01Spreadsheet audit
Case details, insurer information, clinic coordination and status tracking lived in one dense operational sheet.
Design implication
The product needed a traceable case model rather than a digital copy of the spreadsheet.
DoctorAnywhere
One spreadsheet was carrying too much of the care journey.
A concierge team was coordinating referrals, insurers, clinics and patient updates through one complex spreadsheet. I led the move to a case-and-ticket platform that reduced manual work while keeping difficult decisions with the agents.
Service DesignProduct DesignB2B SaaS
- My role
- Lead Product Designer — product strategy, research, service blueprinting, MVP definition, interaction design and validation
- Scope
- MVP → launch
- Focus
- Healthcare · Workflow · MVP
- Evidence status
- The 90% reduction in human error and 80% reduction in manual workload are the verified operational outcomes used here. Usability evidence came from an eight-task test.
Verified operational and usability impact
−90%human error
−80%manual workload
77.5System Usability Scale
My role, the team and how success was defined
A short project brief before the detailed story: the setting, my responsibility, the people involved and the outcome we were working towards.
- 01Service context
- A concierge team coordinated GP referrals, insurer requirements, clinics, appointments and patient updates through a spreadsheet-heavy workflow.
- 02My ownership
- I led discovery, reconstructed the current service, defined the case-and-ticket MVP, designed the agent workspace and validated the critical workflow.
- 03Core collaborators
- Three concierge agents and two stakeholders informed discovery, with Product and Engineering involved in scope, rules and delivery feasibility.
- 04Success definition
- Reduce manual effort and preventable errors while keeping complex clinical and operational exceptions visible to the agent.
What was happening—and why it mattered
The spreadsheet mixed patient context, insurer rules, clinic coordination and status tracking in one place. Errors, slow case completion and unclear hand-offs affected patients and staff, but over-automation would have hidden the exceptions agents still needed to judge.
- A single spreadsheet created a high risk of human error.
- Patient complaints averaged 32% daily.
- Slow case completion contributed to 48% patient drop-off.
- Only 10.5% of requested appointments were secured and proceeded.
The spreadsheet was only the visible problem. The product had to model the service and its exceptions underneath it.
Evidence that changed how I understood the problem
Behavioural, operational and qualitative evidence did different jobs. Together, they helped define what the team needed to solve before choosing a direction.
02Frontline interviews
Three agents and two stakeholders described different needs across case creation, assessment, follow-up and hospitalisation scenarios.
Design implication
The workflow needed clear common stages with room for exception tickets and agent judgement.
03Journey mapping
GP referral and cashless-insurance journeys shared work but diverged at key hand-offs, evidence requirements and escalation points.
Design implication
One platform needed to support multiple service paths without forcing them into an identical sequence.
What customers saw depended on what happened behind it
The blueprint connects the customer journey, product behaviour and operational responsibilities across the same sequence.
StageCustomerFrontstageBackstage
1 · Receive
A referral or specialist request enters the concierge service.
The patient receives acknowledgement and clear next-step expectations.
An agent creates a structured case with source, insurer and referral context.
2 · Assess
Provides missing information when required.
Requests are specific rather than repeated or ambiguous.
The agent validates eligibility, documents and urgency before progressing.
3 · Coordinate
Waits for a suitable clinic or specialist option.
Progress communication reduces uncertainty during coordination.
Tickets track clinic, insurer and internal follow-ups inside the case.
4 · Confirm
Reviews and accepts an appointment option.
Appointment details and requirements are made clear.
The agent records confirmation and closes dependent follow-ups.
5 · Close or escalate
Completes the hand-off or receives support for an exception.
The outcome and next owner are visible.
The case history remains traceable for closure, complaint or escalation.
Scroll horizontally on smaller screens to follow the complete service.
The choices that shaped the work
The artefacts mattered when they helped the team choose. These are the questions, evidence and decisions behind the final direction.
Question
What should replace the spreadsheet?
Evidence
A flat row mixed a patient journey, many follow-ups and several owners.
Decision
Use a parent case for the patient request and linked tickets for actionable follow-ups.
Question
How much should the MVP automate?
