Job seekers build their skill profile by classifying their skills based on a Likert scale ranging from “I never used this skill” to “I always used this skill.” Getting an accurate understanding of a job seeker’s life experience is critical to making SkillLab successful so classifying skills with the app must be fast, easy to understand, and engaging.
We tested many UI options for classifying skills on a one-to-five scale including radio buttons, a slider, and a slider plus a bar chart. We also tested a binary approach with yes/no buttons and swiping left/right, but that hurt the AI reward function. In the end the “slider plus a bar chart” approach offered the best tradeoff between speed and accuracy. We eventually removed the slider and let people simply tap the bar chart which made skill classification faster without affecting data quality.