Project
Your Mobile Exert Right Next to You
The project's vision was to revolutionize T-Mobile's digital customer experience by seamlessly integrating a generative AI chatbot with the new T-Mobile website and native app. By mimicking the in-store mobile experts(ME)' personalized assistance through chat and voice interactions, the hybrid website dynamically adapts its content in response to chat interactions, offering an enriched and user-centric experience.
*Mobile Expert (ME) = Mobile Expert serves customers at T-Mobile retail stores
Info
My Role
Technical Product Designer
UX Researcher
Timeline
Aug 2023 - Dec 2023
Team
4 Software engineers, 2 PMs, 1 UX Researcher
E2E Process
Create Innovative idea to production
Out team rapidly assess our concepts and propose the most promising idea to T-Mobile's leadership
Background
Despite digital platforms, 80% of new customers visit retail stores
This trend highlights a gap—our digital offerings need to more effectively replicate the in-store expertise and guidance that our Mobile Experts provide, addressing the users’ needs comprehensively.
Problem
“Why can't T-Mobile's current platforms deliver a complete digital switch experience?”
Solution - “Revolutionizing digital engagement by blending AI with the genuineness of in-store interactions.”
We crafted a prototype that seamlessly integrates the ChatGPT AI model with the latest T-Mobile public data. This design mimics the experience of engaging with in-store staff (ME) for potential customers, all achieved through intensive user interviews in just 3 weeks.
Mixed UX Research Methods
How we got there?
User persona - “Who’s our user?”
User Story - “What is prospective customer’s switch process?”
Product direction
To make the prospective customer experience a seamless switch experience we decided to build a chatbot feature using generative AI.
90+
Ideas through 2 rounds of ideation sessions
I facilitated UX workshop sessions with internal and external stakeholders to ideate solutions.
Design
To evaluate our idea quickly through MVP test, I defined key scenarios and features with PM. We had 3 weeks sprint to build fully working demo application including design, development, and QA test.
Hand Sketches - “What’s the first image in our minds”
User Flow- “From where to where?”
3 week sprint - “Focus on key scenarios and features for MVP test”
Since we had limited time to create a fully working demo application, I had to find the balance between implementation effort and user experiences. I worked closely with engineers and pm to discuss feasibility and priorities.
Challenge 1: How can we obtain data for personalized recommendations without disrupting the chat experience?
Pros ✅:
Before initiating the chat experience, we can gather essential information to offer personalized recommendations.
Cons 🚫:
This onboarding process is primarily beneficial for first-time users. We may also need to determine the user's intention, whether it's bill checking, making a payment, or making a purchase.
Stakeholders’ opinions
Pros ✅:
The personalized quiz can align with the information provided in other cards.
Instead of redirecting customers to a separate page, they can remain within the chat interface.
We can cater to both the needs of existing and new customers.
The quiz can be tailored based on the ongoing conversation.
Option A: Onboarding before chat start
Option B : Integrate personalized quiz in chat experiece
Software Engineer 🧑🏻💻
“ For offering T-Mobile rate plan or device recommendations within the chat, Option B presents a smoother user flow. ”
Product Manager 💁🏼♂️
“I prefer Option B, which incorporates personalized quizzes directly into the chat process. Additionally, the interactive card-flip feature is engaging and effectively conveys to customers that their responses directly influence the recommendations provided.”
Design Decision
After our stakeholder meeting, we settled on Option B for the design. This choice ensures consistency in both UI and overall user experience. With the integration of the personalized quiz, there's no need to redirect users to another page. Moreover, it addresses a wider range of use cases, from bill payments for existing users to rate plan recommendations for new users.
Option B : Integrate personalized quiz in chat experience
High fidelity design
MVP Test result (KPIs)
100%
Task success rate
Unlike T-Mobile current platform, 3 different user groups were able to complete their tasks without any mobile expert’s support
Go Production
“Yes!” from leadership and proceed to production
Digital Innovation Lab’s success metric is make conceptual idea to production
80%
Faster Average interaction duration
From collect required information or complete a task, total decision making time reduced from 12.5m to 2.5m
Currently, this project is in the production phase!
From Aug 2023 - Dec 2023