expedia logo

CASE STUDY:
Product & Destination Recommendations System

PROJECT CHALLENGES

  • Limited Design Resources Two UX designers across time zones, neither fully allocated, and no dedicated content designer.

  • Compressed Design Timelines Approximately 3–6 weeks to audit, map, and deliver the initial schema, designs, and specifications.

  • New Mental Model Stakeholder discussions revealed this approach was largely new within Expedia, requiring alignment and reframing.

  • Pragmatic Execution As unknowns surfaced, we prioritized workable solutions that preserved the system and schema, even when aesthetics weren’t ideal.

ROLES & RESPONSIBILITIES
»
External & internal research
» Stakeholder reviews
» Competitive & comparative analysis
» Information architecture
» Content audit
» Content framework
» Component design
» Design manager
» Design system specifications & documentation
» Accessibility annotations & compliance

How it started

I knew from previous work on recommendations that Expedia’s were limited and sub-par from a competitive perspective. Every traveler visiting Expedia, regardless of preference, history, and interactions, were exposed to the same carousels and the same product cards.

The previous recommendations experience was described as, “Dumb”, “Unintelligent”, “In the way”, “Doesn’t apply to me, ever”, “Boring” and “Useless.” Internal stakeholders were frustrated with the lack of progress towards a unified, intelligent system while an increasingly disjointed experience was evolving as individual teams attempted to come up with their own personalized solutions.

When I was asked to incorporate an existing recommendation carousel into one of the Post-booking screens I managed, I knew it was the opportunity to begin to create and build a system that initially might solve 70% to 80% of issues and complaints.

Problem statements to keep us grounded

The Problem for Travelers

Travelers today expect personalized, relevant experiences when discovering and purchasing travel products. Expedia’s current experience relies heavily on generic search, with limited personalization and untimely or irrelevant recommendations—resulting in a one-size-fits-all journey that often drives travelers to explore alternatives.

The Solution for Travelers

For travelers, deliver intelligent, timely, and relevant recommendations across the journey—from planning through purchase. Presenting product and destination information in a consistent, uniform way reduces cognitive load and friction, enabling more confident discovery and decision-making.

The Problem for Expedia

Stakeholders partner with domain teams to deliver customized experiences across websites and platforms. However, these teams often struggle with inconsistent traveler experiences, inefficient knowledge sharing across touchpoints, duplicated effort, and slow adoption of domain features—resulting in fewer experiences shipped and tested within a quarter.

The Solution for Expedia

For internal teams, provide a modular travel exploration system that can be deployed across platforms. These reusable modules consistently surface product and destination information across lines of business, eliminating fragmented experiences, accelerating time to market, and enabling comparable performance benchmarking.

A real Shared UI approach & a content first strategy

Historically at Expedia, shared UI components were built as one-off solutions tailored to specific business needs—often limiting reuse across contexts. Early in the project, I chose a different approach: designing a robust, flexible system that could scale across lines of business and support roughly 80% of recommendation use cases.

I adopted a content-first, business-agnostic strategy—defining content requirements, mapping them to display types (e.g., EGDS, shopping patterns), and designing reusable card components and display mechanics (carousel, grid) for implementation across web and app.

What we’re trying to achieve

In order to avoid getting overwhelmed with input and commentary, we created a set of system guidelines or principles to share when necessary. This helped manage expectations outside the team, as well as illicit more relevant, focused feedback and suggestions.

The system is:

  • Line of Buisiness Flexibility
    Support product information across multiple contexts, formats, and lines of business.

  • Scalable, Consistent Design
    Flexible and scalable patterns that accommodate both static product details and dynamic content.

  • Maximal Reuse
    Reusable content configurations designed to work across contexts and recommendation use cases.

  • Continuity Through Familiar Patterns
    Limited to product information travelers already see and interact with on EG, reducing cognitive load.

  • Intentional Purpose
    Designed explicitly as a recommendation card—surfacing the right level of information to support confident traveler decision-making.

The system is not:

  • Not a duplication of existing product or search results cards

  • Not an attempt to support every possible content detail or content combination.

  • Not a reinvention of the Expedia Group Design System; it leverages existing components and patterns.

