2.5×
Customers using lists spent up to 2.5× more than those who did not
01 Case Study
Shopping lists.
Turning messy shopping intent into intelligent baskets
- Role
- Senior Product Designer
- Company
- Waitrose & Partners
- Timeline
- 4 months
- Focus
- Intelligent shopping lists / Behavioural UX / AI assisted basket building
My role
I led the experience strategy, interaction design, prototyping and validation approach, working across customer behaviour, data logic and basket-building flows.
- Defined the behavioural problem and customer planning patterns
- Designed the mobile list creation and basket-building experience
- Created prototypes to test input methods, editable suggestions and customer control
- Worked through recommendation logic, confidence states and refinement patterns
02 Why this mattered
Shopping lists sat at the centre of weekly grocery planning.
Shopping lists were one of the strongest behavioural signals linked to larger basket sizes and repeat purchasing, but existing list experiences still relied heavily on manual product searching and repetitive basket building.
65%+
of customers still created physical lists
56%
used hybrid planning behaviour, combining meals and individual items
New online grocery customers often started with incomplete or ambiguous intent
03 The problem
Real-world shopping behaviour is messy.
Customers rarely think in exact product titles or SKUs. Instead, they create fragmented reminders throughout the week: handwritten notes, screenshots, meals, family requests, generic reminders and brand references. Traditional grocery experiences forced customers to manually translate this intent into products one item at a time.
Slow basket building
Customers translated messy list items into products one at a time before they could start shopping.
Decision fatigue
Too many possible matches for vague inputs made planning feel harder than it needed to be.
Repetitive searching
The same items and brands were searched again each week instead of reusing existing intent.
Low confidence
Customers hesitated when they could not see, review or correct suggestions before adding to basket.
Friction for new customers
Less purchase history meant more guesswork and more manual work to build a first online basket.
How could we help customers move from intent to basket faster without removing confidence or control?
04 Behavioural insights
Designing around how people actually plan.
Planning often happened away from the supermarket. Inputs were messy, ambiguous and created across multiple moments throughout the week. The experience needed to support natural planning behaviour while keeping customers confident and in control.
01
Customers planned before they shopped
Planning happened across notes apps, paper lists, conversations, screenshots and meal inspiration. The experience needed to support behaviour that already existed in the real world.
02
Intent was highly ambiguous
Simple inputs like ‘milk’, ‘bread’ and ‘pasta’ required interpretation, not simple matching. Customers expected the system to understand context, preferences and likely behaviour.
03
Automation only worked when customers stayed in control
Confidence increased when recommendations were editable, alternatives were visible and quantities could be changed easily. Transparency mattered more than aggressive automation.
05 Constraints & trade-offs
Designing within real product and system limits.
The experience had to work with messy inputs, imperfect interpretation and an MVP delivery window. These limits shaped what we could ship and what we chose not to.
Constraints
Ambiguous list inputs
Physical lists were often vague, making accurate product interpretation difficult.
Unreliable OCR
Confidence varied with handwriting, formatting and image clarity.
Thin data for new customers
Without behavioural history, personalisation accuracy was limited early on.
Voice raised the wrong expectations
Spoken input suggested a conversational assistant rather than lightweight list capture.
MVP delivery pressure
Engineering feasibility and timelines limited how complex AI interactions could be in the first release.
Operational complexity
Availability, substitutions and category logic added backend and experience complexity.
Trust in recommendations
Automated suggestions needed to feel transparent, reviewable and easy to challenge.
Control over automation
Customers needed editability, not invisible decisions applied to their basket.
Trade-offs
Editable drafts over full automation
We prioritised AI-generated suggestions customers could review, swap and adjust before checkout.
Voice deprioritised for MVP
Conversational voice was removed from the first release to reduce UX complexity and unrealistic expectations.
Lightweight preferences first
Onboarding captured simple signals instead of building a full recommendation engine upfront.
Mixed inputs over rigid flows
Typing, image upload and quick capture supported how people already plan, rather than forcing structure.
Signals matched customer maturity
Popularity and behavioural cues supported new customers; purchase history mattered more for returning shoppers.
Validate before scaling AI
We prioritised fast MVP learning before investing in deeper AI-driven functionality.
