📈Commercial
2×
Product surface usage — placement change, no feature work
Repositioning the Popular product builder widget to a primary surface doubled usage without changing the product. The intervention was placement, not features — a commercial argument built from product engagement data across 80+ eligible matches.
4×
Product surface usage — next phase target
Surface expansion combined with a behavioural nudge mechanism surfacing real-time engagement volume — modelled at 4× current usage. Business case benchmarked against external A/B results and a Qubit/PwC meta-analysis of 6,700 experiments.
48×
Story point efficiency vs comparable widget build
The proposed mechanism at ~10 story points is projected to deliver significant incremental revenue. The previous comparable build consumed ~60 story points and generates ~40 interactions per week. That ratio was the centrepiece of a formal decision review document challenging business reluctance.
80–85%
Carousel engagement concentrated in positions 1–2
Production analytics showing positions 1–2 driving 80–85% of engagement regardless of how many items are shown. Used to reduce options from 7–8 to 3–4 — cutting technical load, improving page speed. A UX simplification that became a performance and commercial argument simultaneously.
⚙️Operating Model & Delivery
1 Qtr
Design ahead of engineering — sustained
Arrived with design 2 weeks ahead. Embedded into leadership planning cycles — now one full quarter ahead. Dual-track swimlanes protect the pipeline while current-quarter work ships. Creates space for technical feasibility conversations before architecture is locked.
Full system
Design operating model built from scratch
Source-of-truth ticketing. 3-lane delivery standard (Discovery / Design / Dev Support). Quality gates at each stage. Production impact loop as part of definition of done. Async protocol adopted across AU/SE/FR/RO timezones. Not a process document — a working system adopted org-wide.
–20%
Handover time reduced
Discovery to Test to Spec to Dev Support model keeps design active through build. Fewer defects, less rework. Structured ticket model separates exploration from build-ready work — engineering always knows exactly what stage design is at.
First
WCAG embedded in Acceptance Criteria
Accessibility requirements in ACs for the first time — closing years of accumulated debt. Makes a11y a delivery gate, not a retro audit. Targeting 50% app coverage as baseline. Practical spot-check protocol at Stage 3, not theoretical compliance.
📊Measurement & Systems Architecture
L0–L3
Depth-stratified analytics framework
Defined how the organisation measures design ROI — a surface taxonomy anchored to navigation depth rather than static pages. Separates surface effectiveness from module effectiveness in attribution. Reusable across any product vertical with a metadata filter only. No methodology rebuild per product change.
5 pillars
Promotions engine — engineering once, CRM-operated forever
Designed a five-pillar promotion architecture separating token mechanics (engineering-once, versioned, locked) from audience segments, qualification missions, bet targeting, and promo assembly (CRM-operated, zero code per campaign). Removes engineering from every new promotion after the initial token build. Live across GB, FR, RO, NL.
Decision Map
Design authority held at exec level — formally
During a high-stakes FR market launch, presented a formal Decision Map to Heads of and Exec — four open topics, stakeholder ask vs PD position vs proposed v1 decision on each. Held: single card template, MoTD vertical, density via token over variant sprawl. Trade-off rule documented in writing: if something is added, something else is removed or the date moves.
Live
Bet type capability matrix & friction event taxonomy
Mapped which bet types can be placed natively at each depth vs where they require lateral navigation. Defined friction events and cross-level shortcut instrumentation requirements. Handed to Experience Analytics as the instrumentation brief — design defining how the organisation measures itself.
👥Customer & Research
10+
Research studies launched — never done before for UK
Spearheaded the first structured research program for the UK regulated digital product product — equivalent to what existed in FR, built from scratch. 10+ unmoderated studies in the year since launch, each running 100+ participants. A capability that did not exist before, now a standard part of how product decisions are validated.
100+
Participants per study — cross-market agreement on tooling
Before any study launched, drove formal agreement with teams across all markets on Maze as the shared testing platform — faster execution, consistent methodology, comparable data across GB, FR, RO, NL. The infrastructure agreement came first; the research scale followed.
25%
Users adding to transaction flow from homepage
Analytics showed homepage was already a conversion surface — 15% placed a bet directly from it. Reframed page priority across product and content teams without a lengthy research cycle. Data already existed — the work was making it visible and actionable to the right people.
9
Hero journeys → product initiatives — design drove product strategy for the first time
Survey with internal Trading, Ops, and Risk teams identified 9 hero journeys across the back-office platform. Ran 3 workshops (3 journeys each) writing product initiatives directly from real user problems — not assumptions. Design had never driven product initiatives for this platform before. Problems became the backlog.
⚡AI & Workflow Automation
4–6 hrs
Admin time reclaimed weekly — across design management
Jira ticket completion checks, Confluence updates, and stakeholder summaries automated across 3 design managers. AI reads tickets marked done, cross-checks artefacts against agreed Ways of Working, and generates stakeholder summaries in the right tone — replacing what was a manual review and write-up step for every ticket cycle.
2–3 days
Research cycle cut — survey build to synthesised findings
AI used to build the study structure, then analyse the output from CSV exports. What previously took days of manual review and write-up now takes hours. Human validation layer retained throughout — outputs reviewed for accuracy before any finding is actioned or triaged to a product initiative.
–3 steps
Manual steps removed from the design completion flow
Three steps eliminated: artefact compliance check against WoWs, stakeholder summary writing, and CSV export interpretation. Each was a context switch pulling managers out of design work. None required human judgement. Confluence and Jira AI integrations configured via terminal and trained on team naming conventions and artefact standards.
Cross-study
Research compounding — findings linked across initiatives by surface
Building toward a model where every product initiative automatically surfaces relevant research from studies that touched the same surface area — participant counts, comprehension rates, placement findings. Research stops being a one-time deliverable and becomes a live input to the backlog. The value compounds with every new study.
🛣️Pipeline — Commercial & Operational Value
5→1
Event card variants consolidated into one canonical component
Five isolated card variants — each maintained separately, each requiring its own design and engineering changes — replaced by a single configurable model. One change now propagates across all surfaces. Reduced time-to-market and eliminated a class of parallel maintenance debt invisible to leadership until it was quantified.
9 Templates
Market display architecture — all surfaces governed
9 clearly named, component-agnostic templates define every product surface. Replaced an ad-hoc approach with a shared architectural language across design, engineering, and product. Persistent right-hand panel on web is one output of this architecture — the template model is the structural change.
4→1
Interaction model consolidated — fewer states, fewer defects
Four separate interaction lanes — each with their own conditional states and error edge cases — being consolidated into a single unified model. Fewer states means fewer defects, faster QA, and a more predictable experience across bet types and markets. Operational cost reduction framed as a design decision.
0→12
Product initiatives driven by design in 3 quarters
Design was not writing product initiatives. Three quarters of stakeholder influence work — hero journey surveys, structured research, cross-functional workshops writing from real user problems — shifted design from delivery function to roadmap contributor. 12 product initiatives in, more in pipeline.