Carvana Co. (CVNA) Q3 2017 Earnings Report, Transcript and Summary

Good afternoon, and welcome to the Carvana Third Quarter 2017 Earnings Call. [Operator Instructions] Please note this event is being recorded. I would now like to turn the conference over to Alex Wellins with The Blueshirt Group. Please go ahead.

Thank you, Brian. Good afternoon, ladies and gentlemen, and thank you for joining us on Carvana's third quarter 2017 earnings conference call. Please note that this call will be simultaneously webcast on the Investor Relations section of the company's corporate website at investors.carvana.com. The third quarter shareholder letter is also posted on the IR website. Joining me on the call today are Ernie Garcia, Chief Executive Officer; and Mark Jenkins, Chief Financial Officer. Before we start, I would like to remind you that the following discussion contains forward-looking statements within the meaning of the federal securities laws, including, but not limited to, Carvana's market opportunities and future financial results that involve a number of risks and uncertainties that may cause actual results to differ materially from those discussed here. A detailed discussion of the material factors that could cause actual results to differ from forward-looking statements can be found on the Risk Factors section of Carvana's prospectus related to its initial public offering filed pursuant to Rule 424(b) under the Securities Act with the SEC on April 28, 2017. The forward-looking statements and risks in this conference call are based on current expectations as of today, and Carvana assumes no obligation to update or revise them whether as a result of new developments or otherwise. Now with that said, I'd like to turn the call over to Ernie Garcia. Ernie?

Thank you, Alex, and thanks, everyone, for joining the call. As we have in the past, we put a shareholder letter out on the Investor Relations site. There's a fair amount of information in the letter, so it may be useful to have that with you as we go through the call. I'd like to start by acknowledging that despite our unit sales growth of 133% in the quarter, we ended up in the lower end of our unit guidance range. There were a number of drivers of this that we'll discuss -- that we discussed in the letter, but the summary is that August was disappointing, and by September, we were back on track. In fact, for the first time on our history, September was the strongest month in the quarter. Unfortunately, that did not give us enough time to make up the gap. That said, it has us feeling confident, believing that we will be able to make up the difference in the fourth quarter. Therefore, we are reaffirming our guidance for the year despite the soft August. The softness in unit sales in the quarter flowed through our income statement in the expected ways and resulted in us coming near the bottom end of our guidance range for EBITDA margin as well. Outside of the August unit softness and the related effects, the rest of the quarter was full of really good news. Operationally, Q3 was a very strong quarter. We opened our first Carvana-built inspection center outside of Phoenix that will add another 50,000 units of annual capacity at full utilization. We also expanded [indiscernible] out to Phoenix and L.A., connecting the network from coast to coast. Additionally, we opened 9 markets in the quarter, our fastest market opening pace ever. Our operations team is executing well and is currently focused on preparing the business for the significant growth that we expect heading into the first quarter. We continue to make significant strides in technology as well. Over 20% of our customers are now buying a car from start to finish on their phone. For millennials, that number is closer to 30%. We're making significant progress in total GPU. We finished Q3 with $1,742, up $241 sequentially. There are many factors driving these increases, with average days to sale being an important and positive contributor. We moved from 107 days in Q2 to 97 days in Q3. Notably, we exited the quarter with 85 average days to sale in September and now believe we will be around 75 days for the fourth quarter. We're also making significant strides in the financing side of the business. We are proud to announce that we entered into a new agreement with Ally to finance $1.4 billion of customer purchases on our platform, increasing the sum of their commitments to approximately $2 billion. In addition, we closed a $75 million master sale-leaseback agreement that enables us to monetize real estate sitting on our balance sheet. This is clear evidence of our ability to efficiently finance our balance sheet and to reduce cash consumed by CapEx. Lastly, I would like to make sure I'd point to the bigger picture here. There's a lot going on in the business and a lot of numbers to follow. Sometimes, it can get easy to lose sight of what is most important when there are so many moving parts. So I'd like to briefly summarize what we view as most important. We have a product that customers are expressing a clear demand for. In the third quarter, our average customer rating was 4.7 out of 5 stars, and our NPS was 85, which we were able to maintain even as we grew the business by triple digits. We continue to build the brand. We now have 7 vending machines and saw just over 200% growth year-over-year in unique monthly visitors. This increase in traffic is the fastest we have seen since the first quarter of 2015. We continue to grow very rapidly. The trends in market tiers over time remain the same. We grew unit sales by 133% in the quarter, and we expect that to accelerate to over 150% for the fourth quarter. Additionally, our 2017 cohort of markets is our fastest growing cohort yet. We are improving unit economics. We expect total GPU in 2017 to be about 50% higher than 2016 and are achieving that at the same time that we grow units at approximately 140% for total gross profit growth of 250% in 2017. We're doing all of these at nearly the $1 billion revenue level. We're executing on the national roll-out plan. We expect to open over 20 markets this year, nearly doubling, and it's our fifth year in business. We opened our first Carvana-built IC and expanded our logistics network to the West Coast in the quarter, unlocking more markets and additional reconditioning capacity. We are innovating. Our business model benefits from significant technology and infrastructure-driven barriers to entry, and we're working hard every day to continue to extend our lead. Our market is huge. We're growing into a $1 trillion industry that is unbelievably fragmented with tens of thousands players in the U.S., very few of whom are well positioned to respond. The fundamentals are increasingly strong, and we are excited and energized to continue innovating and growing in this unique opportunity. With that, I'll hand it off to Mark.

