IMPACT
3-5x Increase in GP Actions
+88% GP Action Rate
ROLE
Product Designer
TEAM
Small Agile Team of PM and Developers
BetterConsult • 2024 — 2025
Quick Scan Summary
Problem
Our core recommendation features for GPs during consultations had average <10% engagement. GPs saw it as irrelevant, redundant, or buried in the UI.
Experiments
Structural visibility (UI changes) → moved & simplified prompts.
Patient-led prompts → opt-in signals surfaced in GP notes.
Results
The new design (with AstraZeneca Lung Cancer campaign) drove 74% more GP actions than the old design, and 158% more than when no recommendations were shown.
+82% chest CT referrals
140+ additional screened patients (patients likely missed under the legacy flow)
6 new lung cancer diagnoses within 2 months of the new cohort launch
Client reaction: requested to expand the campaign through year-end (deal in negotiation)
Patient-led experiment (with Yaz/Yasmin campaign) led to 88% GP action rate (vs 17% static, 13% none)
Patient Opt-in rate: 31.2% overall
Meaningful user impact: Patients became aware of PBS changes beforehand and proactively initiated discussion about their treatment options.
Impact
Small design shifts delivered 3–5× more engagement and measurable clinical outcomes.
Business Impact
Partnered pharmaceutical campaigns powered by this recommendations feature have contributed $2.1M+ in revenue, strengthening both commercial and clinical partnerships.
Situation
BetterConsult is a digital pre-consultation tool that summarises patient-reported symptoms and medical history for GPs before an appointment. One of its core features is GP Actions To Consider — clinical recommendations tailored to the patient’s context, developed with sponsored partners like AstraZeneca, GSK, and Pfizer.
These prompts were designed to support patients’ preventive care and help identify early health risks. Yet in real-world usage, an average of around 15% of GPs acted on these prompts during consultations, though engagement rates varied by campaign. This didn’t mean the prompts lacked value — but it suggested they weren’t being delivered in a way that felt trustworthy, timely, or actionable within the GP workflow.
Engagement with this feature was also being closely monitored internally, as our sponsored partners expected measurable clinical value from their involvement. Improving GP interaction was not only a product concern — but also important to sustaining trusted relationships with our commercial and clinical stakeholders.
Complication
To understand the problem, we conducted 25 qualitative interviews with GPs, supported by session recordings and usage analytics.
What we uncovered:
💬 "I glance at BetterConsult for what the patient says. I don’t expect clinical advice there."
— Inner metro GP, 15-minute consult style
Misaligned with consultation flow
GPs often viewed BetterConsult as a tool for capturing patient voice, not as a decision support system. Recommendations unrelated to patient's primary complaints felt out of place during time-pressured, problem-focused consultations.
B. Redundant with existing tools
Many prompts surfaced too late — after similar guidance had already appeared via Best Practice, DCP, or even the practice nurse. Some GPs were unaware of the section altogether; others had mentally “tuned it out” after seeing too much irrelevant content
C. Cognitively and visually overwhelming
Even when they did scroll to the section (which happened in only 6% of sessions), the table format and text-heavy design discouraged attention.
📉 Other supporting data
65% of GPs did not scroll far enough to reach the section (Session Recordings)
Among those who saw it, <15% engaged with the content. (Usage Analytics)
85% were unaware of the section, or didn’t trust its source. (User Interviews)
Question
The Product Manager and I posed two key design challenges:
How might we restructure and reposition the section so it aligns with a GP’s natural consultation workflow?
Can we shift from passive content delivery to timely, patient-activated nudges that encourage meaningful engagement?
Answer
I collaborated with the Product Manager, Engineering, Clinical Advisors, and Commercial teams to test these hypotheses through rapid design–test–learn cycles. Each experiment was designed independently to explore different levers of engagement: GP-facing UI changes and patient-facing behavioural nudges.
We designed two separate experiments to validate:
Whether structural and content changes to the GP interface would improve visibility and usability (Experiment 1)
Whether patient-led prompts could indirectly increase GP engagement through timely pre-consultation nudges (Experiment 2)
Answer
Test 1: Making Recommendations More Visible and Usable
Based on early interview findings, I proposed focusing first on structural visibility rather than content changes. Since most GPs weren’t seeing the section at all, we prioritised changes that would bring the content into view and reduce initial cognitive load.
Design Changes:
Moved the recommendations above the presenting complaint summary
Reduced heading copy to short, verb-led actions (e.g., “Consider ordering iron studies”)
Added progressive disclosure (“See more”) to reduce initial visual load
Applied consistent hierarchy and spacing to support scannability
These changes were informed by feedback from time-pressured GPs, who often described long blocks of text as difficult to engage with during fast-paced consults.
We piloted this experiment within AstraZeneca’s lung cancer awareness campaign. Lung cancer remains one of the most prevalent and underdiagnosed cancers in Australia, with many cases detected only at late stages.
📊 What we measured:
Backend logs of GP actions taken
📈 Results:
The new design drove 74% more GP actions than the old design, and 158% more than when no recommendations were shown.
This behavioural lift translated into measurable clinical outcomes. In the AstraZeneca lung cancer campaign, more patients were prompted to undergo chest X-rays or CT scans (+82%) compared to the old design. Within the 2 months of launch, this directly resulted in 6 new lung cancer diagnoses.
Answer
Test 2: Nudging GPs through Patients
In parallel, we also hypothesised:
💬 What if GPs paid more attention when the prompt came from the patient?
We tested this by adding opt-in prompts to the patient questionnaire. If selected, a lightweight message appeared in the GP’s note, suggesting the patient wanted to discuss a relevant recommendation.
This design decision was informed by early interviews, where GPs emphasised that in a typical 15-minute consultation, they prioritise concerns that patients explicitly raise or appear prepared to discuss. This insight shaped our hypothesis that a patient-led signal — not just a system-generated prompt — would better align with how GPs naturally triage and respond to information.
We selected a campaign focused on the PBS listing of Yaz/Yasmin (contraceptive medication) for its high relevance, low discussion burden, and clarity of eligibility criteria.
📈 Results:
Patient opt-in rate: 31.2% overall
GP action rate: Opt-in: 42.6% / Opt-out: 10.3% / Control: 6.8%
Impact: A clear patient-led signal increased GP actions 4× compared to control.
Meaningful user impact: Patients became aware of PBS changes beforehand and proactively initiated discussion about their treatment options.
Answer
What We’re Doing Now
This project is still in progress. Current and upcoming actions include:
Continuing to monitor backend data for GP interaction rates across different cohort types.
Conducting follow-up interviews with GPs to validate perception and effectiveness of design changes.
Expanding patient-facing prompt experiments to other campaigns (e.g., shingles vaccine, cancer screening).
Aligning internal campaign performance reviews with design iterations to refine strategy.
We aim to evolve GP Actions To Consider from a low-engagement add-on into a workflow-aligned, trust-based, and patient-activated clinical recommendation tool.