Citizens Advice Scotland
Research | Content & UX design | Co-design | evaluation
Co-design and prototyping of chatbot content & schema to triage clients and increase capacity
About
Citizens Advice Scotland (CAS) provides a range of free advice services through a network of 59 bureaus run by staff and volunteers. Although some advice services remain face to face, the pandemic further influenced a move to a multichannel approach using phone, email and web chat.

The Civtech programme is funding CAS to develop and prototype digital solutions to improve its services.
Challenge
Effects of the pandemic have created an increase in demand for bureau services, and at the same time reduced the number of volunteers. In this environment CAS needs to find ways to increase capacity for advisers to help complex cases by encouraging people who can to to self-serve.
Aim
To build, test and evaluate an MVP chatbot by launching a pilot on bureau websites. Two approaches would be tested and evaluated: logic trees and AI.

The aim is for the chatbot to identify a person's issue and signpost them to one of over 1000 on the CAS advice site, or to help them request a callback.
My role
Working at SIDE Labs, I was the main point of contact building relationships with the bureaus to facilitate the pilot, including negotiating data agreements.

Over six months I worked with advisers in bureaus, and closely with a developer to create the content and functionality. I analysed user input to recommend design improvements and led the end of project evaluation.
Work undertaken
Research
  • Observations working alongside bureau frontline staff
  • Shadowing advisers on client calls and appointments
  • Semistructured Interviews with advisers to understand service & content areas
Content & UX design
  • Mapping of content from CAS advice site
  • Design of key routing flows for chatbot functionality
  • Weekly analysis of chatbot transcripts, recommendations for design iterations
Co-design
  • One to one online workshops with advisers to review content design
  • Trials to explore CAS Digital content team to design bot flow architecture
Evaluation
  • Evaluation framework design
  • Workshops with bureaus to understand reporting needs
  • Coding of chatbot transcripts and analysis of resolution rate for logic & AI approaches
  • Structure and content for evaluation report
    Outcomes
    Reviewing the evaluation, CAS and the bureaus involved in the pilot were in support of further development if funding could be obtained. You can read more about the outcomes in this summary of the report.

    CAS commissioned us to develop a chatbot for an additional two services - Money Talk Team and the Patient Advice & Support Service.
    Gallery