Thank you, Ernie, and thank you all for joining us today. Full details on our third quarter financial results are available in our shareholder letter. We put a lot of financial detail into the letter. So similar to last quarter, I will not repeat many of the numbers. Instead, I will highlight a few key aspects of our financial model, discuss a few key results and give you more color on our expectations for the remainder of 2017. Unless otherwise noted, all comparisons are on a year-over-year basis. We're excited to report another quarter of triple-digit growth in both retail unit sales and revenue combined with continued margin improvement. Retail units sold totaled 11,719 in Q3, an increase of 133%, and total revenue was $225.4 million, an increase of 128%. We expect retail unit sales growth to accelerate in Q4 to 143% to 168% and are reaffirming and tightening our full year retail unit sales outlook to 44,300 units to 45,700 units, reflecting a growth rate of 136% to 144%. Total gross profit in Q3 was $20.4 million, an increase of 202%. We improved total gross profit per unit sequentially by $241 to a total of $1,742 in the third quarter. We are executing our plan to drive GPU growth and saw increases in a number of areas in Q3, including a reduction in average days to sale and the introduction of our GAP waiver coverage product into our online checkout flow. We expect total GPU to be $1,500 to $1,650 in the fourth quarter, including an estimated combined negative impact of $180 to $220 from our annual Cyber Monday promotion as well as our hurricane relief promotions. This represents a normal seasonal decline in the fourth quarter which we expect to be followed by significant GPU growth in Q1. EBITDA margin was negative 15.9% in Q3, a slight improvement from the prior quarter. We see EBITDA margin improving sequentially in the fourth quarter at the midpoint of our guidance and anticipate further improvement in EBITDA margin in the first quarter of 2018. We believe this outlook demonstrates further progress toward our commitment to improving EBITDA margin and evidence of our clear path to profitability. As we move into the final quarter of 2017, we expect to continue to grow retail unit sales, open new markets and drive operating efficiencies. We expect OpEx to increase in dollar terms as we expand but to decrease as a percent of revenue in Q4. OpEx includes anticipated stock-based compensation of approximately $1.9 million in Q4. We expect very modest increases in share count in Q4. You should use approximately 137 million shares on a fully exchanged basis. Following the lock-up, pre-IPO investors exchanged LLC units for approximately 400,000 Class A common shares. These 400,000 shares will be included in GAAP basic share count in Q4. We ended Q3 with $103.5 million in cash and equivalents. Following quarter end, we entered into a master sale-leaseback agreement that provides for the future sale-leaseback of up to $75 million of existing and future vending machines and IRC construction projects. We currently have approximately $50 million in assets on our balance sheet that are eligible to be sold and leaseback-ed under this agreement and believe this demonstrates our ability to use our balance sheet to provide additional financial flexibility as we continue our rapid growth. We are also pleased to announce the upsize and renewal of our automotive finance receivables facility, increasing Ally's purchase commitment by an aggregate of $1.4 billion to a total of approximately $2 billion. As we round out the year and look forward to significant growth in 2018, this additional $1.4 billion in committed financing for loan originations will allow us to provide a seamless e-commerce experience to many more customers as we grow. Through 3 quarters of 2017, we have made significant progress toward achieving our financial goals. We are now at a time of year where we lay the groundwork for Q1 when we have historically experienced our largest sequential growth in retail unit sales and GPU. Thank you for your attention, and we're now ready for questions.

