Multiple clients · 4 sectors
CRO, Personalisation & Optimisation in Practice
Behaviour-led diagnosis. Sharp hypotheses. Targeted design interventions specific enough to prove whether they worked.
The approach
Signal. Intervention. Impact.
Read the behaviour
Quantitative data identifies where intent drops off. Qualitative signals explain why. The starting point is always: what was the user trying to do, and what stopped them?
Design the smallest targeted change
Precision over scope. The more focused the intervention, the clearer the causal link between the change and the result. Every hypothesis ties a mechanism to an expected behaviour shift.
Measure what actually matters
Choose the metric that reflects the decision being made at that specific funnel moment, not the most convenient one to track. A well-formed test generates useful learning even when the variant loses.
Experiment library
Six engagements across four sectors
Some of my most memorable experiments across finance, education, e-commerce, and SaaS, each designed to solve a specific behavioural problem and improve a measurable outcome. Hover or tap a card to see the full story.
Xero
Returning visitors were seen as new arrivals. A personalised hero recognised prior intent and turned hesitators into trial sign-ups.
Returning visitors saw the same top-of-funnel hero as first-time visitors, with no acknowledgement of prior engagement and no direct path forward.
Hero personalised to visit history. Returning visitors received a message that acknowledged prior engagement and surfaced a more prominent trial CTA.
+23% free-trial registrations
Deakin University
Students close to applying hit a generic homepage. Intent-matched content at the critical decision moment lifted lead CVR by 78%.
High-intent returning students saw top-of-funnel content at the critical application decision moment. Messaging calibrated for no one in particular.
Homepage personalised based on prior course-area browsing: faculty-specific content, matched application steps, relevant social proof.
+78% lead CVR · +178% apply engagement · +18% YoY acceptances
Decathlon
Making commercial value scannable on product listings drove a +2% CVR lift and €1.54M in revenue.
High-intent shoppers scanning PLPs weren't seeing promotional and commercial value quickly enough to influence their click or add-to-cart decision.
Visual value stickers added directly to product listings, surfacing offers, exclusivity, and popular picks at the scanning stage, before the user clicks.
+2% CVR · +0.61% CTR · €1.54M business value
nabtrade
A fractured cross-channel journey was bleeding applications. Aligned messaging and onsite UX fixes increased application completes by 20%.
Users arriving from paid search encountered misaligned onsite messaging and a friction-heavy application flow. The cross-channel experience required users to mentally reconnect.
Cross-channel remarketing aligned with onsite UX improvements, reducing application friction and ensuring message continuity from ad through to form completion.
+20% application completes · +69% CTR · −62.5% CPC
Ubank
Adding reassurance at the start of the application, before the first high-commitment action, increased application completes by 25%.
Drop-off concentrated at the point of first commitment in the application flow. Users weren't yet sure what they were signing up for, or what came next.
A single reassurance step added early in the flow, setting expectations and reducing anxiety before asking for the first high-commitment action.
+25% application completes
Western Sydney University
Two personalisation gaps, Open Day and offer-to-enrolment, both solved by matching the experience to course intent. +43% signups, ~$20M revenue impact.
Generic homepage before Open Day. Personalised offer emails landing on a generic enrolment page, with continuity breaking at the most important conversion moments.
Intent-matched homepage personalisation for Open Day. Personalised email sequence matched to course + faculty, landing on a personalised enrolment journey.
+43% Open Day signups · Largest Open Day in WSU history · ~$20M revenue impact
What this work proved
Four principles that transfer across every sector
Relevance beats generic optimisation
A personalised experience that acknowledges what the user has already done consistently outperforms a generic experience optimised for a broad average. Recognising intent is itself a conversion mechanism.
High-intent moments need clarity, not more friction
When a user is close to committing, the most common mistake is introducing unnecessary cognitive work: too many options, reassurance withheld, value buried. The job at high-intent moments is to get out of the way.
Small UX changes can unlock major commercial impact
The Decathlon sticker work was not a redesign. The Ubank change was a single added step. Precision targeted at exactly the right friction point makes the commercial impact disproportionate to the design effort.
Structural problems require structural hypotheses
The highest-impact tests challenged assumptions already built into the product, about what users needed to see, in what order, before they would act. Holding those assumptions lightly is where the real learning lives.
Booking.com: Design Language Portal
Creating the standards and infrastructure that helped 100+ designers and external partners make consistent design decisions at scale, without needing to ask.