Reshaping search navigation to dramatically improve product discovery

Woolworths is Australia's largest supermarket chain. Search and category pages are critical navigation tools enabling customers to discover and purchase products from a diverse range of over 30,000 items across groceries, fresh produce, household goods, and specialty dietary options on website and mobile web platforms.

Challenges

Woolworths' search and category pages suffered from fragmented navigation that actively hindered customers' ability to find products. The previous design scattered navigation controls across a sidebar and a separate filter box, creating a disjointed experience with measurable customer impact.

Before-state metrics revealed the severity: filter usage sat at just 4%, and critically, 57% of filter applications were unintentional, customers accidentally applied filters without realising it, leading to frustration and abandoned searches. Filter discovery was so poor that most customers never found the tools available to them.

Research participants highlighted the friction:

"I can never find the filters—I have to scroll all over the place and half the time I don't even know they're there." — User testing participant, July 2022

"Sometimes I think I'm seeing all the products but then I realise something got selected and I'm only seeing a tiny portion. It's really frustrating." — User testing participant, July 2022

The business impact: missed revenue opportunities as customers couldn't effectively navigate the product range, particularly during an inflationary climate when showcasing specials was strategically critical.

My Role

I led design and research for this project from June through October 2022, working within a 14-person cross-functional team spanning Atlas (front-end) and Sherlock (back-end) squads. I mentored junior UX designer Victoria Xu throughout the research phases whilst maintaining accountability for all design documentation, specifications, and quality assurance.

My leadership extended beyond the immediate squad to stakeholder engagement with squad leads Meredith Murphy (Sherlock) and Adnan Masood (Atlas PM), navigating technical constraints across Angular front-end and back-end search services. The design completed in October 2022, though front-end capacity constraints delayed production release until March 2023—a 5-month gap requiring sustained engagement to ensure implementation fidelity.

Research & Discovery

I led a comprehensive research programme including Voice of Customer analysis, competitor research (Baymard, Nielsen Norman Group), and extensive technical discovery with engineering teams to understand Angular front-end and search service constraints.

The research methodology comprised three rounds of moderated user testing between July and September 2022, totalling 25 sessions split between Victoria and myself. Round 1 (10 participants, 45 minutes) explored expectations around filters, facets, toggles, and sorting behaviours. Round 2 (10 participants, 45 minutes) evaluated usability and comprehension. Round 3 (5 participants, 15 minutes) validated the "Sold by" filter for Everyday Market products.

Key insights shaped the design direction:

Research revealed customers employ highly varied search strategies depending on context. When asked to find lactose-free cream, 4 of 10 participants searched "lactose free cream" directly, whilst 6 of 10 searched "cream" then applied filters. This validated the need for flexible, discoverable navigation tools rather than prescriptive paths.

"I'd probably look at the filters available to see if there was one about food. I'd probably look at all filters to see if there is a filter that will be able to select all food." — User testing participant, August 2022

Filter categorisation emerged as problematic—9 of 10 participants couldn't distinguish between "Allergens" and "Dietary and Lifestyle" categories. This indicated a need for clearer information architecture in future iterations.

"I think it would be under ethical or sustainable choices, like on The Iconic website. Dietary and Lifestyle doesn't really make sense because it's the lifestyle of the person, not the chicken." — User testing participant, July 2022

Reductive filtering logic was strongly preferred for allergens and dietary requirements, with participants expressing that additive filtering felt "considerably risky" for allergy-related searches.

Feature prioritisation emerged clearly: customers primarily used price-related sorting (unit price, low to high), with strong demand for the Specials toggle. Filter count should reflect total active filters rather than dynamic counts, and all filters should remain visible in a carousel when screen space is minimised.

Design Solution

I unified the scattered navigation controls into a cohesive chip-based system that increased discoverability whilst reducing unintentional filtering. The solution comprised three layers of progressive refinement, each validated through user testing.

Layer 1: Tab content switching

Tab navigation enables seamless switching between product, recipe, and article content types. This replaced the previous sidebar approach which had been removed during a parallel global navigation uplift, leaving no clear way to switch between content types. Testing validated that 9 of 10 participants successfully discovered and used the tab system:

"I found the recipes tab easily and it made sense where it was." — User testing participant, August 2022

Layer 2: Faceted category refinement

Faceted search bars surface popular product groups (e.g., "Chicken Breast," "Chicken Thigh," "6-pack," "12-pack"), allowing customers to recognise relevant categories rather than recall specific filter names. This reduces working memory demands and accelerates product discovery.

Testing revealed strong preference for imagery in facets, with 8 of 10 participants stating images helped avoid confusion, particularly for products with ambiguous names.

"I think pictures would do well here, especially with pasta—not everyone has a good culinary language so they can recognise the pasta that they want." — User testing participant, August 2022

Layer 3: Advanced filtering

The unified chip row consolidates sorting and filtering into accessible, touch-friendly targets. All participants easily applied price sorting, and the total active filter count in the "All filters" button was clearly understood by 9 of 10 participants.

"The filter count makes it clearer. I can see at a glance how many filters I've got active." — User testing participant, August 2022

Special considerations: Specials toggle

The Specials filter received dedicated chip visibility due to strategic importance during inflationary climate. Testing confirmed the toggle was "easy to use and understand," with high anticipated usage in natural settings.

Implementation & Obstacles

The project delivered through two phases: Phase 1 introduced the Specials toggle (target: 30% usage), and Phase 2 unified all navigation controls (target: increase usage above 4% baseline). I created comprehensive design documentation and provided ongoing QA support through the phased release.

The most significant obstacle was navigating Angular front-end and search infrastructure constraints that prevented several planned usability improvements. Technical limitations blocked dynamic filter visibility based on search context and more sophisticated filter categorisation. These constraints required design adaptations whilst maintaining the core experience goals.

"Majority of participants struggled to find 'Free range' in the 'Dietary' filter category... A few participants expected there to be a 'Category' filter so they could find the different types of chicken." — Research findings, August 2022

The engineering resistance to providing early visibility into technical constraints caused rework. In retrospect, I should have pushed harder for earlier technical discovery to surface these limitations before committing to specific interaction patterns in testing.

The 5-month implementation delay between design completion (October 2022) and release (March 2023) created an additional challenge. The Sherlock squad (back-end) completed their work first, but front-end implementation had to wait for Atlas squad capacity. Maintaining design fidelity over this gap required sustained engagement and documentation rigor.

Results

The project ran as an A/B experiment from 19 August 2022 through 9 November 2022, spanning 12 weeks with phased rollout (1%, 5%, 20%, 100%).

Navigation engagement transformed

• Faceted search interactions: +106% increase (measured across experiment period, 4.2M control sessions vs variant)
• Filter chip menu interactions: +171% increase (measured across experiment period)
• Unintentional filtering eliminated: -57% reduction by requiring explicit filter application action
• Specials toggle adoption: 35.6% usage vs 30% target (met objective)
• Overall navigation tool usage: 4.91% vs 4% baseline (met objective, measured across 83-day experiment)

Business impact

• Cart additions post-filter: +22% increase, signalling improved search utility (measured across experiment period)
• Revenue per visitor: Sustained ~2× lift post full-launch (November 2022 onwards, comparing 711K daily average variant visitors vs 682K control)
• Conversion rate: Up to 3× lift during early phases, sustained 2× improvement at scale

Recognition

Amanda Bardwell (Managing Director WooliesX) — Post launch, November 2022

"Amazing results... great improvement in utility and super simple design experience—not easy to achieve."