Exploring how to improve substitutions in Sainsbury’s online grocery shopping experience.
MY ROLE
I contributed to secondary research, user interviews and observations, using journey and empathy mapping to identify pain points. I created wireframes and low- and high-fidelity prototypes aligned with Sainsbury’s design guidelines, refined them through usability testing, and helped define the concept’s limitations and next steps.
USER RESEARCH
We began with secondary research into shopping behaviours and the colleague picking process, followed by interviews with two customers and two colleagues, one observation session, 13 online survey responses and ethnographic research. Our focus became improving customer–colleague communication to support more suitable substitution decisions.
Key findings
Items-per-hour (IPH) targets put pressure on substitution decisions.
Colleagues lacked customer context and direct access to preferences or feedback.
Business priorities could conflict with customers’ expectations around quantity, freshness and shelf life.
Customer
Goals
Receive the exact items she needs
Feel in control of substitutions
Trust that her order will meet her expectations
Pain Points
Limited time to check substitution emails
Frequently has to reject replacements at the door
Picker
Goals
Choose suitable substitutes with confidence
Spend less time searching for items
Make choices that satisfy customers
Pain Points
Pressure to meet items-per-hour (IPH) targets
Uncertainty about following system suggestions or personal judgement
CHALLENGE
Customers reported receiving unsuitable products or incorrect quantities. Limited time to review delivery details could also leave them accepting replacements they could not use.
Pickers had to balance performance targets and business priorities with customer preferences they could not easily access, making it difficult to choose replacements that met individual expectations.
USER JOURNEY
Customers struggled to identify alternatives while browsing and were dissatisfied when order changes or deliveries included unwanted substitutes. Colleagues faced similar difficulties when picking unavailable items, often choosing the closest available option to meet performance targets.
User Needs
As a loyal Sainsbury's customer…
I want to be informed in advance if my items are out of stock
I want the picker to understand my preferences
I want the delivery process to be transparent
As a new picker at Sainsbury's…
I want to reach my KPIs
I want to know in advance which items are going to be out of stock
I want to be aware of customers' preferences in advance so that I can pick what they want
IDEATION
We used Crazy 8s, “How Might We” statements, SWOT analysis and assumption mapping to explore design possibilities. A prioritisation matrix helped us assess ideas against customer and colleague pain points and business goals, guiding us towards a final concept that we developed visually.
Customer manages their shopping list and plans their weekly grocery shopping online via the Sainsbury's app.
Searching their items on app and notice one item to be "Likely out of stock".
Allows substitutions and selects tags that align with their requirements. AI offers recommendations based on past preferences, giving them the option to choose from or add custom tags.
Customer is now stress free after placing their desired order preferences.
The order is received in Sainsbury's system, making customer preference tags visible for pickers to select items.
Pickers can use substitution suggestions and customer tags to quickly find the best available item, reducing search time and maintaining IPH levels.
Customer on the other hand can check their order status on Sainsbury's app.
Items are delivered with satisfactory substitutions leading to a seamless shopping experience.
Solutions
Our concept combines stock alerts, order tracking and preference tags to help customers understand what to expect. Preferences also appear in the colleague picking system, supporting more informed substitution choices.
Low-stock notifications
Alert customers when items are nearly sold out, giving them time to consider alternatives.
Order tracking
Show the order’s real-time progress from checkout to delivery, keeping customers informed throughout.
Smart substitution tags
Let customers set preferences such as “vegan” or “brand loyal” to guide AI-generated substitution suggestions. These tags are also visible to pickers, helping them choose replacements that reflect customer needs.
USABILITY TESTING
We tested the final high-fidelity prototype to validate assumptions, assess usability and understand its value for both groups. Participants completed role-specific tasks using think-aloud, followed by a SUS questionnaire and semi-structured interviews.
Customer tasks
Search for an item.
Explore substitution options.
Complete checkout.
Review an order and reject an unwanted item.
Colleague tasks
Receive an order.
Find the customer’s substitution preferences.
Locate suitable substitutes.
Testing Feedback
Participants wanted clearer information about potential price changes and refunds, advance notice of the reported 12-item limit and one-week lead time for bulk orders, and an order status page that clearly distinguishes in-store collection from doorstep delivery.
LIMITATIONS
Too many low-stock warnings could frustrate customers and lead them to abandon their shop. Accurate predictions require up-to-date stock tracking, historical data and integration with existing systems. Errors in AI recommendations could still result in unsuitable replacements.
Future directions
Next steps include testing complementarity as a recommendation metric by assessing how well an alternative fits with other items in the basket, improving AI accuracy and exploring real-time shelf monitoring to support substitutions across stores. The system could also extend to wider stock management tasks, with potential to improve efficiency and profitability.
Copyright © 2026 Chia An Liu
Let's connect!