Persona Design
Every chatbot has a voice — and therefore, should have a personality. As humans, we’re hardwired to project human traits onto everything, so whether businesses like it or not, users will still assign a personality to their bot even if you have purposely designed one. As a Product Manager for Talkpush's conversational interfaces, I made it best practice to design it before writing or designing any dialogues.
This bot persona represents the brand through its language, tone, and word choice.
To adapt to our customers’ brands, I created a new bot offering called "Persona bot" in addition to the standard bot product. This plan allowed them to choose from a series of “persona bots” designed with a particular set of personality traits in mind. Through machine learning, we trained our Persona bots to respond in a unique way, depending on their personality. For example, an efficient and goal-oriented recruiter like Stan likes to get straight to the point when communicating, answering candidate inquiries in a professional and detailed manner.
Strategic approach
For one of our biggest clients, Transcom, I started by studying and examining their mission statement and existing communication channels, to make sure any personality we came up matches with the goal of their brand. Transcom already had a strong recruitment marketing identity and a recognized mascot in place, called Ces, so it was only natural to digitize that into a personality when redesigning their recruitment chatbot.
Backed by research
To begin, a detailed questionnaire was given to their recruitment team to help us understand their company’s target audience, image, and “tone of voice” to make sure Ces bot was a true embodiment of all those values. Then a workshop was conducted to brainstorm ways on how to convey that “Transcom” feel into the candidate-chatbot exchange.
From there, I came up with a list of adjectives for this "person".
Humorous
Caring
Witty
Upbeat
Conscientious
Kind
Sincere
Everything from the bot avatar to the greeting message and content throughout the conversation centered on this personality.
A key challenge in the project was getting candidates to complete their prescreening interviews. After this rollout, we saw a 20% increase in completion rate. We also saw a significant increase in the chatbot's sentiment score -- the mood of applicants by decoding verbal and sentence structure clues.