Good cars buried in too much inventory.
ACV started with 20-minute auctions. Short by design — enough time to commit, not enough to overthink. This created real urgency. Dealers felt invested when they were winning. The clock ticking down kept them there. Sometimes it made them overbid to win. The tension was the feature.
As inventory scaled, auction relaunches spiked — good cars buried in the feed. So we kept stretching the window: 20 minutes in 2020 crept to nearly three hours by 2025, with options running all the way to 24. More time was supposed to prevent relaunches. The logic was sound. But we solved visibility and killed urgency — we'd traded relaunches for a platform that felt asleep.
Three thousand cars running. No moment to show up for.
Physical car auctions have been event-driven for decades. Sellers bring inventory. Buyers plan their day around it. There's energy: urgency, expectation, the feeling that something is happening.
We'd replicated the inventory without the event. Sellers listed whenever. Dealers scrolled whenever. A good car and the right buyer could both be on the platform and miss each other entirely.
The instinct was to extend listing windows. More time meant more chances to connect, fewer cars needing to relist. The data said otherwise: longer windows didn't sell more cars. The ones that sold closed in about 87 minutes on average; the ones that drew zero bids ran 138 minutes and still went nowhere. And nearly a third of first-run auctions — a vehicle's best shot — attracted no bids at all.
Nobody brought that pattern to product. I did. Not with a solution, just a question: if a third of first runs get zero bids and stretching the window doesn't fix it, the problem isn't time — it's that 5,000 cars are running at once with no moment to show up for. That reframed everything. Instead of asking how to make longer windows work, we started asking when inventory should surface. That's what led to drops: scheduled release windows designed to bring the event back to the experience.
The drop
Drops transformed the dealer experience from reactive to proactive. Instead of scrolling whenever, dealers plan around when inventory surfaces. They know what's coming, they can preview it, and they show up prepared. The platform went from always-on to event-driven.
Previewing upcoming inventory changed dealer behavior as much as the drops themselves. Dealers weren't just reacting to what surfaced — they were planning around what was coming. That pre-drop engagement was the clearest signal the model was working.
Where does personalization live in the drop experience?
We had buyer data—previous purchases, market, inventory type. We could recommend vehicles at the top of every drop. The question was whether to show only recommendations or the full drop with picks highlighted. We chose the full drop. Dealers got their recommendations immediately and could browse everything else. Personalization without access feels like restriction. The logic was visible, not hidden. That earned the buy-in.
Sellers knew their inventory. They didn't know when to list it.
Sellers have been on the platform for years. They have routines and they trust them. But timing was instinct, not data. We knew when buyers were most active by vehicle type and region. The solution was to put that information in front of sellers at the moment they listed.
Sellers know their inventory best. We gave them something new: information about when buyers were actually active. By showing the reason behind our recommendation, it felt helpful instead of bossy.
What we gained, gave up, and are still calibrating
A marketplace with rhythm
Buyers have a reason to show up at specific times. Sellers have guidance on when their buyers are actually there. The platform has structure that continuous listings can't create.
Urgency from structure, not design tricks
Time pressure in a drop window is real — fresh inventory, active competition, the clock means something. That's more durable than countdown timers.
Always-open simplicity
The old SRP was always there. Drops require dealers to understand a schedule. The learning curve was real — dealers who engaged early retained better than those who didn't.
Seller freedom over timing
Sellers can still override the recommendation. But the system has a point of view now — and sellers who list off-schedule see fewer bids.
Personal buyer match vs. marketplace visibility
Relevant buyers bid more. That means some sellers get in front of more qualified buyers, others fewer. Balancing individual experience with healthy marketplace dynamics is ongoing work.
What changed
First 90 days post-launch, compared to the same period the prior year.
The drop model made behavior measurable in ways the old continuous feed couldn't. Dealers either showed up for a drop or they didn't. Sellers either accepted the recommendation or they didn't. That clarity transformed how we read the data.
The metric I watch most carefully is whether dealers come back. A marketplace rhythm only works if it becomes a habit. Once dealers learned the schedule, they had a reason to return.