01
Overview
02
User Research
We began by conducting surveys, interviews, and an observational study at train stations to better understand user pain points. Key challenges included:
App Complexity: Users often struggled with navigating the app, especially when they were in unfamiliar surroundings like stations.
Language Barriers: Tourists and non-residents had difficulty using the app due to a lack of multilingual support.
Station Navigation: Users found it difficult to locate platforms, services, or handle unexpected schedule changes.
Limited Assistance: With fewer help counters at stations, users had to rely on the app for information, and long queues were an issue.
03
Problem Statement
Problem Statement
First-time users, non-residents, and tourists struggle with navigating the SBB app and train stations, particularly when it comes to booking the right ticket, locating platforms, and receiving timely information. This is exacerbated by language barriers and fewer physical help counters at stations.
Design Challenge
How might we design a voice assistant that simplifies the app's functionality, provides navigation assistance, and offers real-time updates, while overcoming language barriers and the challenges of busy, noisy station environments?
04
We explored solutions to help users in complex environments:
Multilingual Support: The voice assistant would support key languages, including English, German, French, and Italian, and offer easy-to-understand commands.
Real-Time Assistance: Users could ask the assistant for train schedules, platform directions, and even discover discounts.
Station Navigation: The voice assistant would help users locate platforms and services within the station, offering directions to things like luggage storage and restrooms.
Push Notifications: Instead of interrupting train journeys with voice updates, we considered replacing real-time updates with notifications that could be checked when convenient.
05
We created wireframes and a flow for how the voice assistant would be integrated into the app. The assistant would support both text and voice input.
Once the design was validated, we built interactive prototypes:
Voice Command Interface: A button in the app would allow users to activate the voice assistant. Commands could include "Book a ticket to Zurich," "Where is platform 5?" or "What discounts are available?"
Multimodal Feedback: The assistant would provide both spoken responses and on-screen text to cater to different environments.
06
Test
We tested the prototype with a range of users to see how they interacted with the voice assistant in different environments.
Key Hurdles & Solutions:
Intrusive Real-Time Updates:
Issue: The voice assistant provided real-time updates on schedule changes and platform information, but this was intrusive in the quiet train environment. The Swiss tend to dislike loud noises, especially in confined spaces.
Solution: We replaced voice updates with push notifications that users could check at their convenience, reducing noise in the train environment while still providing timely information.
Difficulty Hearing Responses in Noisy Environments:
Issue: Users appreciated the hands-free voice input, especially when rushing between platforms, but the voice-only responses were hard to hear in noisy stations.
Solution: We added a visual component—large text on the screen—to display the assistant's responses. This helped users digest information quickly when walking, waiting at escalators, or when they had trouble hearing the audio response.
Voice and Text Input Adaptability:
Issue: Users in noisy environments (stations) found it difficult to interact with voice alone, whereas users on quieter trains preferred text input.
Solution: We reworked the system to allow both voice and text input. This flexibility allowed the assistant to adapt to different environments: voice input for busy stations and text input for quieter, more relaxed settings like on the train.
Accent and Speech Variability:
Issue: Users with various accents, especially non-native speakers, had issues with speech recognition accuracy.
Solution: We fine-tuned the voice assistant to better recognize a wide variety of accents and dialects, making the system more inclusive.
07