New app to fix a broken workflow for 600+ truck drivers
Company
My Role
Sole Designer
Team
PM, product lead, operations team, process team, engineers
Platform
iOS App
Timeline
2 months
Responsibilities
Design Direction
Research
Interaction Design
Visual Design
Prototyping
Overview
Carvana sells millions of cars a year, and vehicle transporters are the ones who move those cars across the country. They're a critical part of the logistics network.
But the app they depended on daily was rated 2.4 out of 5, the lowest of any internal tool at Carvana. It was slow, hard to use, and constantly crashed.
The business also lacked reliable data. There was no chain of custody during vehicle transport, and Carvana had little control over improving the tool.
I led the design of Carvana’s first internal driver app. This was a 0→1 product built from scratch that defined how trips are executed and how data flows through the logistics system.
Our goals
Business
Reduce costs by replacing the third-party app and improve operational efficiency.
User
Improve the driver experience by making trips easier to manage and complete.
Aligning on expectations and defining what mattered for MVP
At kickoff, I worked with product and operations to align on what problems we were solving and what success looked like.
The team had many issues they wanted to fix, but we were already months behind schedule. To help the team stay focused, I asked them to define a clear MVP.
MVP phase
Trip details
Staging status
Scan and inspect vehicles
Load/unload vehicles
Fail vehicles
Swap driver contact info
Future phases
Loader flow
Unload sorting logic
Enhanced notification
Confirmation of preload
Recommendation to leave behind vehicles based on vehicle’s fit and weight
How this new app fits into the logistics system
Cargo Mobile sits between trip planning and downstream teams.
Upstream, the Transportation Management System plans trips and sends trip, vehicle, and route data into the app. This limited what information we could show drivers.
Cargo Mobile then captures execution data in the field. This includes vehicle scans, departure and arrival confirmations, and damage reports. That data flows to planning teams so they can adjust schedules and to repair teams so they can prepare earlier.
I worked closely with product and engineering to map these inputs and outputs. This ensured drivers had the information they needed to complete trips, while downstream teams received reliable data to do their jobs.
Trip
Route
Vehicle
Transportation Management System (TMS)
Cargo mobile app
Scan vehicle
Location
Damages
Signals
Planning & repair teams
One flow that scales for 6 trip types
The logistics system supports six different trip types. Each one has different tasks and information needs.
I gathered requirements and edge cases for each trip type so the new workflow could support all of them. These included standard trips, round trips, swaps, donuts, and multi-city trips with three or four destinations.
By understanding this upfront, I was able to design one flow that could scale across different trips and support edge cases.
6 trip types
Standard
A → B
Same truck
Round trip
A → B → A
Same truck
Swap
A → B → A
Different truck
Donut
A → B → C
Same truck
Multi-city
(3 dest.)
A → B → C → A
Same truck
Multi-city
(4 dest.)
A → B → C → D → A
Same truck
Understanding how drivers actually work in the field
To design an app that actually worked for our drivers, I needed to understand how they move through trips day to day.
I went out to the field and talked with drivers and learned how they work in real conditions. A few interesting insights came out of it:
Drivers are paid by the miles driven, not for completing tasks
Many are not tech-savvy
They work outdoors, often with gloves and glare
Loading vehicles was the hardest part of the job, especially for new drivers
The most important insight was that drivers are only paid for miles driven. And so adding two new tasks meant taking more of their unpaid time. This pushed me to design a fast, efficient workflow while still capturing the data the business needed.
I also identified a gap in the original product brief around vehicle loading. Drivers receive limited training, yet loading is one of the highest-risk parts of the job. I pushed to add in-app guidance and an easy way for drivers to get support when they needed it.
Paid by the mile
They’re paid by the mile, not the hour → quickly getting on the road is important
Not tech-savvy
Most drivers aren’t tech-savvy → simplicity over cleverness
Rough conditions
They wear gloves, work in sunlight, and use their personal phones → design needs to be robust
Loading is hard
Loading vehicles is the hardest part, especially for new drivers → provide help where needed
Adding new tasks without creating too much friction
One major change in the new workflow was adding two required tasks: vehicle scanning and damage reporting. These steps were needed to close gaps in inventory tracking and chain of custody.
But they also added new friction to the driver’s workflow. To minimize this impact, I explored multiple workflow options to find the most efficient and intuitive approach.
After comparing three different flows, I landed on a solution that allows drivers to scan and inspect vehicles one by one, then load all vehicles at the end. This approach improved efficiency while maintaining data accuracy.
Flow 1
Flow 2
Flow 3
Scan vehicle #1
Scan vehicle #2
Scan vehicle #n
Inspect vehicle #1
Inspect vehicle #2
Inspect vehicle #n
Damage?
Load vehicle
Load successful?