[Operator Instructions] Our first question comes from David Lim with Wells Fargo.

So the question that I have is the volume impact that you saw in Texas and Florida relative to your outlook and how much of that was recovered. If you could parse that by the 2 states, that would be helpful so we can get an idea of the -- some of the potential lost sales and gains that you've got in the 2 regions. And then I have a couple of follow-ups on the guidance, if you will.

Okay. So you want to hit the -- we'll hit the hurricanes first then. So I think -- let's start with Harvey. Harvey hit Houston in late August. I think there were clearly probably 10 or 15 days there where sales were significantly depressed. There was also clearly a strong recovery period in September. We've seen that abate a little bit from there, but we expect there to continue to be a recovery for the next month or so to some degree. Irma hit a couple of weeks later in Florida. That was clearly a negative not only for Florida but for the rest of the Southeast. We clearly saw sales affected in Atlanta and throughout the rest of Southeast. So that storm was clearly kind of a net negative. If you add them up, I think honestly, it's -- they're pretty close to probably 0 on the quarter. I think you could make an argument it's a little bit negative, but it's hard to really put that together. So I think Harvey, net positive in sales in the quarter; Irma, net negative. The sum of the 2 of them is pretty close to breakeven.

Got you. And then on the midpoint of your unit outlook, it appears to be obviously maintained but you lowered your revenue guide at the midpoint for the full year. Is this a function of average transaction prices dropping? But also, we're looking at the used vehicle values via Manheim, and they seem to be upticking. I was wondering if you could sort of box that out for us. And then even if the ATP is declining and you're maintaining your full year GPUs, what are the puts and takes in how you're maintaining that GPU? Is it just better back-end efficiency?

Yes, okay. So let's start with ASP declining. So I think, first of all, the way our inventory algorithm works is it looks at cars that are being sold on the site. It looks at demand across our site and across many other different sources, and it tries to figure out which cars are likely to sell. And then we go out and we buy the cars that are likely to sell. So generally, changes in inventory that we're selling are driven by changes in demand that we're seeing on the site. I think there's a number of subtle changes happening all the time. Probably the most pronounced of those changes in the quarter is we did see an uptick in the percentage of our customers that are millennials in the quarter. We probably are approaching -- about 50% of our customers now are millennials. Those customers tend to prefer cars that are around $2,500 cheaper than average, and so that's affecting ASP to some degree. I think that's probably the largest of several factors that's driving it. I don't think that the general kind of used car indices and their movement are a significant contributor there. I think it's much more driven by the cars customers are demanding. And then as it relates to kind of ASP impact from GPU, I think we've done a little bit of work on that. And in general, for every $100 of ASP movement, we tend to see about kind of $2 to $4 of GPU movement. So it's not a huge driver of GPU. I think the reason we call it GPU and end it with unit is that in automotive retail, the unit is largely what drives kind of the gross profit. And so I think in general, things look really good. We feel like we're clearly on our path to our $3,000 midterm GPU goal. We ran a little ahead of that path this quarter. I think we're starting to show some progress in several parts of transaction and, in general, feel like we've got good things in front of us there.

And my final question is referring to your performance in wholesale gross profit per unit. It looks like -- according to our number, as you guys broke even, I think you typically make about $100, $105 per unit on that business line. What happened in the quarter related to wholesale?

