Context
CFC (Cash for Cars) is Copart's consumer platform for selling damaged/old vehicles.
Problem
"Users keep dropping off midway and finalizing each offer requires multiple follow ups which is expensive"
3–6 calls per transaction were required to finalize pickup details
Out of the 275k users only 10% made it to the final offer screen
Okay, so people aren't satisfied with their offer but why?
Research
To understand what users were actually experiencing I conducted 4 rounds of usability testing.
I conducted usability interviews with one another designer to validate the assumptions set by the team and pinpoint the exact emotional and decision making moments during the entire journey and what exactly triggers trust or distrust?
My point of confusion
Every participant said the flow was easy to use but kept asking me follow up questions.
01
Users need clarity to trust the offer and proceed
The platform was missing important information that is crucial for sellers in feeling confident in moving ahead.
old final offer screen
"Is this final? Will the offer change after verification and when do I get paid?"
02
Lack of control leads to anxiety
Most of the questions in the flow were simply yes/no which made the sellers confused and anxious about what detail is taken into account to determine their cars value.
old damage question screen
"I replaced my side mirrors so they do not match, does that affect my offer?"
03
Selling a car is a huge emotional and financial decision that CFC does not account for.
Most platforms including cash for cars focus on speed and giving an "instant" offer but what users care about is fairness and thoroughness more than getting a quote in 10 seconds.
old homepage hero section
"A lot of websites don't ask so many questions so this was nice, felt reliable."
Data analysis
To improve the experience holistically, I wanted to first understand the accuracy of the pricing model to make sure we are giving out the most competitive offers.
I pushed for multiple meetings and collaborated with the product manager and data analyst to understand how the pricing model calculates vehicle value, we were able to pinpoint the exact minimal data needed to generate an accurate and early estimate.
Most important details
Mileage
Title type
Extent of damage
Area of damage
Least important details
Keys available
Pincode
Starts or drive
Disqualifying details
Has a lien or loan
Missing parts
User flow
Changes to the flow and questions to improve trust and accuracy.
01
CFC does not accept cars with loans so I moved the question up to clean up the funnel data and prevent sellers from wasting their time and effort.
Finding out at step 11 that you're ineligible destroys trust and guarantees user never comes back plus this was messing with funnel conversion data as were counting leads that could never convert.
Question no. 3 right after License plate
Do you have a loan?
Question no. 11, the last question before final offer
02
Image integration for extent of damage.
The business wanted to integrate image upload at the end and fasten verification later because they were afraid adding it earlier would lead to drop offs since it is a high effort activity.
My light bulb moment
What if we added vehicle upload as an alternative to damage questions with the option of skipping?
Without any visual proof the model assumes the worst case scenario leading to lowball offers but this would allow the AI to price the car closer to its true value.
Pros
- Avoids price changes/surprises later on.
- Users have more control over the answers because they can "prove" the condition of their vehicle, instead of a simple yes or no.
- More competitive prices so higher chances of leads converting.
- Self service, users don't have to manually select check-boxes for every scratch, dent, and mechanical issue.
Cons
- Anxiety Over AI Judgment.
- Blurry images, harsh shadows or bad angles can cause the AI to fail.
- Drivers might feel self-conscious about uploading photos of a messy interior, a dirty garage.
- Self service, users don't have to manually select check-boxes for every scratch, dent, and mechanical issue.
Design
I am happy to share the designs for this over a call. If you are interested please reach out.
