.png)
Transforming digital referral journeys at Bupa dental
Scope of work
Discovery and research
User experience (UX)
User interface (UI)
User Testing
Facilitation and workshops
Company
Bupa
Year
2022
My role
Co-lead designer (1 of 2)
An in-person model with digital opportunities
Bupa has an internal cross-referral initiative connecting Health Insurance members to Bupa-owned services such as Bupa Dental. The model was entirely done manually.
When members visited a Bupa Health Insurance retail store or contacted the call centre, staff could verbally offer a free $40 voucher to visit a Bupa Dental clinic for the first time.
The problem? Approximately 70% did not follow through with a booking or purchase.
Once customers left the store, momentum dropped and conversions did not follow through.
The challenge
How might we redesign the referral experience so members receive personalised, in-context referrals at the right time and in the right channel?
For customers, we needed to find a solution that could make the referral they receive relevant and timely, as well as help them get more value from their cover.
For the business, we explored how we could increase acquisition into Bupa-owned services and improve the conversion from referral to appointment.
.jpg)
Digital referral redesign
As the lead designer within the innovation team, I:
-
Framed the opportunity alongside commercial and service leads
-
Partnered with researchers to leverage their guerrilla interview and research discovery insights
-
Mapped the end-to-end referral journey across retail, contact centre, and dental
-
Translated behavioural insights into experience principles
-
Designed and led three MVP experiments end-to-end
-
Worked closely with engineering to ship, launch, and measure each iteration
This was a test-and-learn program designed to validate demand before investing in a solution for scale.
Phase 1: Framing the right problem
By collaborating with our researchers, we started discovery through guerrilla interviews in selected Bupa health insurance retail stores to talk to customers and staff alike. We uncovered three core behavioural insights and opportunities:
The referral was out of context
Customers came in to resolve their health insurance needs, with dental not front of mind. The offer felt secondary and disconnected from their immediate intent.
Opt-in did not equal action
Verbal acceptance in-store did not translate to bookings. There was a clear drop-off between “Yes” and actually claiming the $40 referral voucher and visiting a Bupa dental clinic for a service.
Momentum was lost after exit
For those gave verbal acceptance and have not booked any dental service, follow-up calls happened days later, often at inconvenient times. Without immediacy, intent cooled and customers forgot. Only 14% of customers actually followed through.
Phase 2: Designing our experimentation strategy
Rather than investing in a fully integrated solution upfront, we designed a structured test-and-learn pipeline to validate demand.
Through pretotyping workshops, we identified the smallest possible experiment that could test our core hypothesis:
Would a contextual digital referral drive higher follow-through than an in-person offer?
We introduced a simple baseline experience:
A $40 dental referral surfaced as a banner in selected digital channels, leading to a lightweight opt-in form.
We then tested where and how it performed best before committing to any final solution for scale.
Phase 3: Iterative test, learn, refine
We ran structured test-and-learn cycles and co-design sessions wiwith messaging consultants and engineers to strengthen the AI’s reliability in real world conditions.
This included facilitating keyword-mapping and semantic workshops with consultants to improve accuracy and relevance. We captured synonymous language used in real customer conversations and mapped policy sections commonly referenced together, ensuring the AI could understand intent in context, not just keywords in isolation.
In parallel, we partnered closely with engineering to refine prompt strategy, establish guardrails for scope control, and structure content semantically to improve contextual accuracy.
Because OpenAI is a pre-trained model with no inherent Bupa context, this phase required deliberate knowledge and systems design. It was the critical step that transformed a promising proof of concept into a reliable MVP consultants could trust and use.
Phase 3: Running the experiments
Experiment 1: Placing offer at the myBupa Health Insurance app
We placed the referral banner inside the logged-in member portal (MyBupa), particularly within Extras pages where members were already reviewing benefits.
Behind the scenes, referrals were manually processed by our team. This was a deliberate Wizard-of-Oz approach to test demand before automating. This channel significantly outperformed retail-only referral, with increased referral conversion from 14% (baseline) to 46%.
Experiment 2: Placing offer in paid socials channel (Facebook)
We tested referral banners via paid social channels targeting members externally. Results showed lower conversion and weaker intent compared to authenticated environments.
Experiment 3: Booking Flow Integration
We explored reducing friction by shortening the journey between referral acceptance and booking.
A contextual digital first referral model
Based on experimentation, we identified authenticated member environments as the strongest scalable pathway. The redesigned referral model focused on three core shifts
$118k
revenue generated from experiment MVPs
Across three MVP experiments driving new patient acquisition to Bupa Dental.
380+
new Bupa dental customers/ patients acquired
Bupa health insurance members who converted via this digital referral experiment project
$2.1M
projected annual revenue if scaled and strategically operationalised
If implemented as an always-on authenticated member experience.
7k+
potential new Bupa dental patients annually
Helping more Bupa health insurance members expand their access to dental care and maximise their policy coverage
Impact from Pilot to Scale
What I learned from this project
Pretotype small, learn big. Yes, that's PRE-to-type. Refer to Alberto Savoia's amazing book, the right it for more amazing methodologies!
It's okay to be scrappy
Not every experiment needs a perfect backend. Sometimes a manual workaround is the fastest way to learn what people actually want.
Don't overbuild too early
It’s tempting to design the perfect integrated system upfront.
But investing in heavy build before validating demand increases risk.
We proved demand before investing time, money, and engineering effort.
The right solution will reveal itself, if you allow yourself to try
The best-performing channel wasn’t obvious. Testing different paths gave us real signals instead of assumptions.
Other work
Designing an AI Agent for Bupa's consultants
Transforming consultant workflows with $1.02M impact
Simplifying course fee discovery at Macquarie University
Designing a 3-step calculator to replace complex PDFs
Design leadership in practice
Creating spaces, tools, and rituals that help design thinking spread beyond the design team.
.png)
Let's make something meaningful together
Made in sunny Radelaide!
© 2026 by Denise Tan