Amazon – Interactive Styling Installation

Retail · Voice & Conversational UX · Physical–Digital Experience

Year
2019

Conversational UX starts with data structure, not dialogue

Context and product

As part of an interactive in-store installation, Amazon created a voice-guided styling experience powered by Alexa. In a futuristic “dressing room” environment, customers could describe their style preferences and receive personalised denim recommendations from Amazon’s extensive catalogue.

The exeprience:

The experience was designed to feel conversational and exploratory rather than transactional—inviting users to talk about style, not search for products.

Problem space

Traditional product discovery relies on screens, filters, and typing. In a physical retail environment, this approach would not have worked.

The challenge was to:

  • Enable hands-free, voice-led discovery

  • Translate complex product data into natural, conversational questions

  • Create an experience that feels intuitive, playful, and fast in a noisy public space

  • Deliver reliable results despite very real-world constraints such as acoustics and crowd noise

Approach and process

The largest part of the work happened before any actual interface design:

  • Sorting, structuring, and prioritising product data
    Identifying which attributes (fit, cut, rise, wash, style) were meaningful in a spoken interaction

  • Designing the conversational logic
    Mapping how Alexa would guide users from vague preferences to a concrete recommendation, and help and deal with error recovery and ambiguity

  • UX writing & voice scripting
    Writing and refining voice prompts, confirmations, and responses
    Translating and adapting the entire experience in English and German, accounting for linguistic and cultural differences, rather than direct translation

  • Scenario design & testing
    Different test scenarios were developed to anticipate possible user behaviours such as unclear answers, changes of mind, and interruptions.

Design focus and decisions

  1. Turning structured product data into human-friendly dialogue

  2. Designing questions that feel open but still lead to meaningful results

  3. Supporting voice interaction with visual cues in the physical space

  4. Keeping cognitive load low in a public, time-limited setting

Real-world implementation challenges

One of the most interesting aspects of the project was the on-site setup at the exhibition.

Test scenarios had been conducted in controlled environments, but the live setting introduced:

  • Unexpected acoustics

  • Background noise and overlapping conversations

  • Different user behaviour in a public space

This required last-minute adjustments to voice prompts, pacing, and error handling to ensure Alexa could reliably understand users and keep the experience flowing.

Outcome and impact

My role

UX Designer focusing on conversational UX, data structuring, UX writing, and on-site implementation.
Worked closely with engineers, data teams, and exhibition staff.

The installation demonstrated how voice interaction can meaningfully support product discovery in physical retail environments.

It was a particularly rewarding project that combined conversation design, information architecture, UX writing, localisation, and hands-on delivery—from concept through live implementation.