Spicy Boys
Turning a 15-minute line into a 2-minute kiosk order
Zero spice errors. Sub-2-minute orders. After three hours of field research in 98°F Austin heat, I designed a self-service kiosk that runs orders in parallel and locks in spice choices on screen because the bottleneck was the window, not the fryer.
- Context
- Proactive Concept, Field Research
- Role
- UX Research, UX/UI Design
- Timeline
- 4 Weeks
- Tools
- Figma, Contextual Inquiry, User Interviews
- Platform
- Tablet Kiosk, Outdoor Food Park
Summary
100% of final-round testers (3/3) ordered in under two minutes with 100% spice-level accuracy. Spicy Boys' biggest bottleneck wasn't the chicken, it was the line. I spent three hours in the lunch-rush heat, designed a self-service kiosk from what I saw, and validated it with 9 unique, time-pressed workers. One change made the difference: a mandatory UI gate eliminated the truck's most common fulfillment error.
Core problem
One cashier, 98°F heat, and a line of fifteen. Manual ordering meant 15-minute waits, observed frequent queue abandonment, and spice levels lost in translation at a loud outdoor window: six of eight customers I interviewed had been burned by the wrong heat level.
Final outcome
Every tester ordered in under two minutes with 100% spice-level accuracy, successfully eliminating the truck's most common fulfillment error via a mandatory UI gate.
Context
Background
Self-service kiosks are proven at major chains but missing at independent food trucks. Spicy Boys operates in busy Austin food parks where lines form in direct 98°F sun. Customers are time-pressed and ordering in a loud, chaotic environment.
A kiosk extends their bold brand into a self-service touchpoint built for glaring sunlight and speed, without sacrificing customization accuracy.
The problem
- •The Problem: 15-minute average waits in 98°F heat. Observed frequent queue abandonment; roughly 1 in 3 potential customers walked away before ordering.
- •The Agitation: "Hot" means different things depending on the cashier. 6 of 8 customers interviewed had received the wrong spice level. Regulars lost trust.
- •The Solution: Offload order-taking to a kiosk where customization can't be misheard.
The approach
A self-service tablet kiosk that runs multiple order streams in parallel and locks in customization visually. Crucially, offloading the intake process frees cashiers from shouting over crowd noise, allowing them to be reallocated to higher-value tasks like expediting and quality control.
- •Mandatory Spice Selection: Gates checkout, eliminating the most common error.
- •Transparent Pricing: A sticky running total prevents checkout hesitation.
- •Confirmed Modifiers: Visual receipt before payment replaces failed verbal confirmation.
- •Order Numbers: Replaces name calls swallowed by crowd noise.
Scope
Project goals
Deliverable: A validated kiosk prototype in 4 weeks.
- •Parallelize order intake to fix the window bottleneck
- •Free staff from repetitive order-taking so they can focus on prep and expediting
- •Eliminate spice-level miscommunication
- •Survive a loud, outdoor food park with zero staff training
Constraint: Faster intake shifts pressure to the kitchen. Success was measured by throughput and accuracy, anticipating a future Kitchen Display System (KDS).
The process
01 · Empathize
Three hours in the heat revealed the real bottleneck
Observing the lunch rush showed the line was the product's weakest link. Active ordering took ~1.5 minutes, but the queue-to-food cycle hit 15+ minutes.
Customers weren't comparing kiosk taps to fry time. They were comparing their lunch break to an unpredictable wait.
- •Contextual inquiry: 3 hours observing peak lunch rush (11:30 AM–2:30 PM). I watched queue behavior, overheard customization conversations, and timed wait patterns.
- •Customer interviews: 8 intercept interviews (5–10 min each) with people waiting in line about ordering frustrations and self-service willingness.
- •Owner interview: Operational bottlenecks, staff training limits, and what it would take to trust a kiosk at the window.
Time pressure: Office workers on 45-minute breaks reported spending half their break in line. Several had timed it: 20+ minutes from queue to food. Unpredictable waits made people leave before ordering.
Spice miscommunication: 6 of 8 customers mentioned past disappointment with heat levels. "Hot" meant different things depending on who took the order. Regulars had stopped trusting verbal confirmation entirely.
Environmental friction: Limited queuing space, outdoor noise making verbal orders easy to mishear, and customers with dietary questions feeling guilty for holding up the line.
The owner confirmed the operational reality I was already seeing: faster order intake doesn't speed up the kitchen. Kiosks without a Kitchen Display System would flood the prep line. That constraint shaped how I framed success around throughput and accuracy, not false speed comparisons.