Evidence
Rules repeated, but exceptions still required agent interpretation and coordination.
Decision
Automate structure, states and traceability first while keeping judgement with the concierge team.
Question
Where should scope stop?
Evidence
Referral, insurer, appointment and hospitalisation paths could make a first release too broad.
Decision
Prioritise the smallest coherent case lifecycle and validate eight critical tasks before scaling scenarios.
One problem, viewed through service and product
The operating service and the interface formed one system. Each lens addressed a different part of the same problem.
Service Design
Shaping the ecosystem, hand-offs and operating model.
- Mapped GP referral, cashless insurance and escalation journeys
- Combined stakeholder hypotheses with frontline interviews
- Designed current- and future-state service journeys
- Clarified roles, states and hand-offs across the concierge operation
Product Design
Turning service decisions into clear, usable product behaviour.
- Defined the MVP around cases, tickets and status stages
- Translated service rules into end-to-end task flows
- Designed and tested low- and high-fidelity prototypes
- Built a scalable agent workspace using the design system
How the work moved from uncertainty to a decision
Understand how the work was really done
Referral, insurance, patient-source and escalation paths were mapped alongside the working spreadsheet. The differences between documented process and daily practice helped define the actual service problem.
Test the team’s assumptions
I facilitated stakeholder mapping, wrote a 15-question guide and interviewed three concierge agents and two stakeholders about case creation, assessment and hospitalisation.
Choose the right product model
Current- and future-state journeys led to a parent case for the patient request and smaller tickets for accountable follow-ups. That model became the boundary for the first coherent MVP.
Validate the critical workflow before expanding it
Stakeholder and agent reviews shaped the low-fidelity flow. An eight-task usability study on the high-fidelity product then identified the issues to fix before delivery.
Three connected views for the complete concierge workflow
The high-fidelity design turned the service model into a practical agent workspace. Read together, these screens show how an agent creates a structured case, coordinates the referral and closes the loop with a confirmed appointment and traceable history.
01 · Structured intake
Create a case without recreating the spreadsheet
A stepped intake captures GP, patient and insurer information in the order agents need it. Policy benefits and exclusions remain visible at the point of entry, reducing rework before the referral progresses.
02 · Active coordination
Manage the referral from one operational view
The case header, appointment ticket, supporting documents, ownership and suggested specialists sit in one workspace. Agents can move the request forward without reconstructing context across separate tools.
03 · Confirmed outcome
Close the loop while preserving the case history
The booked state keeps the chosen specialist, appointment time, follow-up actions and status trail together. Agents can change or cancel the appointment while retaining a clear record of how the case reached its outcome.
The constraints that shaped the work
These constraints affected the direction, the order of work and the compromises made with the wider team.
Healthcare work is full of exceptions
The tension
A standard workflow could make urgent or unusual patient situations less visible.
My response
The case lifecycle was standardised, while tickets, notes and escalation states kept exceptions explicit.
The live service could not stop
The tension
Agents still had active patients to support while moving away from a familiar spreadsheet.
My response
The MVP followed their existing mental model and focused first on the manual work carrying the greatest risk.
Many parties, one workspace
The tension
Patient, referral, insurer and clinic information all needed to be available without reproducing another dense all-purpose view.
My response
Information was grouped by task and case stage, with ownership and history kept traceable.
What the evidence shows
8critical tasks tested
83.3%error-free completion
5agents and stakeholders interviewed
2service journeys mapped
The 90% reduction in human error and 80% reduction in manual workload are the verified operational outcomes used here. Usability evidence came from an eight-task test.
What changed
01
The spreadsheet-heavy operation became a structured digital workflow.
02
Manual workload fell without removing agent judgement from complex cases.
03
Patient requests, clinic updates and status history became traceable in one place.
04
The platform created a base for later provider-recommendation work.
What I learned and would do next
I translated a complex human service into a scalable product model without automating away the judgement that made the work safe.
The useful abstraction was not a better spreadsheet. It was a case lifecycle with smaller, accountable tickets for the work inside it.
I would track time in stage, reopened tickets, exception categories and gaps in patient communication to decide what to automate next.