familiar with the relatively new process of buying a vehicle online. This approach may not be ideal for tasks that need to be performed more often and are more likely to be learned over time; it could be frustrating in the medium/long term.