Carvana: Ramping
The sell-side is badly mismodeling growth and operating leverage.
Carvana reported its third-quarter results on November 2nd. This was another strong quarter of Step 2 progress. For those unfamiliar with Carvana, Step 2 is part of management’s three step plan to fix the company’s financials and return it to growth.
Step 1, completed in the second quarter of last year, returned the business to positive adjusted EBITDA.
Step 2 is focused on achieving efficiencies to maximize the unit economics aka adjusted EBITDA per unit while managing retail volume to a more or less flat level.
Step 3 will be about returning to volume growth.
I’ll get straight to the point here: despite being an 11-bagger last year—and the single best performing stock in the entire world—CVNA is still a phenomenal investment. This is an incredibly compelling long-term story, which I have covered previously, but in this post I’m actually going to cover why this is super attractive short-term story. I almost never have a specific short-term view of my holdings or rely on catalysts, but between the fundamental changes going on, the still widespread investor skepticism, and the sell-side’s mismodeling of growth and operating leverage, it seems like seven moons are aligning here.
The Short-Term Story
Over the last 18 months, Carvana had to shrink in order to get lean and efficient. That meant doing multiple rounds of layoffs to bring headcount in line with a lower but stable level of retail unit volume. It meant cutting both inventory and advertising by over 50%. When Carvana cuts advertising, naturally site and app traffic declines. And when it cuts inventory, it becomes even less likely that the traffic that does visit finds the car they are looking for. Not surprisingly, retail units declined 24% last year. That self-inflicted harm to sales was necessary so management could fix the cost structure.
The opposite will be true when the business returns to growth. It will hire more logistics and delivery personnel, which will enable faster delivery times, including same-day delivery. Faster delivery times increase sales conversion. It will also increase inventory, which means customers are more likely to find the car they want. That increases sales conversion, which means customer acquisition costs fall, which means the business can increase advertising to bring in more top of the funnel traffic. More traffic that is converting higher will enable continued growth of inventory, which increases sales conversion, drops CAC, enables more advertising, and the positive cycle runs higher still. In contrast, this will be a self-inflicted benefit to sales.
So one big question has been “When is Step 3 going to start?” The company has been in Step 2 since last April. Management’s most recent public comments indicate they are still in Step 2, but very recent evidence suggests Step 3 may be starting right now.
There have been several posts on LinkedIn by Carvana HR advertising jobs and hiring events. Here are some of them:
Consistent with that, job listings on carvana.com have increased materially. Tracking them each day shows that they increase and then fall as they get filled and that cycle repeats.
The alt data from Alternative Alpha also shows a few things. First, average advertised delivery times have come down in recent weeks. In November and December, the average advertised delivery times were 6.3 days. In the last 10 days, it has averaged 5.3 days and has touched high 4s some days. This is consistent with the improving delivery speeds, including the continuous rollout of same-day delivery in more markets as you can see from these press releases.
Feb 07, 2024 Carvana Brings Same Day Vehicle Delivery to Oklahoma City
Jan 31, 2024 Carvana Accelerates Car Buying with Same-Day Delivery in San Antonio
Jan 24, 2024 Carvana Raises the Bar in Online Auto Retail with Same Day Delivery Expansion to Birmingham
Dec 11, 2023 Carvana Launches Same Day Vehicle Delivery Offering to Columbus and Cincinnati Area Residents
Nov 29, 2023 Carvana Expands Same Day Delivery Offering to Central Florida Residents
Nov 16, 2023 Carvana Expands Same Day Delivery Offering to Dallas-Fort Worth Area Customers
Nov 13, 2023 Carvana Debuts Same Day Vehicle Delivery In Greater Atlanta Area
My view is that hiring more delivery drivers and improving delivery speeds may have been the first part of Step 3. That increases sales conversion without needing to increase inventory or advertising very much. It is a lower CAC way to grow. That also seems consistent with management’s guidance on days to sale, which was 66 in the third quarter and is expected to come down further to the “low to mid 60s” range. Sales increasing faster than inventory would accomplish that.
In addition, and not to bury the lede, but Alt Alpha data also shows sales and orders starting to pick up in recent weeks. While one might wonder if this is just a normal seasonal ramp, this hiring will be unwound after tax season, and Step 2 will continue, I think that is very unlikely. All signs point to the end of Step 2 being near. And for this to only be a normal seasonal ramp that unwinds again, it would probably have to mean Step 3 isn’t planned to begin until sometime in the second half of the year, which seems unrealistic based on management’s commentary and the available evidence.
On inventory, Monday’s data from Yipit showed a large increase in inventory, including cars bought from now defunct competitor Vroom. That is why CVNA was up about 10% on Monday.
On top of all this, founder and CEO Ernie Garcia did an interview with CNBC last week that resulted in the headline: A year after bankruptcy concerns, Carvana is leaner and ready for its Wall Street redemption. It’s worth a read. To sit down for a wide-ranging interview just two weeks before earnings that results in a headline like that seems bullish to me. It is hard to imagine not having more good things to report on February 22nd after setting the table like that.
