The university has received comments from some users of its website that the content is not always relevant to them. To improve engagement, it has been decided that information will be gathered that will support a more personalised user experience for all visitors to the website.
The website team want to ensure that any information they gather about users based on their behaviour, or from Google analytics data, is consistent with the university's policies and its committment to safeguarding and protecting the privacy of its website users while being accessible and providing a positive user experience to everyone.
The team are concerned about possible unintended consequences so have decided to start by doing a User Jeopardy analysis to identify:
- What problems might surface?
- What questions they need to ask to understand these problems?
- Which groups of users they need to include in the research?
Personalisation can be characterised by considering the questions:
- What should be personalised?
- For whom do we need to personalise?
- Who does the personalisation?
Explicit personalisation, where the users participate by making choices, should do something sensible if no choices have yet been made. Implicit personalisation, where the system automatically generates personalised content, should similarly do something sensible on first use even if no information has yet been gathered.
1. Discuss which distinct groups of website visitors you might have. Are these groups exclusive, or might they overlap?
2. Discuss how fairness and equity might be lost. What might happen? Who to? Could you prevent it? If it happened, how would you know? What would make it less unfair? What might make it worse?
3. Discuss what choices might need to be made, and who by.
4. Discuss how agency might be lost if the design makes a choice hard to make. What might happen? Who to? Could you prevent it? If it happened, how would you know? What might make the choice easier? What might make it harder?
5. Discuss what data users might need to provide, and how often.
6. Discuss whether the benefit justifies the data provided, or would it be disproportionate? What questions would help you to know? Who should you be particularly careful to ask?
7. Discuss who is responsible, and what kind of consent needs to be given.
8. Discuss how accountability might be lost, or mis-assigned. What might happen? How would you know?