Mark Jenkins : Sure. I could take that one. So I think we have a multiyear period, I would say, honestly, 2018, ’19, I think, late ’20 and 2021, where we operate pretty tight inventory. Average sale, call it, as low as the high 50s, up through the mid- to high 60s, and we’re — and that’s on average. We’re pretty consistently operating in that range. I think that we probably experienced a little blip with COVID in early ’20 and then certainly moved materially off of that normalized range in 2022 because we just overbuild inventory for the interest rate environment that ultimately came about and the sales that we ultimately ended up executing in 2022, we meaningfully overbuilt inventory for that. And so I think the — what we’re seeing with these cars above 90 days, I do think is really a function of us overbuilding into relative to sales and sort of maintaining that for most of 2022.
Now we’ve clearly adjusted, clearly marched out inventory. The cars that we hold relative to the cars that we’re selling are in a much more normalized range. As everything sort of works through, we think that’ll lead to a much more normalized average sale, a much more normalized share of cars sold in less than 90 days. And so I think really the name of the game is returning to our historical norms after a pretty significant outlier year in 2020 that really kind of culminated in Q1.
Operator: The next question is from Zachary Fadem with Wells Fargo.
Unidentified Analyst: This is pitch-hitting for Zach. Wanted to bucket advertising savings in a bit more detail. Can you break out how much you might be saving from shifting to different ad spend channels like more digital versus just absolute reductions in ad spend? And then separately, a more broader one on market share. As you right-size the business, who do you think is picking up some of the market share you might be giving up?
Ernie Garcia : Sure. So I think at a very high level, I think what we’ve tried to do in marketing is kind of manage or uncertainty as best we can. And so what that basically meant was pulling more away from digital channels than brand channels. And what I mean by managing our uncertainty is I think brand channels tend to have a long and difficult to measure payoff, but we think that payoff is significant. Building a brand is an incredibly difficult thing to do. It’s an incredibly valuable thing to do. Direct channels and various advertising channels vary in their level of directness, but direct channels tend to have a shorter, faster payoff that is much more measurable. And so as we’ve gone through this environment, we’ve tested many different channels.
Some of those tests take kind of a global form where we do large A/B tests of using a channel or not using a channel. Many of those tests take kind of a market level forum where we pick a subset of markets that look similar to other subset of markets and we run one marketing channel in one set of markets and a different marketing channel and a different subset of markets, and we try to get a sense for how those are going. And those sorts of tests are a little bit less susceptible to errors and attribution. And I think we’ve just tried to kind of continually learn because this environment has to be different. It’s been an environment where car prices are higher. It’s been an environment where our inventory is smaller. Both those things mean lower conversion.
It’s been an environment where there’s been less competition for various marketing channels, which means clicks are less expensive. And that varies by the marketing channel, even by the kind of sub-channel and side of any given channel. If we think about , for example, there are many sub keywords and sub kind of campaigns that you can run. So we try to be thoughtful about running many different tests and learning. I think that we feel like we’ve been pretty successful in cutting a lot of expense out. And I think we’re excited by that. I think we’re also starting to see that with GPU climbing up pretty significantly and our variable costs dropping pretty significantly, our contribution margins are starting to look better. And so I think some of those trades could change a little bit in terms of what marketing channels we’re supposed to be utilizing.
So we’ll keep learning. And I think over time, we will most likely grow marketing spend from here, and I think there’s a reasonable chance that it could even go up at the per unit level just given what’s most efficient given our higher level of GPU and our lower levels of variable costs. And then as it relates to kind of market share, what I would say there is I really think the most important point here is this market is enormous. And so we’re probably right now on the order of about 1% market share. We’re down from where we were, let’s say, two years ago, but we’re down a similar amount to the market overall. I think quarter-to-quarter and kind of year-to-year that there can be some variability in those numbers. But if you look back kind of to a more normalized environment, we’re probably down at a similar level to where the market was overall.
And so we’re probably about 1%. If you look at it over the last 6 months, 9 months, 12 months, we’ve certainly give up some. We were probably a little higher than 1%, and we’re probably back down to 1% now. But then where that is going is to a mix of the other 99%. And I think the most important dynamic there is just that this is a very, very large market. And now is not a time when everyone is focused on growth here and we get appropriate not to be focused on growth. But I do think that if we allow ourselves the intelligence, it’s worth thinking about what that will feel like again because we do have a differentiated customer offering that customers love. Our NPS is high. As we get more efficient, we’re seeing benefits to NPS there. There will be a time when it’s time to grow inventory again, it’s time to turn up marketing because our GPU is high and our variable costs are low, and it will be very difficult to replicate the machine that we’ve built, and we’re incredibly well positioned for that time.
So I think when that time comes, we’ll look to kind of take that volume from that — the entirety of that market again, that very, very large market. And the great news of being in the market that big is very few players will be able to identify exactly where it’s coming from because it is just such an enormous pool that we are drawing from. So I think it’s hard for us to say where the very small amount of market share that we’ve given up is gone, but I think we’re very well positioned to take it back when it’s time.