David, so we think we had a great -- just to clarify one aspect of that question. So what we've reported is a $418 profit per wholesale unit sold in Q3, and so that's actually our highest per wholesale unit sold measure ever. I think that we have a number of interesting things happening in the wholesale side of the business, including integrated Carlypso technology that we talked about a little bit on the previous call in order to better understand all of the options and features as well as the wholesale and retail market valuations of cars that customers might bring to the site. And so we're doing some interesting things there and had a good quarter on the wholesale front.

Next question comes from Mark Kelley with Citi.

The first one is, last quarter, you called OEM spending as the reason to widen the unit range that you were forecasting for Q3. In the shareholder letter, you talked about atypical seasonal patterns and some of the moving pieces from the hurricane that you just walked through a little bit more. But can you talk about what you saw on the OEM side of the business? And then second, can you just talk through the tax receivable agreements a bit? I think that would have been triggered with the 400,000 shares that were converted from LLC units. So any of the moving pieces there in terms of how that affects the financials would be helpful.

Sure. So I think -- first, let's start with kind of where were our expectations wrong in the quarter. I think they were clearly wrong in August. If we look at the quarter, we kind of knew July at the time that we guided. August was soft relative to expectations. September was strong relative to expectations but not quite as strong as August was soft. And so I think we were left trying to figure out what do we think happened in August. I think we're hesitant to too confidently attribute August softness to the OEM spend just because we haven't heard other groups talk about that. That said, given the data we look at, it does seem to be the clearest driver that we're able to find. So for perspective, in previous years, we've seen a discontinuous 20% or 30% increase in top-funnel metrics, whether it's traffic or accounts or any top-funnel metric that we look at that generally pops up somewhere in the middle of August. We tend to see that kind of pop up for a couple of weeks and then abate over the next several weeks. And we generally see a big burst in sales that shows up in August. When we go back and look at that in the past, we saw that according to Nielsen data in 2016, in 2 weeks in August, the OEMs roughly doubled their broadcast television spend, and that kind of corresponded precisely with when we saw our top-funnel traffic increased. When we look at that this year, we do not see the same kind of discontinuous, 2-week burst of OEM broadcast spend that occurred in August. In fact, in the same 2 weeks this year, the OEMs cumulatively spent about $320 million on broadcast marketing, whereas last year, in those 2 weeks, they spent about $640 million in broadcast marketing. That, I think, is the clearest explanation for why sales were soft in August. But again, we haven't heard of people talk about that. So we're hesitant to put too much of it on there. I think what we know is that September was strong and August was not as good as we would have liked, and we're feeling really good about heading into the fourth quarter. And then do you want to hit actually...

Yes. And then Mark, I can hit the TRA question. So basically, the way the TRA works, when LLC unitholders exchange units for common shares, a tax asset is created based on the step-up in basis between the current share price and the pre-existing tax basis on those units. That tax asset then, in the event Carvana has positive taxable income, can lead to tax savings for Carvana that gets shared between pre-existing LLC unitholders who exchange shares as well as Carvana public shareholders. I think the most important thing to realize there is that the TRA and the UP-C structure that gave rise to the TRA, we view them as big positives for shareholders down the road when and if Carvana generates taxable income. And there are cash tax savings that can be shared with public shareholders as well as pre-existing LLC unitholders.

All helpful. And then one more quick one if I could squeeze it in. It looks like -- maybe I missed it, but are you guys breaking out advertising spend any more? And if so, I guess, what was it in Q3?

Sure. So Mark, we are breaking out advertising spend in the MD&A section of our financial statements. In Q3, that number was $15.5 million. So a large part of that advertising spend was based on our accelerated market ramp. Obviously, we launched 7 new markets in Q2. We launched another 9 new markets in Q3, and we're advertising in all those markets. And that, obviously, is a key driver of that advertising spend number.

Next question comes from Colin Sebastian with Robert Baird.

I wonder -- maybe as a follow-up then, could you talk about the decisions to accelerate the new market expansion and the impact that this should have near term on both revenues and expenses? And then going back to the GPU, I was wondering if you could break out the improvement there that came from vehicle sales versus other revenues. And then I have one quick follow-up.

