Letting Users Describe With Words: A Human-First Approach to Voice-Based Deviation Reporting

Sammanfattning

Deviation reporting is a critical component of Quality Management Systems (QMS) in industrial environments, yet underreporting remains a persistent challenge. Workers face barriers including time pressure, physical constraints, cognitive complexity, and form-based interfaces that fail to accommodate the realities of production work. This thesis explores whether voice-based interaction can reduce these barriers by allowing workers to describe deviations in their own words, rather than navigating structured digital forms. Following a Design Thinking framework, this study conducted semi-structured interviews with five quality management professionals to identify pain points in current reporting practices. Further developed and evaluated three conceptual approaches to voice-based reporting, and produced a high-fidelity prototype of the selected concept: a conversational mobile reporting flow supported by a digital AI colleague. The prototype was evaluated through a moderated usability test with five participants using the System Usability Scale (SUS) and four project-specific success criteria. Results indicate strong user acceptance of the voice-based approach. The prototype achieved a mean SUS score of 90.0, rated as Excellent, and received positive ratings across all four success criteria: Speed & Efficiency (4.2/5), Simplicity & Natural Language (4.4/5), Accessibility & Mobility (4.0/5), and Feedback & Confirmation (4.6/5). Participants found that speaking reduced cognitive load and allowed for more natural self-expression compared to manual form entry. Key areas identified for further development include context-adaptive dialogue design, handling of ambiguous input, and real-world testing in noisy production environments. The findings suggest that voice-based deviation reporting is not only viable but represents a meaningful improvement over current methods for users who find manual input to be a barrier to consistent reporting.