Redesigning Copart CashForCars InstaQuote: the moment someone sells their car online.
Most people came here wanting to sell their car, and almost none of them did. This is what builds trust in the quote flow, and this is where the experience breaks.
What I did
I led research across 25+ FullStory replays, 4 user interviews, and 4 usability sessions, then mapped three core insights, redesigned the end-to-end flow, and pitched to Copart stakeholders. They pushed back on the first version based on pricing model constraints, so I revised within those limits and the second flow was approved.
Context
A 10% completion rate was costing Copart inventory at scale.
158.2KVisited
-23.9K 15.1% off
134.3KBegan details
-42.6K 31.7% off
91.7KFinished Qs
-6.9K 7.5% off
84.8KSaw offer
-68.4K 80.7% off
16.4KAccepted
Only 10% made it to an offer, a 10% completion rate costing Copart inventory at scale
The quote flow stopped before the offer did
Problem
The quote flow stopped before the offer did
Sellers thought they were finishing online, but the experience didn't collect enough upfront, so Copart's team picked up where the UI left off.
01
What was missing
Vehicle details the offer depended on never made it into the first submission. Nothing could move forward without someone verifying them.
02
3 to 6 calls per deal
Staff called sellers back repeatedly to fill gaps the flow skipped, not to close, but to catch up.
03
Cost to the business
Manual follow-ups slowed closes, frustrated sellers, and pulled the team into low-value work instead of acquiring inventory.
The calls weren't policy. They were a workaround for a flow that ended too early.
Nine in ten sellers dropped out before ever seeing an offer
CashForCars drew serious traffic every month, but the funnel didn't convert it.
01
~160K monthly visits
People arrived ready to sell, and most never made it to a quote.
02
Conversion cliff
Only one in ten reached the final offer screen. The rest left without knowing what their car was worth. Of those who did see a number, fewer than 20% accepted. 84,800 people got an offer and still walked away.
84.8Koffer screen viewed
68.4K dropped off80.7% drop-off
16.4Kaccepted
19.3% conversion
Key takeaway
Most people who saw a number still didn't take it. The offer screen wasn't the finish line, it was another drop-off.
03
Trust broke early
A damage question with no room to explain. A phone number before any value. The flow asked for commitment it hadn't earned.
Understanding who was dropping off, and why, meant starting with the people behind the numbers.
Target user
Who I'm designing for
People across the US who want to sell a car online. They want it quick, fair, and free of pressure. Skepticism is the default. They arrive having already compared prices elsewhere and already half-expecting to be lowballed.
Who they are
US sellers who want the transaction done without a middleman.
What they want
A fair number, explained, with no pressure to commit immediately.
What gets in the way
Doubt. About the offer, about the platform, and about whether their description of the car will cost them money.
Research
What happens when someone tries to sell their car online?
I used three methods to understand where the experience was losing people and why: 25+ FullStory session replays, 4 user interviews with people who'd sold or tried to sell, and 4 moderated usability sessions on the live InstaQuote flow.
01What builds or breaks trust when selling through an online platform?
02Where are the emotional peaks and decision points in the selling journey?
03What information and visibility do sellers need to feel the process is moving forward?
04How well does InstaQuote support completing a sale without calling for help?
Insights
Every method pointed to the same underlying problem: the flow asked for trust it had never built. Three themes explained why.
Insight 1
Transparency
Without seeing how the number was calculated, sellers assumed the worst.
The flow gave users a number with no explanation. Selling a car is too significant for that to feel fair. Several participants assumed a low offer meant the platform was taking advantage. They didn't see any evidence the number reflected their specific car.
I need a better value for my car. I know it's going to depreciate, but at least I want value for it. I don't want to sell it off at a ridiculous amount of money.
I replaced the side mirrors so they do not match, how does that affect my offer?
When I started looking at these things, I noticed that there was a wide price range, especially for similar cars. How do I know what mine is actually worth?
What this meant for the design: the offer screen needed to show its work, not just the answer.
Users didn't trust a modal to hold their offer. Several tried to screenshot it before doing anything else.
Nobody reads “Accept offer now” and feels calm. This is the moment they most need to think, and the copy tells them to hurry.
Insight 2
Control
Sellers needed to describe their car in their own words. Yes or no wasn't enough.
A yes-or-no damage question doesn't match how people think about their car. Sellers wanted to explain a replaced mirror, hail damage, or a door scratch. When the flow wouldn't let them, they assumed the offer would be wrong. They feared losing money because of it.
I want to also put in the severity of the damage to get a better offer.
It did not ask me the extent of damage, does it not matter?
What happens if I don't agree with the offer? When do you negotiate?
What this meant for the design: sellers needed to represent their car, not categorize it.
Yes or no forced a lie. Every user had a ‘but’ the form wouldn't let them say.
Insight 3
Trust
Sellers wanted to know what they were walking into before the flow asked anything of them.
Sellers came expecting to get a quote, not to hand over personal information mid-flow. When the phone number field appeared before any offer, most didn't know why it was there. Three out of five navigated away to find an explanation and couldn't find their way back.
I'd only proceed once payment processes, ownership transfer, and inspection steps are fully clear and documented.
There's still a part of me that's holding back. But if I see something, evidence beyond doubt that this is genuine, I have no reason to be scared.
Three out of five left the form to find out why it needed their number. None of them made it back.
User flow
A clearer path from quote to offer
I mapped the end-to-end InstaQuote journey after research and reframed steps where sellers dropped off or lost trust. We pitched the first version to Copart stakeholders. They pushed back: their pricing model couldn't process both external damage and mechanical conditions simultaneously, something we hadn't known going in. I revised the flow within that constraint, holding onto what research showed sellers actually needed. The second version was approved and now guides key screen design.
Transparency
Show how the offer is calculated before asking sellers to commit.
Control
Let sellers describe damage in their own terms, not yes-or-no gates.
Trust
Explain what happens next before collecting a phone number.
Design
Research, flow redesign, and pitch are complete. Wireframes, UI polish, and prototyping are in progress. This section updates as work ships.
What the design has to do.
The flow is approved and screens are in progress. What I know going in is that the offer screen needs to show its work, the damage questions need to let sellers speak in specifics, and the phone number needs to come after value, not before it, so the design earns what the research promised.