Damage?
Load vehicle
Load successful?
Damage?
Load vehicle
Load successful?
My thoughts on this flow
Makes drivers go through the vehicle lineup twice, which slows them down.
Leveraging AI to explore more ideas quickly
With these insights, I began exploring different user flow concepts. To move quickly, I shared the product brief and my own design prompt with AI to generate a wide range of ideas.
Many of the concepts were too complex or didn’t hold up in real-world conditions. But one direction stood out: a dashboard-style flow that clearly laid out all required tasks in a clear, linear way.
This approach worked well for drivers who are not tech-savvy and need simple guidance, with an interface that is easy to use in field conditions.
I shared this early concept with stakeholders to gather feedback and align on a direction. It became the foundation for the designs, which I continued to refine to meet real needs.
Dashboard layout
Balancing driver and business needs in my explorations
I explored multiple iterations for each core screen and workflows.
Across the designs, I was constantly balancing speed versus data quality, flexibility versus structure, and simplicity versus comprehensiveness.
For example, I explored separate versus combined steps for scanning and damage reporting tasks. I chose to combine them into a single flow because it matched how drivers worked in real life.
On the Trips screen, I chose clarity over showing too much information so drivers could quickly see what mattered. On the Load Vehicles screen, I explored batch loading for speed but later chose individual loading to reduce errors.
For every decision, I referred back to driver needs and focused on creating designs that were simple, intuitive, and easy to use.
Trips
Trip details
Load vehicles
Final design
Version 1
Reason for choosing the winner
Clear info matters more than saving space in this case
Navigating trade-offs with stakeholders
Throughout the project, I worked with stakeholders to navigate many trade-offs. User needs and business needs often conflicted, so my job as the designer was to help the team understand the impact of each decision.
I pushed back on decisions that would hurt drivers without clear business value. But I also made compromises when the business value clearly outweighed the impact on drivers.
By referring to user insights and real-world constraints in the conversations, I was able to help stakeholders see the bigger picture and align on the right trade-offs.
Trade-off #1
Scan + inspect
Trade-off:
Keep scan and inspect tasks separate vs. combine them for efficiency
Final direction:
Product wanted two steps. But it didn’t make sense in practice. I decided to combine scan and inspect into one seamless flow to increase efficiency.
Separated tasks
Combined tasks
Trade-off #2
Load vehicles
Trade-off:
Batch load for drivers’ speed vs. single load for operational accuracy
Final direction:
Drivers wanted to load multiple vehicles at once because that’s how they do it. But operations was concerned it would lead to mistakes in reporting damages.
Eventually I conceded since this friction protected the integrity of the damage data and the additional time should be negligible.
Batch load
Single load
Final designs
This is where everything comes together. After exploring, testing, and refining, these are the final designs. The new Cargo Mobile app lets drivers complete trips with ease, accuracy and confidence.
Vehicles list
BOL
Tasks are as clear as 1, 2, 3—with only the info you need and extra help when you need it.
Scanned vehicles (5 of 9)
2021 Ford Mustang
Red
CA325478
VIN
Does this vehicle have any damages?
Yes
No
2023 Toyota Camry
White
CA325478
VIN
Does this vehicle have any damages?
Yes
No
2021 Chevrolet Camaro
Does this vehicle have any damages?
Yes
No
2023 Honda Accord
Does this vehicle have any damages?
Yes
No
2022 Nissan Altima
Does this vehicle have any damages?
Yes
No
2021 Tesla Model S
Does this vehicle have any damages?
Yes
No
2021 Subaru WRX
Does this vehicle have any damages?
Yes
No
2021 Dodge Charger
Does this vehicle have any damages?
Yes
No
Continue
Scanned vehicles (1 of 9)
2021 Ford Mustang
Red
CA325478
VIN
Does this vehicle have any damages?
Yes
No
Report Damage
Where’s the damage?
Select all that apply
Roof
Bumper
Side panels
Glass
Upload at least 1 photo
Add
Sideview mirrors
Other
Submit
Cancel
Scan and inspect one by one or in batches. Whatever works best for the driver.
Support when they need it, from reviewing loads to loading vehicles
Design impacts on some of the metrics we care about
The product brief listed a dozen metrics that I kept in mind while designing. However there are no actual data yet since the app is still being implemented at the time of the writing.
But here is how the design will potentially impact the following metrics:
On-time departures & arrivals
I designed a simple, efficient flow that lets drivers complete tasks quickly and get help when issues come up
Explicit departure & arrival signals
I designed buttons that drivers have to interact with before they can successfully depart or arrive at the facility
Trip damage rate
Drivers are now required to verify vehicles and report any damages to create a clear chain of custody
Want to see more of my work?
Check out my case studies, or reach out to me at
