Project overview
Bupa UK was consolidating 15 fragmented internal systems into a single Microsoft Dynamics platform used by more than 1.200 customer service agents across multiple channels in all regions.
I led the end-to-end design of the Identity & Verification (ID&V) workflow. The ID&V process is one of the most operationally critical journeys within the platform as it’s the first touchpoint to protect customer data.
Duration: 3 months
Outcomes
90%+ reduction in ID&V-related QA failures
10+ second reduction in handling time
100% adoption across contact channels
Reduced onboarding time from 1 hour to 25 minutes
*Some mock-ups are representative recreations. Due to confidentiality constraints of legacy/internal systems, the original interfaces cannot be publicly shared.
The goal
Reduce handling time when serving customers
Reduce Data Protection QA outcomes
Reduce training time and costs for new agent onboarding
Improve operational consistency
Support vulnerable customer interactions
Increase agent confidence during live customer calls.
My role
Senior Product Designer (solo)
I led:
Discovery and workflow research
Interaction and UI design
Accessibility improvements
Prototyping and testing
Cross-functional collaboration with Product, Engineering, QA and Operations
Rollout and adoption support
The challenge
Customer service agents handled highly sensitive customer interactions across multiple disconnected systems.
Verification workflows were inconsistent, difficult to follow, and heavily dependent on agent memory and personal judgement.
This created operational problems across the contact centre:
Verification approaches varied between teams
Customers were over-verified due to compliance anxiety
QA failures increased operational risk
Agents relied on tribal knowledge rather than workflows
Vulnerable customer handling lacked contextual support
The problem wasn’t training.
The workflow itself created ambiguity.
Understanding the workflow
I worked closely with customer service agents, QA teams, operations managers and stakeholders across multiple channels.
Research included:
Live call observations
Workflow walkthroughs
Empathy mapping
Existing system audits
QA reviews
Prototype testing
One of the clearest behavioural patterns was defensive behaviour.Because agents lacked confidence in the workflow, they compensated by:
Asking excessive verification questions
Repeating steps unnecessarily
Relying on memory instead of tooling
During research sessions, agents repeatedly described uncertainty during live calls.
“I prefer to ask more questions just to be safe.”
Ethan, Customer Service Agent
“I hope QA won’t listen to this call.’’
Mary, Customer Service Agent
The workflow needed to reduce uncertainty.
Not just improve usability.
The before - Existing experience
The original verification flow contained multiple operational and usability issues:
Inconsistent layouts and reading patterns
Poor visibility of verification outcomes
Low contrast and accessibility issues
Limited guidance during live calls
No clear progress indicators
Agents often couldn’t tell:
Whether verification had succeeded
Which steps were mandatory
What had already been completed
What required attention next
Step 1: Find the customer by adding customer details
Step 2: Select the customer from the results
Step 3: Complete the identify and verification process
Step 4: Agent loads up customer’s dashboard with mixed feelings
This resulted in cognitive load during customer interactions that slowed down decision-making.
The after - Designing for operational confidence
The redesigned ID&V workflow introduced:
Guided verification flows
Clear pass/fail visibility
Simplified reading patterns
Accessible hierarchy and contrast
Mandatory step visibility
Reduced navigation friction
Step 1: Find the customer by adding customer details
Step 2: Select the customer from the results
Step 3: Complete the identify and verification process (Passed ID&V example)
Step 3: Complete the identify and verification process (Failed ID&V example)
Supporting vulnerable customers
One of the most important additions was contextual support for vulnerable customer interactions.
When vulnerability indicators were present, the interface surfaced high-level vulnerability information at the beginning of the interaction.
This helped agents:
Recognise vulnerable situations earlier
Adjust communication approaches
Reduce customer distress
Follow safeguarding guidance more consistently
Every interaction decision focused on helping agents quickly understand:
What they needed to do
What had already happened
What the outcome was
What required attention next
The AI solution with Merlin Co-pilot
To reduce cognitive load during customer interactions, we introduced Merlin Copilot. This is contextual AI knowledge assistant embedded directly into the workflow.
Based on customer intent and interaction context, Merlin Copilot surfaced relevant operational guidance in real time.
For example:
Vulnerable customer calls surfaced safeguarding guidance
Claims interactions surfaced claims handling processes
Verification scenarios surfaced operational compliance guidance
Instead of forcing agents to search through large knowledge bases mid-call, relevant information became available directly within the workflow.
Validation and Rollout
The workflow was tested iteratively with agents and operational teams throughout development.
Testing focused heavily on:
Verification confidence
Ease of navigation
Handling speed
Operational clarity
Accessibility
The rollout demonstrated that operational adoption increases significantly when users trust they can complete workflows reliably and efficiently inside a single experience.
This became one of the most important strategic learnings from the programme.
“I never thought I’d get excited about ID&V!”
Molly- Customer Service Agent
I could use that!
Kirk- Customer Services Director
“The Microsoft guys were so impressed - I think we might be doing something a bit different to what they have seen!”
Steve- Chief Insurance Architect
Metrics
90%+ reduction in ID&V QA failures
Guided workflows improved process consistency and significantly reduced operational errors one year after launch.
1.
10+ second reduction in handling time
Simplified workflows reduced hesitation, repetitive questioning and navigation friction.At enterprise call volumes, even small time reductions create significant operational impact.
2.
100% operational adoption
All agents adopted the new workflow across contact channels. The simplified workflow and improved visibility increased confidence and usability during live customer interactions.
3.
58% reduction in onboarding time
Training time reduced from 1 hour to 25 minutes through simplified workflows and clearer guidance.
4.
Key learnings
The project reinforced how strongly operational adoption depends on end-to-end workflow confidence.
Users adopt systems they trust they can complete tasks within reliably and efficiently, not isolated features.
Operational tools also directly influence confidence, stress levels and decision-making.
Reducing uncertainty became just as important as improving efficiency.
Working within Microsoft Dynamics limitations reinforced another important lesson: Just because we can build it, doesn’t mean we should.
Final reflections
This project changed how I think about enterprise adoption and operational design.
Initially, the focus was on improving individual workflows and operational features. Over time, it became clear that adoption only happened when users trusted the broader workflow experience and felt confident completing tasks within the platform.
The project reinforced the importance of designing for operational confidence, reducing cognitive load, and supporting real-world complexity inside enterprise environments.
It also demonstrated how thoughtful UX decisions inside internal tooling can significantly reduce operational risk, improve compliance outcomes and create measurable business impact at scale.