02 · Define
Behaviors and constraints that shaped every screen
The design was driven by two recurring behavioral patterns observed in the queue:
The Lunch-Break Worker: Needs speed above all.
The Spice Enthusiast: Needs consistency.
From the research, I narrowed the design work to three questions:
03 · Ideate
Why customization is a first-class step, not a modifier
Before any screens, I scoped the kiosk to one signature sandwich order with customization as a required gate in the flow, not a buried sub-menu. Four weeks meant perfecting the core path, not building mediocre versions of five features.
The architecture separates ordering from payment, forces spice selection before Add to Order, and ends with an order number instead of a name call.
03 · Ideate
Every screen had to earn its place
The user flow was ruthlessly prioritized. If a screen or interaction didn't directly mitigate time pressure, improve spice accuracy, or account for outdoor glare, it was cut from the prototype.
The core loop guarantees speed without sacrificing the customization accuracy that regulars demand.
04 · Prototype
Testing structure before brand polish
Low-fidelity wireframes to validate layout and interaction patterns before committing to visual design. I studied kiosk conventions from Shake Shack and Sweetgreen for patterns customers already understand, and Spicy Boys' own menu board for category structure and spice labeling.
These wireframes became the script for Round 1 usability sessions.
Step 1 of 3
Welcome
One primary action on entry. Truck name and time stay visible in the header.
04 · Prototype
The modal that eliminated the truck's #1 error
This is the core research fix. Required spice level and protein sections gate the Add to Order button. Users cannot proceed without choosing a heat level. Making it mandatory adds one deliberate tap, but it eliminates the "I thought I said medium" complaints that 6/8 interviewees reported. Optional removals sit below so required decisions stay visually dominant.
04 · Prototype
The validated flow: welcome to order number
Clickable high-fidelity prototype in Figma covering the complete journey. This is the exact flow 3 participants used in the final validation round. Every tester ordered in under two minutes with 100% spice-level accuracy.
Step 1 of 7
Welcome
Spicy Boys logo and a single Start Order tap target. No account, no friction.
05 · Test
Iterating from wireframes to a validated product
Testing was conducted on-site during the lunch rush over three weeks. I conducted guerrilla testing with 9 unique, time-pressed workers recruited directly from the queue (3 different participants per testing round). They were given a timed, unmoderated task: order a customized spicy sandwich and a side.
Round 1 (Week 2): Lo-Fi Wireframes
Goal: Validate category navigation.
Result: Users found items quickly, but all 3 participants skipped the optional spice modifier because it lacked visual weight. This directly led to mandatory gating.
Round 2 (Week 3): Mid-Fi Clickable Prototype
Goal: Test the mandatory spice gate and checkout flow.
Result: 100% accuracy on spice levels, but the 3 new testers hesitated at checkout because they couldn't track their running total while customizing items.
Round 3 (Week 4): Hi-Fi iPad Prototype
Goal: Validate speed and trust with full branding in a realistic outdoor environment.
Result: 3/3 participants completed the order in under 2 minutes. The sticky cart subtotal resolved the checkout hesitation flagged in Round 2.
05 · Test
Three crucial iterations that shaped the final product
- •Required spice selection: Users cannot add to cart without choosing a heat level. Result: 100% accuracy.
- •Sticky cart total: Persistent subtotal keeps pricing visible, addressing the checkout hesitation.
- •Order number: Replaced verbal name calls, which fail in loud food parks.
05 · Test
The bottleneck moves to the kitchen next
Speeding up front-of-house shifts the bottleneck to prep. The critical next phase is a Kitchen Display System (KDS) to throttle kiosk orders so cooks aren't flooded at lunch rush.
Results
Before: 15-minute waits, frequent queue abandonment, and widespread spice errors.
After (Validated Prototype)
The final prototype delivers the speed of a fast-casual chain with the accuracy regulars demand.
Looking back
Retrospective
Four weeks from first site visit to validated prototype.
Key Takeaways
Be where your users are
3 hours in 98°F heat revealed that the problem was trust and environment, not just wait time.
Required fields aren't always bad UX
Forcing spice selection added friction but prevented catastrophic downstream errors.
Constraints force clarity
Stakeholder pushback shaped the use of order numbers over names.
Systems thinking
Fixing the ordering bottleneck frees staff for higher-value tasks, but immediately exposed the need for a Kitchen Display System to pace the kitchen.
See it in action
Explore the final high-fidelity prototype and interact with the complete user flow.
Next case study
Taste Tester App
Designed a single spin mechanic that solves 3 distinct dining anxieties in a two-week sprint.
UX Designer (Class Project) · 2 Weeks