Great. So first, let's talk about the market openings. I think there are several things driving that. One, the new cohorts continue to perform very well. As I said, 2017 is now our fastest ramping cohort that we've ever seen. Secondarily, we continue to make a lot of progress on GPU. I think the way we think about kind of the economic cost of opening a new market is when we open a market, we have a little bit of CapEx to connect that up to the logistics network. The largest single cost that we incur is marketing. That varies across markets, but that's the biggest cost we incur. And then we start to basically make positive revenues that are driven by the number of cars that we sell times that GPU. As GPU increases, as the speed of sales in these new markets increase, that just makes kind of the cash investment and the payback time for these new markets decrease. And that has us wanting to open more markets faster. I think additionally and very importantly, the ops team has done a great job at continually getting better at opening these markets faster. They've made several kind of innovations in the way that we open new markets. And so I think we're also operationally capable of opening markets faster. So I think that's also driving the decision. Mark, do you want to...

Great. And then to hit the second part of that question, so the components of GPU improvements -- that's the $241 in sequential improvement, $76 of it came from the used vehicle; $39 of it came from wholesale sales of vehicles, as we discussed earlier; and $125 of it came from increases on other gross profit. There, I'd like to hit just a couple of drivers of those gains. So as we discussed, we increased vehicle gross profit in part due to a decline in average days to sale. That's a key part of our path to our $3,000 midterm gross profit per unit goal. And so we showed positive progress there. On other gross profit component, I think a really exciting thing that happened in the quarter was the roll-out of our GAP waiver coverage product in the online checkout flow to customers in most of our states. We've seen really great take-up of that product, better than we were expecting, and think that serves as a really nice example of our ability to launch new products onto our mobile checkout process or our desktop checkout process. And we have great evidence that customers find those products valuable, and they enhance our profitability.

Okay. And then maybe just a quick follow-up. With the expansion of the national ad spend, wondering what you're thinking there in terms of ROI and if that's changing the trend line at all in terms of the out-of-market vehicle purchases.

I think it's still a little early to comment, too, as to how it's directly impacting sales. I think we gave you the statistic that we've grown traffic by 200% year-over-year, which is the fastest that's it's been since Q1 2015. I think that, that is driven by this and maybe some other marketing innovations that we're putting forward. I think we've historically seen -- as we access new customers whether it's through new markets or new marketing channels, we generally see kind of the top funnel fill up first and we see conversion rates kind of slow. And then as customers get more comfortable with the brand, we see those conversion rates increase. And so I think the traffic is clearly showing up. I think we'll expect to see some of those benefits heading into next year on the national marketing front, but it's clearly working the top of the funnel. There's no doubt about that.

Next question comes from Sharon Zackfia with William Blair.

A couple of questions on the softness you saw in August and then the recovery in September. Was that pretty broad-based in what you saw, I mean, if you exclude the hurricane impact? And when you see that kind of softness in August, what kind of levers do you try to apply to kind of get the customer back on the web or back ordering cars in your model?

Yes. So first, as far as softness in August and strength in September, my fault for not clarifying. In general, we are kind of controlling as best we can for hurricane effects when we give you kind of those directional impact. Definitely, there was recovery from Harvey in September and then there was some negatives from Irma in September. We're trying to massage that out when we say, "August was soft and September was strong." So those, I think, were real trends. They were broad-based. That's not just kind of the hurricane effects playing out. As far as levers, I mean, there are many levers in the business. I think we're still coming up on a 5-year-old company. We want to make sure that we're doing the right thing for the company in the long run. And so I think we work hard to try not to overreact too much here or there. If our expectations are off a little bit in August, we don't want to try to pull levers too heavily -- too heavy-handedly to try to make up for that just to kind of make sure that we hit a quarter. We want to make sure what we do is right in the long term. So I think there are several levers in the business, but in general, we're focused on pulling the right levers with the -- a longer time horizon than kind of making sure that we're hitting any given month's or quarter's numbers.

And then a follow-up on the decision to accelerate growth, I mean, how, if at all, does that impact the path to free cash flow manifesting itself? And Mark, do you still think you have adequate financing capabilities to bridge that gap other than going back to the public markets?

