UX Research across 2 retail platforms
Product Discovery · Mixed Methods · Usability Testing · Card Sorting · Emotion AI · NPS/CSI · Stakeholder Management · Research Operations · Service Design · B2C · eCommerce · Grocery
As Head of UX Research at X5 Digital, I led product discovery for a digital department with 20M MAU grocery retailer with retain chains. Scale comparable to Tesco and Sainsbury's combined running on one shared tech platform.
Over 18 months I built an iterative discovery system to test multiple concepts in parallel, using AI-driven emotional analysis to increase signal quality, while reducing research and product costs through smaller, faster validation cycles.
Over 18 months I built an iterative discovery system to test multiple concepts in parallel, using AI-driven emotional analysis to increase signal quality, while reducing research and product costs through smaller, faster validation cycles.
Project examples
Homepage Discovery and first widgets for the unified platform
Card sorting · Usability testing · Quant testing · Beta · CSI analysis · Sep 2023 → Nov 2024 · role: Researcher and servise designer
The problem wasn't just UX — it was prioritisation
The homepage is where delivery, loyalty, promotions, in-store marketing, and CVM compete for space - each owned by a team with its own KPIs. Same pattern across every surface. The answer was a widget platform: reusable content units any business unit could populate, refresh, or test live - deployed across the product wherever possible. The homepage shipped first. Discovery ran in three iterations to derisk it - from strategic frame to post-release tracking.
The problem wasn't just UX — it was prioritisation
The homepage is where delivery, loyalty, promotions, in-store marketing, and CVM compete for space - each owned by a team with its own KPIs. Same pattern across every surface. The answer was a widget platform: reusable content units any business unit could populate, refresh, or test live - deployed across the product wherever possible. The homepage shipped first. Discovery ran in three iterations to derisk it - from strategic frame to post-release tracking.
Delivery navigation concept
Card Sorting · Qualitative UT · Quantitative test with AI emotional response analysis · Nov 2023 → Q1 2024 · role: Researcher and servise designer
Using AI to settle what qualitative testing alone couldn't
Using AI to settle what qualitative testing alone couldn't
NPS as a product signal
Survey design · Open-text analysis · AI-assisted tagging · Backlog integration
I relaunched the NPS and CSI methodology: survey design, segmentation by feature area, open-text analysis with AI-assisted category tagging.
The biggest change was in how findings moved through the teams - we put the changes into the PDLS. Research, design and POs ran joint prioritisation for the critical issues. Also where design work on the relevant flows was already in progress, findings fed directly into it without a handoff gap.
Evidence:
I relaunched the NPS and CSI methodology: survey design, segmentation by feature area, open-text analysis with AI-assisted category tagging.
The biggest change was in how findings moved through the teams - we put the changes into the PDLS. Research, design and POs ran joint prioritisation for the critical issues. Also where design work on the relevant flows was already in progress, findings fed directly into it without a handoff gap.
Evidence:
- NPS score improved 30%
- 2025 target hit in Q3 2024
- Research directly informed backlog changes following churn and other studies
Context & Approach
X5 is one of the harder environments to move fast in. Two major retail chains with different audiences run on the same backend. Unified delivery, omnichannel loyalty programme, personalised promotions, CVM - user flows and business interests combined on the same apps.
Everyone wants to cut time-to-market by skipping design and research iterations. It can work for a while but you'll eventually find yourself in the worsening loop of growing systematic mistakes. So we replaced stakeholder decision meetings with iterative concept and feature testing. Fewer debates, more data — and discovery moved from a bottleneck to the thing that made the whole system faster.
The same approach — discovery as decision infrastructure, embedded across teams, measured end-to-end — is what I bring to the next role.
The same approach — discovery as decision infrastructure, embedded across teams, measured end-to-end — is what I bring to the next role.
More cases