So How Profitable Is Growth Going To Be?
This is the other big question. Thanks to its Step 2 efforts, Carvana is already the most profitable it has ever been on a unit economics basis. Sustainable and ongoing non-GAAP GPU is already well over $5,000, the highest level ever, thanks to lower inbound and transport costs, buying an even higher percentage of cars from customers, additional revenue streams like adding a long-distance delivery fee, and other changes. And on the fulfillment side, variable opex has decreased as vehicle miles traveled has come down. They have paired picks ups with drop offs, improving efficiency. Over 40% of customers are now opting to pick up their cars at Carvana vending machines, which saves last mile delivery costs. So as a result of these and other efficiency projects, go forward volume growth is going to come with higher-than-ever incremental profits.
I think non-GAAP GPU should be in the neighborhood of $5,600 this year, down from recent quarters as the higher than usual amount of loans sales normalize. It may end up higher to the extent they retain the residual certificates of their loan securitizations longer (see Appendix II of their last shareholder letter). Then we have to think about the variable costs of fulfilling the next unit and incremental advertising cost per unit. On the November call, management disclosed some new data, including operations expenses per unit. This is essentially the more variable component of SG&A.
Here is a multiyear chart showing operations expenses per unit.
It’s been coming down nicely over the last year. But also consider that it was around the $2,000 level from 2018 through 2021 when Carvana was growing retail units at a 70% annual rate and investing meaningfully ahead of that demand. There were a ton of growth expenses built into that $2,000 per unit. So I think this will continue to decline. Ernie actually thinks operations expenses in dollars will decline even with volume growth, which means they would plummet on a per-unit basis.
And if you read the definition of operations expenses I posted above, it includes semi-fixed expenses resulting from the under-utilization of logistics capacity. As you can see in the graphic below, logistics infrastructure is only 18%-24% utilized right now.
So I think operations expenses per unit can fall substantially with more progress on operations costs and as the utilization of that logistics infrastructure improves with volume. Once we are at a steady state on operations expenses, I think they might be $1,000 or lower on an incremental basis. I think incremental advertising per unit might be $800 or lower, which includes an assumption of ramping ad spending. If anything that could be too high because higher inventory and selection and lower delivery times increase sales conversion. So all together, we might have $5,600 of non-GAAP GPU minus $1,000 of incremental operations expense per unit minus $800 of incremental advertising per unit, which would drive $3,800 of incremental adjusted EBITDA per unit. Even on a GAAP basis, that should be over $3,300. They won’t report these figures because they are incremental figures, but we’ll be able to do the math. Even with modest growth in overhead expenses, I think the incremental profit on the next retail unit sold should be something in this neighborhood.
What are the implications? Well, if Carvana were to grow retail units by 100,000 to ~415,000 units, it might grow adjusted EBITDA by $350 million to $724 million. If it grows by 200,000 over some period of time, it would add on $700 million so adjusted EBITDA would be almost $1.1 billion. And so on.
That is why the sell-side consensus expectations are so interesting.
I’ve written several times (here, here, here, here, and here) over the last year how sell-side expectations have been way behind the curve on Carvana. That seems likely to continue. For the fourth-quarter, I have $98 million versus the $58 million shown here.
For 2024, sell-side consensus is only $373 million. That is up only $38.5 million versus last year. That reflects a consensus expectation of 345,000 retail units sold this year, which is up only marginally from last year’s ~315,000 level. Remember, 315,000 reflects the intentional reduction in unit volume from slashing inventory and advertising both by over 50% and other numerous other efficiency initiatives that sacrificed unit volume. To think they enter Step 3 and only improves retail volume by 9.5% is really, really conservative.
This is also implying that the 30,000 incremental retail units drive only $38.5 million of incremental adjusted EBITDA. That’s an incremental adjusted EBITDA per unit of only $1,285. That’s even lower than the actual adjusted EBITDA per unit I have them reporting in this fourth quarter, not even on an incremental basis. Simply put, the sell-side is badly mismodeling both retail unit growth and the incremental profitability per unit, which will be driven by the high operating leverage made possible by their efficiency initiatives and ramping into their severely underutilized existing capacity. Importantly, there is a lot of room or error with the incremental unit economics above without changing the conclusion.
Of course, management will have lots of optionality. They can reduce prices and lower GPU to grow faster if they want to, which would drive up volume at the expense of some profit per unit. That’s what Ernie says when he uses the term “competitively advantaged unit economics.” Simply put, they can afford to do things like that others can’t.
I am currently estimating around $700 million of adjusted EBITDA in 2024 and just over $1.1 billion in 2025. That’s based on 401,711 retail units this year and 522,224 retail units in 2025.
Carvana reports fourth-quarter results on Thursday, February 22nd after the close. About 29% of the Class A share are sold short.
Disclosure: Long CVNA
Disclaimer: Disclaimer: This post is for entertainment purposes only and is not a recommendation to buy or sell any security. Everything I write could be completely wrong and the stock I’m writing about could go to $0. Rely entirely on your own research and investment judgement.