Yes. So I'll start that one. So I think going back to the economics of an incremental market, there are clearly investment upfront in CapEx and then there are clearly negative cash consumers early on as we advertise, is our biggest expense, and we don't have a lot of sales to offset that. I think when we're thinking about how new markets impact our capital plan, what's important is for us to try to calculate what we think the cumulative cash invested in any given market is at the point in time when we expect to be troughing in cash. And I think that's a consideration that we're taking into account as we decide to open up additional markets. To briefly dive into the cash question, I think our expectations, broadly, have not changed. We continue to feel like we've got the ability to continue to see through our plan. I think the $75 million master sale-leaseback agreement is clear evidence of our ability to creatively and efficiently finance our balance sheet and our CapEx. And we think that going forward, we'll continue to assess the capital needs of the business, like we always have. We'll continue to assess the options that are available to us through all the different capital markets, and we'll do what's right for the business at any point in time.

Next question comes from Dan Salmon with BMO Capital Markets.

A couple of questions. First, in the shareholder letter, you highlighted once again days sales outstanding coming down and targeting it to again come down in the fourth quarter. Maybe if you could call out maybe 1 or 2 things that's particularly driving that lately, that would be great. And then just to return to the accelerated marketing -- market opening rate of hitting about 9 this quarter up from 7 in the second quarter, do you think 9 is the run rate? It sounds like you feel like you can get even better than that on a quarterly basis. And then maybe lastly, if you'll humor me on one market in particular, any early read on Los Angeles in particular? I know it's only been 2 months, but that would seem like a market that is particularly rich for your model.

Dan, this is Mark. I'll start with your question about average days to sale. So I think the reduction in average days to sale that we've seen and are projecting for Q4 is a direct result of our inventory strategy. So since Q4 2016, we've left website units approximately flat. And meanwhile, we've nearly doubled the number of markets that we operate in and have grown sales by 110% as well if you just compare Q3 to Q4 unit sales growth. That's the plan we laid out. It's basically playing out the way that we forecasted. When you have a business model that allows you to leave inventory flat and grow sales rapidly, that does decrease inventory turns. And so I think that's what we're seeing.

And I'll just jump in. I just want to make sure I kind of highlight that again. I do think that's a big deal and that's a key actually to the business model. So again, we've roughly doubled markets over the last 12 months. Inventory has been flat. We've grown sales by 130%. Inventory has been flat. If you look at any of our metrics prior to the fourth quarter of 2016, they all benefited from growing inventory. So whether it was market share that benefited from growing inventory because we're putting marketing dollars out there and we're getting customers come to our site. And when they show up, they have an increasing likelihood that they're going to find the car that they want. That benefited market share. If you are looking at the way the tax trended prior to the fourth quarter of 2016, it benefited from growing inventory because of those same effects. If we have higher conversion rates on customers that were attracted to the website, we're going to see lower tax. The last 12 months has really been, in a way, an investment in moving down turn time. And that investment has taken the form of creating at least a relative headwind in sales and in tax for us, but we've been able to kind of fight through that going from 97 days in the third quarter to an expected 75 days in the fourth quarter. We're clearly making a lot of progress there. As we stated in the past, our goals were to get in line with traditional dealers which historically have had a range from 30 to 60 days. When we get into that range and we feel like it's time to turn inventory back up, we start to grow inventory at the same pace that we grow sales. So if we double sales, we're going to double inventory and all those tailwinds come back. So I think that's an exciting piece of evidence that's starting to show up and something that we're really kind of proud of. As it relates to kind of Los Angeles, I think Los Angeles in the 2017 cohort, as we said, in general, those markets are trending faster. We're very excited about the West Coast, obviously huge population centers in the West Coast, I mean, debatably great customer bases for us in the West Coast. We also know that historically, as we've gone into markets that are further away from markets we've already opened, those markets tend to ramp a little slower as we build brand. So I think we want to be careful about getting too forward leaning on that, but in general, we feel really good. Another point I would make there is we're telling you that these markets are ramping faster. That's the trend that's been in place for a long time. It's certainly a trend that's holding in 2017. We don't really need that trend. As long as markets ramp consistently, we're going to build a really, really big business here. And so we're excited about those trends continuing to hold. For market openings in the future, we're going to stick with kind of guidance for the fourth quarter and for this year for now. So we don't want to look too far into the future, but I do think that we've clearly got into a place we believe that we can comfortably open market at a similar pace to what we opened this year or an accelerated pace, and we'll make that assessment over the next several months then give you more visibility on our next call.

