INDUSTRY
HealthTech
MY ROLE
Product Design
TEAM
PM, Domain Specialist, PO, 2 x Lead Dev, Architect
TIMELINE
2024 - 2025
ABOUT
RDx by Topcon is a web platform for eye-care professionals. The platform enables eye clinics to manage patient care and allows doctors to perform eye diagnostics remotely, by operating the Phoropter device online.
BACKGROUND
To perform a remote eye exam, a doctor should be assigned to the patient in the system. The assignments are done manually - either a store personnel or a remote doctor should pick a patient/doctor from the list. The manual work required training and was always dependent on user behavior, which can be inconsistent.
PROBLEMS
Manual assignment created misalignment between store personnel and doctors, and was inefficient, especially in peak hours
Doctors were cherry-picking "easy" patients or not selecting any patients at all
Confusion and misalignment impacted overall business performance, leading to hurt revenue, especially for high-volume customers
IMPACT
Reduced assignment time by 87% by introducing auto-assignment functionality
66% increase in doctors’ productivity - from the average 12 to the average 20 exams/doctor/day
32% increase in total user engagement
Part 1: Letting the system do the heavy lifting
On the left - store user selects a doctor for the patient; on the right - the Patient list view for the doctor to select a patient
DISCOVERY
After a few user interviews, we understood that the problem is bigger than just time-to-complete — often, users don’t even start the task.
How might we increase assignment frequency without extra pressure, integrating it naturally into users’ workflows?
DESIGN DECISION
A series of workshops led to the decision to automate it — write an algorithm that assigns doctors to patients based on the longest wait times. Users won't be forced to make the decisions; instead, the system will handle the complex thinking, allowing them to focus on care.
IMPLEMENTATION
We designed and integrated automation seamlessly, preserving familiar user workflows. Manual assignment remained possible and played along with auto-assignment. We also left room for edge case flexibility, accommodating, e.g., contact lenses workflow, which requires performing an exam for the second patient, while the first one is on the contact-lenses-fitting hold.
Final design in action - a doctor has been assigned to the patient automatically
Part 2: Bringing light to patient removals
DISCOVERY
The new auto-assignment functionality aimed to boost productivity by distributing patients more efficiently. However, it didn’t address a critical issue — doctors could still easily remove patients from their queue.
❌ No indication for store users of why a patient was removed, confusing them regarding possible follow up actions
❌ It opened the door to underperformance — doctors could simply decline patients without explanation. These patterns could go unnoticed by management.
INSIGHTS
The Comments section was originally designed to improve communication between store users and remote doctors. Qualitative data from end users, customer interviews, and discussions with CS helped us to understand that the function wasn’t being used to its full potential. Deeper discussion also revealed that customers had turned to a third-party chat for any store-to-doctor communications.
I saw an opportunity to solve three problems at once:
✅ Create a place to communicate removal reasons, as well as to track the follow-up actions
✅ Shift customers from a third-party chatting app to our product and increase the Comments functionality usage
✅ Move patient-related conversations out of the general store-to-doctor chat to a secure, patient-specific channel for information protection
DESIGN DECISION
We revamped the section by automatically feeding in patient removal reasons and connecting them with any follow-up messages. This kept the original commenting functionality intact but added more context, making conversations more meaningful.
Building on that logic, we also integrated cancellation reasons from the store side, turning the space into a traceable record of activity and communication. To support this shift, I addressed key UX and UI gaps — renaming the section to Activity Log, improving the message flow and scannability, refining indicators, and making minor visual enhancements for better usability.
VALIDATION
Post-launch validation showed a +15 pts boost in user satisfaction score (CSAT) and a 32% increase in the total user engagement in the Activity Log section.
Despite the numbers, we also improved the product usability – secured sensitive data by moving conversations into a patient-specific channel.
Final design
Drag the arrows to see before and after of the Activity Log section
REFLECTIONS
💬 In B2B2C content, despite the customers' decision drivers, it is important to talk with the end users, too. Advocating for user-centred design decisions based on feedback and finding a middle ground between business and users was something that had to be fought for to make the final solution excel.
🤝🏽 Partnering with engineers and the architect early in the process helped dramatically address challenges, share findings, and discuss ideas for such a complex feature. It is crucial as we are trying to focus on the outcome (allow users to focus on patient care than on the flow management) rather than a specific output (automated assignment).
🪄 Engaging customers is valuable to truly meet their needs, improve the product experience, and strengthen the relationship. I took the lead of co-creating solution in the absence of core PM team members due to illness. During a busy pre-Christmas week, I organized a round of workshops with clients to finalize the details and lock the core functionality.