  • Not limited to a single line of business, context, or use case.

  • Not a “boil-the-ocean” effort—acknowledges necessary outliers and additional card styles where appropriate.

Audits & analysis

Extensive audits, both internal and external, were performed to capture the following details:

  • Location (e.g., Home page, Search results)

  • Mechanic (e.g, List, Carousel)

  • Display patterns and versions

  • Content types (e.g., Title, Rating)

Information requirements

I was familiar with information (content) requirements gathering in the Expedia ecosystem, having done a similar exercise for the Activities Search Redesign. Obviously this was a much larger, more extensive task.

The content had to be identified, broken down, cataloged and defined across the four lines of business: Lodging, Activities, Cars, and Flights; we later added Destination and Location (Geography).

The goal was to be able to build consistent, LOB agnostic content framework, starting with the "Core” set which could be scaled down (Abbreviated) or up (Long form) as necessary. The amount of content was referred to as “Size.”

Overall, the exercise looked something like:

  • Identify shared (common) and unique content types

  • Deconstruct content types to elemental level (e.g., “Review count (x,xxx)”)

  • Define “core” content types and corresponding card visual display; should be LOB agnostic to avoid bias

  • Subtractive and additive methodology to determine both minimum and maximum content and card sizes

  • Scale up and down to determine the scope of application

  • Create a working taxonomy

tde content design screenshot

Design requirements

Once I was confident in the overall framework, the next hurdle was applying the framework to different Lines of Business (LOB) to determine if it was viable and consistent. This design requirement phase included:

  • Comparing and aligning LOB content requirements and card displays to ensure appropriate scale and continuity; adjust and edit as necessary

  • Define content regions and corresponding content within said regions; lock content design

  • Pricing display spike, Badge display spike

  • Stakeholder education (show and share) and stakeholder inputs

tde design requirements screenshot
tde design requirements scale exercise screenshot

Design specifications

As the framework application neared finalization, I began to consolidate and lock design specifications; identifying regions of the card with corresponding content types, followed by text specs and logic. This was repeated for each line of business.

Specifications include:

  • Content size

  • Card size and corresponding dimensions

  • Card regions (Product, Pricing)

  • Card specifications and business rules

  • Container specifications for carousel and grid

  • Elements (and their variants) design specifications; LOB examples

  • Weekly status meetings with engineering

Figma components & variants

  • Creation of the local component library with variants

  • Build out each LOB example for card size and applicable content sizing

  • Weekly status meetings with engineering

TDE lodging card figma examples 1
TDE activities card figma examples 1

Accessibility annotations & compliance

To ensure proper assistive tech experience and remain in compliance with WCAG 2.0 standards, comprehensive accessibility annotations were provided and implemented.

The System Catalog

As part of the Expedia Shared UI Library, the recommendations system consists of the following card styles, configurations, and display capabilities, all prebuilt.

Configurations

  • Highlight/dynamic content

  • Primary title

  • Secondary title

  • Details list

  • Left/Right justified pricing

  • Badging

  • Title, Price, Rating, Pickup location (über)

  • With and without background fill or border

Card style

  • Long form

  • Foundation

  • Short form

  • Abbreviated

  • über abbreviated

Display type

  • Carousel

  • Grid (list)

  • Dialog overlay

Teams may use and/or leverage the 
pre-built card configurations in grid or carousel, or the individual elements. While benefiting from removal of varied experiences, speed to market entry, and comparable benchmarking features.

TDE catalog examples

Successful test results

In the wild locations and variants (in green outline)

  • Trip screen: Similar stays

  • Home screen mobile: Recently viewed, Destination, Last minute Flights

  • Home screen desktop: Recommended stays, Destinations, Last minute Flights

Results: Nearly $6M in annualized GDP identified during initial A/B testing.

“We built a dynamic platform capable of showcasing a wide range of experiences without ongoing development effort — but that scalability was only possible because of the content framework Jen designed. Her strategic thinking transformed a complex product challenge into a flexible, future-proof system, enabling us to scale efficiently and sustainably. Without her vision and structure, we would never have achieved the scale or efficiency we did.”

Neha mucchal

Neha Muchhal
Senior Product Manager, Personalization