Skip to Content
Case Study

At-Seat Food & Drink Delivery on Trains

As the sole designer, I challenged management's proposed solution and designed the UK's first at-seat food delivery service on trains, doubling catering revenue within 6 weeks of launch.

RoleUX Lead & Product Owner
Timeline6 weeks
Year2020
TeamAvanti Stakeholders, Vendor Developers
+100%Catering revenue
+95%Customer satisfaction
+93%Staff satisfaction
01

The Brief Was Wrong

When Avanti West Coast took over from Virgin Trains, new management identified an opportunity to increase standard class catering revenue. Their proposed solution: a "trolley tracker" to help passengers locate the trolley and decide whether to wait or walk to the cafe.

I owned this project end-to-end: strategy, budget, product decisions, all UX research, and every design artefact from journey maps to final UI screens. A PM supported on logistics and a white-label supplier handled the build, but all strategic and design decisions were mine.

The problem was that this contradicted research I had already conducted. Customers didn't need to find the trolley: they needed food brought to them.

Photo of existing trolley service on train: cramped aisle, limited selection

What the research already showed

Previous customer and staff research (surveys, interviews, on-board shadowing) had surfaced a clear picture of the real barriers:

"I don't want to lose my seat or leave my belongings"
"By the time I get to the cafe, they're out of what I want"
"The trolley is heavy and hard to maneuver through crowded carriages"
  • Passengers treat their seat as their space: leaving it felt risky
  • Expectations had shifted: Deliveroo and Amazon had raised the bar
  • Staff found trolleys physically demanding with limited stock range
  • Most passengers didn't know where the cafe car was or what it stocked
Research findings summary: customer and staff quotes from previous research

The challenge: Convince stakeholders their proposed solution missed the root problem, and secure buy-in for something more ambitious: in a company with no in-house development team.


02

Research & Discovery

Rather than starting from scratch, I synthesized existing internal research and layered in competitive analysis to build the case for a different approach.

Existing Research Synthesis

I collated data from previous customer surveys, staff interviews, and on-board observation sessions. The pattern was consistent: the seat was the customer's anchor point. Everything else: the trolley, the cafe, the menu: was secondary friction. The insight that shaped everything: passengers expect Deliveroo-level convenience even at 125mph.

Affinity map from research synthesis: customer and staff pain points clustered by theme

Competitor Analysis

I analyzed food ordering across airlines (BA, Emirates), delivery apps (Deliveroo, Uber Eats), hotel room service, and international rail operators. The finding was significant: no UK rail operator was doing at-seat delivery. This wasn't just a product improvement: it was a first-mover opportunity.

Competitive landscape matrix: ordering experience vs transport context

Proposition Canvas & Stakeholder Workshop

I created a proposition canvas mapping the gap between current capabilities and user needs, then facilitated a workshop with executives to present findings and challenge the trolley tracker brief. The data made the case. By the end, we had executive sign-off to pursue at-seat delivery as a UK rail first.

Proposition canvas showing current state vs desired state: used in stakeholder workshop

03

Strategy & Design Principles

The product had two users with different contexts: a passenger on their phone trying to order without thinking too hard, and a staff member on a moving train managing multiple orders at once. Both needed to be designed for independently.

Mobile Web, Not Native App

QR code access via mobile browser. No download friction, works on any device, easier for a vendor to maintain. The trade-off: slightly less performant, no offline mode: was acceptable given the short journey durations.

Design for Movement

Every decision was filtered through the constraint of a moving train: large touch targets, high contrast for varying light, minimal text input, glanceable information for staff. Physical context drove interface decisions more than anything else.

Ruthless MVP Scoping

Six weeks with a vendor development team meant hard prioritization. The must-haves were QR access, full menu with photography, real-time stock availability, payment, staff order queue, and seat-based delivery logic. Loyalty integration, group ordering, and dietary filtering were pushed to phase two.

Story map from prioritization workshop: MVP scope vs phase two backlog

04

Design & Prototyping

Customer mobile app: menu view, product detail, cart, and order confirmation

Customer Interface

The customer flow had to feel effortless: scan QR, enter seat number, browse, order, done. I kept the information architecture flat: home menu, product detail, cart, checkout, status: with large product photography and minimal text input. The seat number entry included a contextual helper ("Look for the number above your head") after testing revealed occasional confusion.

Customer app screens: seat entry with helper text, menu browse, and checkout flow

Staff Tablet Interface

Staff needed to manage incoming orders while moving through carriages. The tablet interface prioritized glanceability: a live order queue with seat numbers prominent, one-tap completion, and a visual seat map for locating passengers quickly. Large buttons accounted for potential gloves and the physical demands of the role.

Staff tablet screens: order queue, order detail with seat finder

Vendor Handoff

With no in-house developers, I worked directly with the white-label supplier's team. I built a component library with design tokens, wrote detailed interaction specs and edge case documentation, and ran daily design reviews during the build phase. Where technical constraints required compromises, I made real-time decisions about what to flex and what to protect.

Component library: buttons, cards, order states, and seat map elements

05

Validation & Iteration

I tested prototypes with 12 train staff members, 8 frequent passengers, and 3 accessibility users before development began. The concept landed immediately: both groups understood it without explanation.

Task completion rate: High across all user groups from the first session
Key friction point: Seat number input and delivery time expectations

Two iterations came directly from testing: a seat number helper prompt reduced input errors, and adding an estimated delivery time with a caveat about train occupancy managed expectations before frustration could occur. Checkout was also simplified from five steps to three.

Before/after screens showing seat number helper and simplified checkout

Pilot Launch

Rather than a full rollout, we launched on two services for a 12-week trial to prove value before committing. This gave us real-world data without the operational risk of an immediate fleet-wide deployment.


06

Results & Impact

After 6 weeks of trial on two services, executives approved immediate rollout to all 150 daily services:

+100%Catering revenue

Incremental revenue vs the traditional trolley service: driven by higher basket sizes and more transactions per journey.

+95%Customer satisfaction

4.8/5 average app rating with 92% repeat usage on return journeys.

+93%Staff satisfaction

Staff reported reduced physical strain, better tips, and greater sense of control over their workload.

Industry-firstUK rail adoption

The service was subsequently adopted across all UK long-distance train operators.

Metrics dashboard: trial results across revenue, satisfaction, and usage
"Game changer. I can finally enjoy a meal without leaving my family to watch our bags."
- Passenger feedback, post-launch survey

07

What I Learned

🎯 Always validate the brief

The best solution often isn't what stakeholders initially ask for. Data gave me the standing to push back and propose something better.

🖼️ Show, don't tell

Prototypes and a proposition canvas convinced stakeholders where a written recommendation wouldn't have.

🚆 Design for the physical context

Train movement, varying lighting, gloves, and multitasking shaped every interface decision. The constraint became the design.

Ruthless scoping improves products

Six weeks forced prioritization that made the product better. Scope creep would have delayed launch and diluted the core experience.

👥 Design for all users, not just the obvious one

Investing as much in the staff interface as the customer app was critical: staff satisfaction directly drove service quality.

What I'd Do Differently

Earlier accessibility testing: some issues were caught late in development that required rework. Building in accessibility reviews from the first prototype would have been more efficient.

More route diversity in the pilot: the initial trial ran on premium routes. Passengers on shorter or more price-sensitive routes had different needs that surfaced only after full rollout.