Buying blind: $20k decisions, no test drive
ACV Auctions is a digital wholesale car marketplace. Dealers buy vehicles they've never touched, from sellers they've never met, on a time limit. The average vehicle sells for $18,000–$24,000. The goal isn't to make decisions faster, it's to make them more confident — for both sides. Misrepresent a vehicle and buyers dispute it. Underrepresent one and sellers take their inventory to Manheim.
The VDP is where that decision happens. Everything a dealer needs to bid — condition, photos, inspection data, pricing — lives in one place. Get it wrong and they don't bid. Get it right and they come back.
I was the only designer on a 27-person initiative across two teams (Marketplace & Inspections): 4 PMs, 20 engineers, 2 data scientists, ops stakeholders. End-to-end ownership — from defining the problem to shipping the system.
The VDP was making good vehicles look bad.
The VDP was a data dump, not a narrative on whether to buy or walk away. Every finding at the same visual weight, no positive framing, no baseline for what was normal. The design had evolved reactively: disclose everything, protect against arbitration. Over time, features were added, not designed. It looked thorough. It eroded buyer confidence.
Data without a story is just noise.
A clean, sellable unit presented identically to one with structural damage. Buyers had no way to distinguish "this is normal" from "this is a problem." The interface offered data. It offered no judgment.
Bids dropped, abandonment increased, sellers moved to Manheim and OpenLane. I brought a different diagnosis to leadership: this wasn't a UX problem, it was a trust problem. Good inventory was being undervalued because every vehicle looked like a risk. Fixing that wasn't a redesign. It was a platform-level investment, and I made the case for it.
We rebuilt how dealers read a car — not just what they see.
The same disclosure list reads completely differently depending on the vehicle. On a 120k-mile truck, 20 findings mostly mean normal for age. On a 3-year-old SUV, five findings hit differently. Equal visual weight was never the right call — dealers need to know what something costs, not just that it exists. I proposed a new condition model: separate tiers for known costs, open unknowns, and normal wear. It didn't exist anywhere in the product or the industry. It had to be defined, defended to legal, and agreed on with engineering before anything was built.
The hardest part: legal alignment
What helps a dealer act — a number, or a sentence?
Condition scores are the industry standard — every platform has one. My research revealed dealers couldn't explain the difference between a 4.1 and a 4.3. A number strips the context you need to act on it. Rather than polish a broken model, I proposed replacing it entirely with a narrative: an AI-generated paragraph that tells you what the vehicle is, what it'll cost, and what to watch out for.
Single owner, 20,150 mi — low for a 2022. Known recon: front-end damage and cracked head lamp — estimated $1,650. Three items require diagnosis: ABS fault, emissions readiness failures, and malfunctioning display screen — keep these in mind when you bid. Wear consistent with normal use.
The summary only includes what the inspection confirmed. If it wasn't documented, it doesn't appear. We agreed on that rule with engineering before anything was built.
The old VDP showed everything — and answered nothing.
The old VDP wasn't built around how dealers think — it was built around what was easy to add. Information piled up with no hierarchy. No market context, no seller history, and to run the numbers dealers had to leave the platform entirely. I defined five questions every dealer asks before they bid, and restructured the entire page around them. That framework became the template for how we'd think about information hierarchy going forward.
The specs, history, and options that set the baseline before anything else.
1FAFP31N57FML5AGU
Condition structured by severity — not a flat list of everything.
System fault detected.
Malfunctioning display screen
Cracked bumper, grille, trim + loose mirror casing.
Cracked head lamp
Paint work on bumper, fender, hood, liftgate
Irregular wear front-left and rear-right
Demand and velocity data for the dealer's own region — not national averages.
Seller track record so dealers know if the inspection can be trusted.
A profit calculator built in — no leaving the platform to run the numbers.
What we gained, gave up, and are still calibrating
A page that thinks like a dealer
The page is organized around the five questions dealers actually ask — in the same order they ask them. No hunting for what matters. The structure does the work.
A trust loop between buyers and sellers
Showing seller history and arbitration rates rewards good sellers and gives buyers a reason to come back. The data was already there — we just put it where the decision was being made.
The "neutral presenter" position
The old page showed everything and let dealers decide what mattered. The new model makes prioritization calls for them — tiering findings, surfacing what's costly. That's more useful, but it means we own those calls when they're wrong.
Seller control over how findings land
Structuring condition into tiers means the platform decides what's highlighted and what's routine. Sellers with a lot of Normal for Age findings worried they'd look worse than they were. We had to earn that trust too.
Useful defaults vs. false precision
The profit calculator pre-fills recon, transport, and retail estimates. Helpful — but dealers can anchor on numbers that don't fit their market. Useful and misleading can look identical.
Buyer transparency vs. seller comfort
Arbitration rates help buyers make better decisions but put sellers on edge. Where transparency builds trust and where it feels like public judgment isn't a fixed line — it needs ongoing calibration.
What changed
90-day results post-launch, compared to the same period the prior year.
The arbitration drop matters most — it's directly tied to the condition redesign and had the clearest business cost. The walk-away time is the one I find most interesting: dealers who passed did it faster, which means the page was helping them read a vehicle confidently in both directions.
We deliberately don't track time-to-bid as a success metric. A faster bid isn't a better bid — a dealer who spends four minutes reading a condition report and buys with confidence is exactly what we want. The sell-through rate and arbitration numbers tell that story better: dealers are buying the right vehicles and actually moving them.
Beyond the metrics: the condition model and trust framework we built here became the foundation for AI features that came after it. This wasn't just a page redesign — it was infrastructure.