top of page
Machine Learning (AI/ML) Strategy Development
Led the Workshop in conducting thorough user research to create a future in-vehicle HMI experience with core values and principles for machine learning. This project led to another collaboration project in China and helped shape China Ford’s current voice strategy and system.
My Role
The Team
Time Frame
Project Lead
Workshop Facilitation
Project Management
Research Analyses
Strategy Development
Strategy Presentation & QA
3 Researcher (Agency)
1 UI Designer (Agency)
1 UX Researcher
1 UX Designer
3 System Engineers
1 Prototyper
Feb. 2022 - Jun. 2022
(5 months Project)
IMPACT
The Machine Learning strategy has a great impact on Business and Usability
Drove Product Strategy
“ The outcome of the Machine Learning strategy workshop can be a beacon for our long-term product strategy”
- Director, Product Development
Impacted China Ford’s current voice strategy and system
11
Big Category
1500+
VUI flow
Stats from China Ford
CONTEXT
Understand the goal and be the bridge between developing new strategy v.s. existing plans
The department had planned 3 workshops to determine what the company wanted to do as a strategy in Augmented Reality, Machine Learning, and Screen layout exploration.
I led a cross-functional team of industrial, UX designers, researchers, guest engineers, prototyper & agencies working remotely in the Machine Learning study.
The goal of the Workshop was not to design any screen or single experience but to develop a future AI/ML customer experience strategy and a set of design principles to guide product development.
It had lots of attention and stakeholders within the company. Since I was the lead of a brand new workstream, firstly I met with the product experts and invited them as guest team members to define the project scope and process together.
Led the cross-functional team
My focus:
Workshop Facilitation
Project Management
PROCESS & CHALLENGES
Remote collaboration, split the workstream and trust my team
The COVID pandemic changed the way we work, remote team collaboration became a new norm, and it had a great impact on the design process, especially in qualitative data analysis and conceptualization phases.
Therefore, after experimenting with different collaboration tools and platforms (Bluescape - Figma - Miro) our team has changed the strategy of one person leading one work stream, and the other primarily offering critiques and feedback. Trust among team members is crucial when work is split.
Define Scope
Collect Data
Findings
Strategy
Led:
our team
Stakeholders:
Research Engineers
Feature Owners
Feature Planning
Platform: BlueScape
Led:
Research Agency
Platform:
Online Video recording tool, Figma, Miro
Led:
UX Researcher
Platform: Miro
Led:
UX Designer
Platform: Miro
DEFINE SCOPE
Understand the product and target users
Product
117 1-page descriptors of concepts inspired by experience themes targeting brands, segments, and domains provide a structure of aspirations, Jobs to Be Done that Ford can target to deliver value to customers.
LUXURY
AUTONOMY
TRUCK
GUIDE ME
Target Users
Define Target Users from 69 themes, and 12 separate global ethnographic studies targeting different vehicle segments and attributes, for internal ideation and external evaluation of hypotheses.
RESEARCH SCOPE
Break the existing box, think from the experience
First, we looked into all domains and features based on desirability rankings then regrouped them into new categories by what users want for them.
User the categories as the key research domains to develop compelling use cases for the studies, while having the central question always in mind: How might we use data to personalize and tailor the in-vehicle experience to the individual?
Stuff Tracker
Voice
Personalize ADAS
Lane Guidance
Contextual Suggestions
Procedural Music
Smart Navigation
Vehicle Needs
Grow with User
COLLECT DATA
Remote interviews accompanied by stimulus
Remote interviews accompanied by stimulus (non-interactive prototypes with synthesized speech). 26 participants across 19 interviews Participants were recruited nationwide across all four key segments (sedan, CUV/SUV, truck, luxury).
FINDINGS
Convince stakeholders with data
The old studies and findings drove by the Engineer of the product which more focus on technology and features. In this UX Researcher led the analysis activity, I encouraged the team to think from the point of users' focus on their workflow and needs.
With a lot of persuasion and pitch along with user research insights to support. Our stakeholders understood and learned from our new findings.
Thoughtful & Pertinent Suggestions
Users appreciate thoughtful and pertinent suggestions or assistance both outside and in the vehicle.
Digital Assistant
Users prefer interacting with a DA that is authentic, respectful, accountable, contextually aware, and integrated with their other services and devices.
Integration with Customers’ Data
ML can provide added value but is not necessary to establish a foundation.
Trust
Ford has not earned their trust when it comes to technology experiences, and they are less forgiving of issues with Ford’s tech when compared to those provided by big tech companies.
Core Value
The core value in these suggestions comes from context-aware design and integration with customers’ data.
ML can provide added value but is not necessary to establish a foundation.
STRATEGY
Actionable strategy is the key to win
The new strategy suggests providing value to customers immediately, even before the machine learning infrastructure is ready to deliver. This three-stage approach shows the far-term direction but also provides short-term actionable to-dos that can start to work right away.
We also list out System Behavior V.S.Impact which was important for our stakeholders to understand in their familiar language for better collaboration.
v1
Integrated with vehicle
v2
Integrated with user’s data ecosystem
v3
Predictive suggestions
SYSTEM
BEHAVIOR
IMPACT
Phone app provides reminders to schedule an appointment
App connects with user’s calendar and schedules appointment
System provides reminders at the time most likely to be acted on
High value, simple task, hard for user to do by themselves
Medium additional value, schedule suggestions may not be aware of other factors not present in the personal data ecosystem
Small additional value, will need learning & feedback time to get it right, getting it wrong is equivalent to non-predictive experience (v1 or v2)
PRINCIPLES
Find opportunity from the industry's specificity and interests
We had given principles from two angles based on the car industry's specificity and the interest from internal.
Digital assistance is a hot topic at the time but facing lots of implementation difficulties due to the limitation of our 3rd parties supplies in the U.S. With the success of presenting this project to our own leadership, we got the chance to introduce the strategy to other regions of Ford market like Europe and China. And the China market luckily doesn't have to be constrained as the U.S. does. So we started another project together about Digital Assistants. and this project Impacted China Ford’s current voice strategy and system.
REFLECT & LEARNINGS
Work closely with cross-functional partners at the right time and check back regularly
It had lots of attention and stakeholders within the company about this topic. I met with the product experts and invited them in the very early stages to understand their needs and set the goals together. Invited them as guest team members to come back at project checkpoints to make sure we are single in our aim and let them experience the way we work and the reason why we work in a particular way.
The pre-work gave us a great result about the wide range of acceptance.
Mentoring others by asking the people-centered and goal-driving questions
My role in this project was to lead the team and drive the process instead of designing the pixel-perfect screen and collecting the research data by hand. After experiencing the difficulty of working as a big team online, I dedicated split the workstream and let one team member be in charge of one activity and regroup as a team to offer critiques and feedback.
During the meeting time, I emphasized our stage and work goals and focused on how to target, collect, or analyze questions and data that are people-centered and solve the needs of users.
Set the goal in stages and itemize the tough ones for implementation
Proposing a strategy for a hot and new topic in a company that is huge and has lots of legency like Ford is always a challenge. To have a realistic impact, I had to balance my audience's acceptance level and stakeholder's interest and workflow. In the end, I gave a three-stage plan starting with the right away actionable item and a vision for a long-term goal. It emphasized the area that had a higher priority in the middle stage.
Read More
bottom of page