Redesign Range Functionality — Recommendations Engine
Redesigning a core product interaction that users had abandoned for Excel
This project was part of a "Design Thinking in Action" portfolio presentation for Workday, showcasing how design thinking methodology was applied across three complex enterprise design challenges. The range functionality redesign at dunnhumby was one of the most impactful — taking a critical data table interaction that was so complex users abandoned it entirely, and transforming it into something intuitive, accessible, and genuinely useful.
Tools used to establish the audience & identify user needs
Interviewed users and shadowed their usage of the tool to observe real-world behaviour — seeing firsthand how users struggled with the existing interaction model.
Surveyed users to quantify behaviours and usage patterns, turning qualitative observations into measurable data to build the case for redesign.
Collated user data to build a detailed persona that captured the range planner's goals, digital literacy, experience levels, and key pain points.
Facilitated a design workshop with internal users, the PM team, and development to create user flows and identify what was known and unknown about the problem space.
Problems and user needs identified
- Existing product was complex and hard to use — most users gave up and exported to Excel
- Learnability was low and accessibility was not considered in the original design
- Lots of unused functionality that added noise without adding value
- Users were time-poor and needed fast, decisive tools to complete tasks
Steve's workday is often stressful as his focus is spread across several categories. He juggles his time between creating Customer Decision Trees (CDTs) and collating insights from different dunnhumby products.
- Grow category sales and margin through deeper understanding of product relationships
- Optimise shopping experiences by considering the customer mission when building planograms
- Streamline product ranges to improve profitability, basing decisions on CDT insights
- Build trust by presenting robust supporting evidence for proposed changes
- Drive consistency across stores
- Takes a long time to build and iterate the CDT then translate it into the planogram
- Existing product was complex, required extensive ramp-up, and was not accessible
- Time-poor — needs fast tools to complete tasks efficiently
Solving 6 interaction states — intuitively
The existing design relied on a dense visual key to communicate six different states — using combinations of cell backgrounds, border styles, side icons, and indicator dots that users simply couldn't decode in the flow of making commercial decisions. The interaction model was so complex that advisors admitted they exported to Excel instead.
After the research phase and all-day workshop, I designed a new interaction model that replaced the visual key with clear, labelled action controls — inline in the table, without requiring any external reference.
Product should remain in the recommended range as-is.
Exclude from range — user confirms to keep product out.
Product is currently within the recommended range.
Product is currently outside the recommended range.
User overrides recommendation to force product into range.
User overrides recommendation to force product out of range.
The design also introduced a Reset action — allowing users to revert any override back to the original recommendation, giving them confidence to experiment without fear of losing the baseline.
Agile retrospectives driving continuous improvement
The closure phase was embedded into the agile process — rather than a single project close-out, the team ran retrospectives at the end of every two-week sprint using Trello to capture what went well, what needed to change, questions and discussion, and action items to carry forward.
- People love it
- Hit the launch date
- Increased app usage after launch
- Were able to deploy improvements quickly after main launch
- Scope not clear with a super tight deadline
- User confusion around the existing app
- User confusion around how sending updates would work
- Product's breakdown of related changes to teams didn't work well for developers in practice
- No clear definition of success/failure for the project
Action items from retrospectives were fed directly into the next sprint — ensuring the design continued to improve iteratively based on real-world usage and team learnings.
From Excel workaround to a product users actually love
The redesigned range functionality was received positively by users — the retrospective feedback captured "people love it" and "increased app usage after launch." By removing the complex visual key and replacing it with clear, inline action controls, the product became genuinely usable for the first time.