The chatbot team existed for more than a year when I joined them. It consisted of a product owner, a Conversational UI Designer and a number of developers. There was already a lot of knowledge available, but there was still a need for physical design. The Conversational Designer indicated not to have time for design. The tricky thing about designing this chatbot was that it not only had to comply with the brand values of the company, it also had to meet the 'wow' factor. The first design that the Conversational Designer and I made in one afternoon was therefore rejected. The second attempt was made in a design sprint and contained almost all features, but still missed the 'wow' factor. It also turned out that not everyone thought the same about what the 'wow' factor should be and whether it should be in the design or in the micro animations. The third design gave me more freedom because they said that the Chatbot was a complete new product and therefore not designed yet.
Avatar considerations
For the Avatar design it was important that the visitor should not get the idea that it was a real person but a robot. Through research into other similar companies, I saw that various Avatars were used: an abstract avatar, an illustration, a picture of a person and a robot. An abstract symbol was found looked cold and a photograph of a person was too misleading. We chose a robot and an illustration as a starting point. We selected 4 Avatars, the company already provided. We also looked at a mix of young/old, male/female and various skin colours. An older man with glasses showed a more senior status while a young woman looked empathetic. New robot figures also have been made. Several tests were carried out on the work floor and a name for the bot was requested.
In the department where I worked, I had designed an identity in the form of a sticker with an illustration. The product owner asked me to add this illustration to the other illustrations I was investigating. In addition, the Conversational UI Designer and I decided to research both types of avatars in a poll with the last question which the respondents preferred; an illustration or a robot. It became clear that the previous illustration had a small majority of 51% of the votes preferred an illustration instead of a robot.