The challenge was not matching products. It was recommending with confidence.
06 Hypothesis
If we could interpret messy intent, we could reduce planning effort.
We believed that if we could capture shopping intent naturally, interpret ambiguous inputs and generate editable draft baskets, we could reduce planning effort while increasing confidence, basket completion and repeat usage.
- 01
Capture shopping intent naturally
- 02
Interpret ambiguous inputs
- 03
Generate editable draft baskets
- 04
Keep refinement lightweight
07 Design principles
Reduce effort — not control.
AI should accelerate basket building without removing customer ownership.
These principles shaped how the experience balanced intelligence, trust and customer control.
1. Start with intent, not products
Customers think in meals and household needs, not exact product names.
2. Keep refinement lightweight
Editing quantities, swapping products and removing items needed to feel effortless.
3. Make the system feel collaborative
The experience should feel assistive rather than automated.
08 Solution
The solution.
AI that understands messy intent. An experience that keeps customers in control.
Understand anything
Natural language and notes are turned into smart, shoppable suggestions.
Refine with ease
Edit, add or remove items in seconds. Always your list, your way.
Seamless to basket
Everything flows straight to your shop, ready to buy.
09 Key product decisions
The decisions that made the experience work.
Trust came from restraint, clarity and control — not just automation.
01 We intentionally avoided full automation
Customers needed editable drafts, not invisible decisions. Recommendations stayed transparent and reviewable.
02 We prioritised mixed-input behaviour
People plan in many ways — typed lists, meal ideas, notes and screenshots — so the experience had to support messy real-world input.
03 We used confidence-based recommendation logic
High-confidence suggestions could speed up basket building, while more ambiguous inputs surfaced alternatives.
04 We designed for new customers first
New users had less history and more uncertainty, so the journey needed lightweight guidance and confidence-building support.
The real breakthrough was not automation. It was editable intelligence.
10 Validation and iteration
Validation and iteration.
We tested early concepts, refined the experience in response to feedback, and validated the solution with real customers.
Our validation process
- 01
Research
User interviews and concept testing
- 02
Prototype
Low-fi to hi-fi prototypes across mobile
- 03
Test and learn
Usability testing with customers
- 04
Iterate
Refine, retest and improve confidence
What we learned (and how we responded)
Customers expected flexibility
“I want to tweak items before adding them to my list.”
→ What we did: Gave customers edit control before adding to basket.
Visibility improved trust
“Seeing what was added helps me feel confident.”
→ What we did: Made suggestions visible, reviewable and editable.
Grouping improved comprehension
“Grouped suggestions are easier to scan.”
→ What we did: Grouped items by occasion and category.
Proof
What gave us confidence.
The direction was supported by a clear behavioural pattern: customers were already planning shops outside the product. The opportunity was to translate that existing intent into a faster, editable basket-building experience without removing customer control.
Behaviour already existed
Customers were already using paper lists, notes, screenshots, meal ideas and household reminders to plan shops before opening the grocery experience. This showed the product did not need to create a new behaviour. It needed to support one that already existed.
Ambiguity was the real problem
Inputs like milk, bread or pasta showed that customers rarely think in exact product names. The experience needed to interpret intent, suggest likely matches and keep alternatives visible where confidence was lower.
Control protected trust
Testing showed that customers were more comfortable with AI-assisted suggestions when they could review, edit, swap, remove and adjust quantities before committing anything to basket.
The proof was not that customers wanted automation. It was that they wanted less effort while still feeling in control.
11 Impact and outcomes
From fragmented planning to editable basket creation.
The concept connected a high-value customer behaviour with a faster, more flexible way to build baskets. Instead of asking customers to search one product at a time, the experience translated messy intent into editable, shoppable suggestions that customers could review, refine and add to basket.
12 Reflection
Automation alone did not create trust.
The biggest shift was not helping customers create lists. It was helping them turn intent into a basket.
Customers valued speed, but only when they retained visibility, flexibility and control. That insight reframed the experience from a simple list-making tool into an intelligent, editable shopping assistant.
Control built confidence
Editable suggestions made AI feel helpful rather than prescriptive.
Visibility increased trust
Customers wanted to understand what was suggested and why.
Intent became the real input
The product worked best when it responded to messy human planning behaviour.