Modernising the Consumer Referrals System
Summary
Modernising the consumer referrals system by redesigning an outdated and fragmented legacy workflow into a streamlined, single-section interface on a new platform.
Outcome
Achieved an 84% reduction in referral-related fields, which dramatically decluttered the user interface.
The new design reduced the time to complete a referral by 18%
The share of users finding the referral process "Extremely Easy" more than tripled from 19% to over 67%.
The Project
As Citizens Advice migrated its core services from the 15-year-old Flare21 system to its new bespoke Casebook platform, the entire consumer referrals process had to be redesigned from the ground up. The goal was to replace a clunky, outdated, and inefficient legacy system with a modern, intuitive, and streamlined workflow for our specialist advisors.
The Problem
The existing referrals process was a major source of frustration and inefficiency for advisors. Through observational research, I identified several critical issues:
Fragmented Workflow: Key information was spread across three separate screens, and advisors had to save the case after each step before proceeding. This was slow and unintuitive.
Forced Sequential Actions: The system only allowed advisors to make one referral at a time, even though cases often required multiple referrals (e.g., a referral to one body and a notification to another).
Risk of Duplicate Data: Advisors were forced to save a case before making a referral. If a client provided more information later in the call, advisors often had to update the case and resend the entire referral, creating duplicates.
Confusing and Outdated Options: The system was cluttered with redundant and confusing dropdown options, making it difficult for new advisors to learn and forcing experienced advisors to rely on memorised shortcodes.
Users & Audience
Consumer Advisors: The primary users, working in a fast-paced call centre environment where they are expected to understand a client's issue, provide advice, and complete a referral in under 10 minutes.
Partner Organisations (e.g., Trading Standards): The external recipients of the referrals who rely on accurate and complete information.
My Role & Responsibilities
As the Product Designer, I led the end-to-end redesign of the referrals journey. My responsibilities included:
Conducting observational research and task analysis to map the existing workflow.
Auditing legacy fields and collaborating with subject matter experts to redefine the referral logic.
Creating user flows, wireframes, and a high-fidelity clickable prototype in Axure.
Conducting usability testing with advisors and iterating on the design.
Producing detailed handover documentation, including logic diagrams and acceptance criteria, for the development team.
Process
1. Discovery & Task Analysis
I started by conducting observational research, watching advisors use the old Flare21 system in their busy work environment. The advisors were working under significant time pressure, with multiple screens and documents open at once. I used task analysis to map every click and decision, which clearly highlighted the high cognitive load of the existing process. Advisors were relying on manual workarounds and knowledge that wasn't captured in the system, creating an error-prone workflow.
2. Audit & Simplification
Working closely with a subject matter expert, I conducted a full audit of the existing referral types and fields. By reviewing internal process documents and analysing the task analysis findings, we were able to map out the logic for each of the seven referral types. This deep dive revealed dozens of redundant and confusing fields that had been added over the years. This work allowed us to consolidate two legacy fields ("RNS type" and "Case class") that were serving the same purpose into one simplified "Classification" field.
4. Testing & Iteration
We tested a high-fidelity prototype with six advisors. The sessions confirmed the new flow was much simpler, but also revealed a key flaw in my initial design: I had linked the case classification too rigidly to the referral type. Users explained that a single case might require both a "criminal" and a "civil" referral. I iterated on the design, moving the classification down a level to give advisors the flexibility they needed.
Solution
Based on the research, I designed a new, single-section solution within the Casebook platform that directly addressed the core pain points of the old system. The key features of the new design were:
Consolidated Interface
All referral actions were brought into a single, easy-to-use section, removing the need to navigate between multiple pages.
Multiple Referrals
Advisors could now create as many referrals, notifications, or signposts as needed for a single case, all at the same time.
Flexibility
The "forced save" was removed. The new design allowed advisors to complete tasks in any order and only save the case once the call with the client had ended.
Clarity
The confusing legacy options were replaced with a clear, hierarchical structure, and fields were auto-populated with information entered earlier in the form, saving advisors valuable time.
Usability Testing & Iteration
I built a fully clickable prototype in Axure and conducted usability testing with six advisors.
The sessions validated the core design but also uncovered a key flaw in my initial logic. I had initially linked the "case type" (e.g., criminal or civil breach) to the type of referral that could be made. The testing revealed that a single case could have both criminal and civil elements.
Based on this feedback, I iterated on the design, decoupling the case type from the referral itself. The final design moved the "Classification" (Criminal/Civil) down a level, allowing advisors the flexibility to make any combination of referrals needed for a single case.
Development Handoff
To ensure a smooth transition from design to development, I created a comprehensive handover package. A suite of documentation to give the engineering team clarity and context.
Walkthroughs & User Flows:
I ran walkthrough sessions with the development team to explain the user journey and the thinking behind the design decisions.
Detailed Acceptance Criteria
For each component and its variants (e.g., the different states of a Trading Standards referral card), I wrote clear acceptance criteria. This defined the expected behaviour based on user actions, leaving no room for ambiguity.
Logic Diagrams
For the most complex part of the system—the Trading Standards referrals and notifications—I created a detailed logic diagram. This visual flow mapped out all the conditional rules (e.g., "if the trader and client are in the same area, then..."), which was then used to break the design down into manageable Jira tickets.
Results
The redesigned referrals system was a huge success, directly addressing the core inefficiencies of the legacy process.
The new system is vastly more intuitive, with the share of users finding the referral process "Extremely Easy" more than tripling from 19% to over 67%.
The new system has successfully streamlined the referral process. It is now 18% faster for the typical user.
The audit and simplification work resulted in an 84% reduction in referral-related fields, dramatically decluttering the interface.
The new, intuitive workflow was praised by training staff, who noted it would significantly reduce the time needed to train new advisors. One commented: "To actually use wasn't difficult - wouldn't take long to train someone on this."