Challenge
The service needed to be as trustworthy and clear as GOV.UK itself, offering consistent, distraction-free guidance to users while adapting to the unique requirements of conversational interfaces.
My brief was twofold:
- Backstage: Figure out how to adapt backstage processes and technology to create a robust, scalable, and trustworthy chat experience.
- Frontstage: Explore what it would mean from a front-end perspective “if the GOV.UK website itself could chat with visitors” reflecting its clarity and reliability in a conversational format.
Approach
I began by aligning cross-functional teams, including content designers, developers, and policy experts, to define the chatbot’s purpose and scope. Inspired by GOV.UK’s no-distractions service ethos, we focused on delivering precise, relevant responses to user queries while reducing “dead ends” in the chat flow.
Key steps included:
- Starting with user needs: Collaborating with researchers to understand common user pain points. Insights from user research shaped every design decision.
- Backstage innovation: Simplifying integration with both HMRC and govuk’s existing content management infrastructure to ensure chat responses remained accurate and up to date.
- Iterative design and testing: Prototyping both the govuk-inspired IxD and the conversational flows and testing them with users to ensure clarity and utility.
Outcomes
- 24/7 assistance: Delivered a chatbot capable of handling a significant volume of inquiries, reducing strain on helplines.
- Trust and satisfaction: Users described the experience as clear and dependable, echoing GOV.UK’s tone of voice.
- Efficiency gains: The chatbot saved HMRC an estimated £5 million in support costs and paved the way for further adoption of conversational UIs across government services.
Impact
The HMRC chatbot project demonstrated the value of combining human-centred design with robust backstage processes. By embedding GOV.UK’s design principles and values into a conversational experience, we were able to deliver immediate operational benefits.
The view from the research team
Read Ruth Drake’s great blog post on designing helpful chatbots
Can I help you solve a similar problem? dug@goodlookslikethis.com