O2O Experience Redesign
IKEA Food Mobile
A mobile app that moved the physical queue from IKEA's restaurants to customers' phones — and recovered the sales lost to it.
Role
UX Designer
Timeline
4 Months
Team
2 PMs, 3 Engineers, 1 Researcher
Tools
Figma, Miro, Jira
The Problem
A Business Mandate Behind the Queue
In 2021, IKEA China launched its Digital Hub — a company-wide initiative to migrate customer touchpoints from physical to digital. Membership, loyalty programs, and customer data were being unified under one ecosystem. The food court, generating significant daily revenue but operating entirely offline, was the natural next candidate.
The Messy M-Line
The urgency came from the floor. During peak hours, the M-line — IKEA's internal name for the winding physical queue — was actively costing the business. 30% of customers left without ordering. Average wait times exceeded 15 minutes. The physical queue had become the restaurant's biggest competitor.
This created a dual mandate: design a mobile ordering experience that recovered revenue lost to walkaway customers, while serving IKEA's sales goals — which meant menu layout and item sequencing mattered as much as the flow itself.
"The physical queue was our biggest competitor."
15+
Minutes
Average wait time during peak hours
30%
Walk-aways
Customers who leave without ordering
3×
Ordering Steps
Customers had to queue separately to order, pay, and pick up
What We Found
On-Site, Before the Screens
Before designing anything, we spent time in IKEA restaurants during peak hours — observing how customers moved, waited, and made decisions under real conditions.
.png&w=3840&q=75)
.png&w=3840&q=75)
Finding 01
The sequence was designed around the kitchen, not the customer
Customers decided what they wanted while browsing the store — well before reaching the counter. But the system forced them to queue first, then decide. The wait and the decision were happening in the wrong order.
Finding 02
Managing food and finding a seat was a two-person job
Food was ready before customers had settled. The system treated pickup as the endpoint — but for customers, it was actually the midpoint. They still needed a table. The ordering system had no way to accommodate that sequence.
Finding 03
Once in line, there was no recovery path
Adding an item or changing an order after joining the queue meant starting over. The system had zero tolerance for changed minds — and customers knew it, which created anxiety before they even arrived at the counter.
Design Direction
Move the Queue, Not the Kitchen
The M-line existed because ordering was tied to a fixed place and a fixed moment. Move that moment earlier — to wherever the customer already is in the store — and the line dissolves. That was the design direction: not to speed up the queue, but to make it irrelevant.
Principle 01
Integrate, don't replace
The store layout was fixed. Every screen had to work in context — legible under overhead lighting, usable while standing, and connected to the physical signage customers already relied on.
Principle 02
Convert waiting into browsing
A customer who orders while walking the store isn't waiting — they're shopping. The goal was to make the wait invisible, not just shorter.
Principle 03
Layout is a lever
Sales goals were part of the brief. Menu sequencing and combo placement weren't just UX decisions — they were revenue levers. That tension was designed into the system deliberately, not around it.
The Solution
A New User Journey
The research surfaced three breakdowns in the existing experience: customers were forced to make ordering decisions too late, the wait after ordering was anxious and phone-bound, and the final handoff between app and staff had no reliable protocol. Each part of this solution targets one of these.
The Digital Queue
↳ Ordering happened too late in the journey
Intelligent Ordering
Customers had already decided what they wanted while walking the store — not at the counter. We moved the ordering moment to match: a menu designed to confirm a choice already in progress, not initiate one under pressure.
- High-quality imagery for confirmation, not discovery
- Combo placement designed to serve the sales mandate without slowing the core task
The Menu Sequencing Decision
The Debate & Hypothesis
To meet the revenue mandate, the PM hypothesized that prioritizing high-value combo meals at the top of the menu would increase Average Transaction Value (ATV). I initially challenged this, advocating for prioritizing high-volume, popular single items (like $1 ice cream), assuming that frictionless, familiar choices would maximize total orders and overall sales.
Letting Data Decide
To resolve our differing assumptions, I proposed an A/B test to let customer behavior drive the decision, and designed the two variants.
Group A (Control)
The default sorting logic prioritized individual items based purely on their highest historical sales volume.
Group B (Variant)
The alternative logic prioritized high-value combo meals at the top of the menu to encourage upselling and drive up the ticket size.
Revenue over Volume
Combos Drove Higher Ticket Sizes
The data validated the PM's hypothesis: Group B (Combos) slightly decreased total users, but successfully increased the ATV, driving a +2.38% increase in overall turnover. This was a profound learning moment for me—it proved that in a physical dining context, optimizing for "perceived value" can outperform optimizing for "pure conversion/speed."
Status & Anxiety
↳ Waiting was anxious and tethered to the phone
Bridging the Physical and Digital
After ordering, customers needed the confidence to step away. When early prototypes caused anxious phone-checking, we rebuilt the screen around one extreme priority: an oversized pickup number. By actively redirecting attention to the overhead displays, the phone becomes a bridge to the physical room, not a replacement for it.
Pick-up Experience
↳ The digital-to-physical handoff had no reliable protocol
The Physical Handoff
The last friction point was identity: how does staff know which order belongs to which customer without a verbal exchange at a crowded counter? We replaced the confirmation step with a QR code that syncs directly with the kitchen display — making the pickup handoff work cleanly whether the bistro was quiet or overwhelmed.
The Transformation
Re-engineering the Journey
None of the furniture moved. No signage was torn down. The architectural layout of the restaurant stayed exactly as it was. What changed was the sequence of actions customers had to take within it — and when they had to take them. By shifting the active steps earlier in the visit, the time between ordering and eating became time customers spent doing something else.
Launch
25 Stores in 3 Months
The rollout plan was ambitious by design: deploy to over 25 stores — both new locations and stores already in operation — within a single quarter. The digital rollout team operated as a bridge between customers, store staff, and the product team, carrying on-site observations back into the design process in real time.
Internal rollout presentation — IKEA China Digital Hub
Impact
By the Numbers
¥25M
Revenue
Total sales generated through the Food Mobile platform
1.8M+
Transactions
Total orders completed through the app since launch
500K
New Users
New customers acquired through the rollout period
25
Stores in 90 Days
Deployed across 25 locations in under 3 months
The most counterintuitive result: making the menu harder to skim — by pushing combos above individual items — drove higher revenue than optimizing for speed. The data pointed to something more specific: in a dining context, customers aren't optimizing for speed — they're optimizing for perceived value. That's a different design problem entirely.
On the A/B test that changed the menu strategy
Reflection
This project taught me that O2O design has two surfaces: the one you can prototype, and the one you can't. The app was the visible part. The invisible part was the operational chain it depended on — kitchen staff, signage timing, network reliability in a large retail space. We had to design for both simultaneously, which meant working with store operations managers as much as with engineers.
The A/B test result was also a useful corrective. Our instinct was to optimize the menu for speed — show people what they usually order, get them out fast. The data disagreed. Customers given more prominent combo options spent slightly more and didn't drop off.
If I were to run this project again, I'd involve operations staff in design reviews from week one — not as stakeholders to present to, but as co-designers of the handoff experience.
Me on-site during the rollout.
Next Project
Google Fitbit — Celebrating Workout Achievements