University chatbot governance: usability and accessibility

Introduction

Chatbots are becoming an everyday tool for supporting users, providing instant support, and aiming to streamline processes. These guidelines aim to ensure that chatbots deployed across the institution are user-friendly, accessible, and aligned with university standards.

Purpose and scope

These guidelines provide a framework for designing, developing, and maintaining chatbots that support students, faculty, and staff. They cover principles of user experience (UX), accessibility, conversation design, data privacy, and governance. They assume that user research has been undertaken that supports the need for a chatbot.

Key principles

User-centred design

  • Chatbots should be designed based on user needs and behaviours. Target audience must be clearly defined to tailor the chatbot's persona and dialogue to meet their expectations.
  • Provide intuitive and clear interactions that minimise confusion.
  • Allow users to navigate conversations easily, with the ability to correct errors.

Accessibility and inclusivity

  • Follow WCAG 2.2 AA standards for accessibility, as set out in the accessibility regulations that the University follows.
  • Ensure chatbot responses are structured for clarity and readability.
  • Ensure that screen readers, keyboard navigation, and voice input are as effectively supported as mouse or touch input
  • Use high-contrast text and avoid reliance on colour alone for meaning.
  • Ensure the chatbot functions correctly in different viewports and at different text sizes.
  • Consider that some users may take longer to complete tasks than others when deciding on timing out the users’ session.
  • Ensure that your ‘tone of voice’ is maintained when writing non-visual text, such as image alt text or screen reader prompts.
  • Ensure chatbot interactions are compatible with assistive technologies.
  • Provide accurate alternative text for images and captions and transcripts for multimedia content are provided in chatbot responses.
  • Enable voice interaction (speech-to-text) and text-to-speech options where appropriate.
  • Review Microsoft’s In Pursuit of Inclusive AI.

Transparency and ethics

  • Clearly indicate when a chatbot is being used.
  • Disclose chatbot limitations and provide a way to escalate to human support.
  • Avoid biases in AI responses and ensure inclusivity in content.

Consistency

  • Use a standardised tone of voice aligned with University communication policies.
  • Maintain consistency in chatbot behaviour across different platforms (e.g., website, mobile app).

Conversation design standards.

  • Use simple and concise language to ensure clarity.
  • Provide suggested responses or buttons where applicable to guide users.
  • Offer multiple ways to ask common queries (e.g., variations of "How do I apply?").
  • Allow users to rephrase or refine their queries without restarting the conversation.
  • Design chatbots to handle errors gracefully by offering alternative solutions (e.g., don’t blame the user if they make a mistake) User testing and feedback.
  • Prior to build, user testing must be carried out on wireframes and prototypes before build and deployment.
  • Regularly test chatbots for functionality, UX, and accessibility compliance.
  • Conduct periodic reviews to update chatbot knowledge bases.
  • Provide users with a mechanism to report issues and suggest improvements.

Conclusion

By following these guidelines, the university ensures that chatbots provide a seamless, accessible, and valuable experience for users. Regular updates and collaboration across departments will help maintain high standards in chatbot interactions.