Next question comes from Ron Josey with JMP Securities.

So just as a follow-up for that, maybe Ernie and Mark, talking about DSOs. And just can you remind me the seasonality associated with days sales outstanding particularly as you -- I think Mark said you've laid the groundwork for 1Q. And I think typically, you prepare for 1Q with greater inventory in 4Q. So can you just talk about that in terms of seasonality? And then on the pricing trends, I know we talked about that with pricing going lower with more millennials going out there. But I'm curious if you think this is more of a permanent pricing headwind, meaning we should expect pricing to come down. And if so, has this changed your inventory mix in terms of what you're looking to buy. And then last one is real quick. We've gotten some questions with all the hurricanes and floods. Like how does that impact your purchasing of used cars, like ensuring they're not flooded cars?

Thank you. Lots of good questions there. So first, on seasonality and days to sale, I would decompose that into 2 pieces. So one side is what's happening on the supply side, how many cars are we buying. That's kind of half of the equation on days to sale. In general, we've been basically replacing cars that we've sold over the last several quarters and we're holding inventory flat. In the fourth quarter, generally speaking, depreciation rates accelerate and car prices drop. And so it's a good time of year to increase sales -- or excuse me, to increase purchases. We plan to do that to some degree in this fourth quarter. So I think there will be a little bit of an increase in inventory here. That doesn't -- immediately speaking, that's actually helpful to days to sale because you have more fresh inventory. Over time, you've got to make sure that you're careful to not get too heavy with one cohort of inventory, but that's something that we'll be very cognizant of and careful about. On the demand side, basically, just selling more cars is good for days to sale. And we're heading into Q1, which is every year, for our -- since our inception, it's been the biggest quarter-over-quarter growth rate. Q2 is generally the second largest quarter-over-quarter growth rate. So we're heading into a lot of expected demand here over the next 6 months or so. So we expect that to also be helpful, and that's kind of how days to sale works. On the ASP front, I don't know -- what I do know is we're going to continue to watch what's happening there. We're going to give customers the cars that they want. The trend with millennials is something that started to show up in kind of April, May. We weren't really sure if it was real. It's continued since then, and so we've continued to buy the cars that they're demanding. I don't think it's particularly important. I think what is important is that we make sure that we're the best place for anybody to buy a car regardless of what type of car they want to buy, regardless of what generation they're from. And as long as we do that, then we're just going to kind of follow the demand. I think our best guess is probably that demand will look similar to how it looks today, but I think it's likely to migrate and we'll update you as that changes. As it relates to purchasing cars from hurricane-related areas, we've been very careful about that. I think we're a little bit lucky there because one of the benefits of our national logistics network is that we're able to buy cars anywhere in the country and then ship them to whichever inspection center we'd like to recondition them at and photograph them at and hold them at prior to sale. So we've been able to avoid areas where those cars are popping up with higher frequency, but that's definitely something that we're being careful about.

The next question comes from Steve Dyer with Craig-Hallum Capital.

Ryan Sigdahl on for Steve. Most of our questions have been answered but one here for you. Doesn't look like you guys are running lower than your 7,000-ish vehicles in your website for parts of the quarter. And was that just due to buy-sell timing and refurbishing? Or were there other issue there with procurement or other things?

Yes, that was largely a transitory timing issue. So I wouldn't take too much from that particular couple hundred unit variation in website units.

Next question comes from Sameet Sinha with B. Riley FBR.

A couple of questions. So the other revenue GPU saw some nice increases. Can you talk about some of the new products you have planned for the next few quarters? It seemed like the GAP waiver worked really well. And my second question is as it relates to guidance, your adjusted EBITDA loss guide went up about $13 million and you're adding about 5 new markets. Should we assume that all of that relates to that -- those new market opens -- other incremental new market opens?

So that question was hard to hear on our end. So I'm going to do my best to answer what I think you asked. You asked about the full year EBITDA guidance, is that correct? Or Q4 EBITDA guidance?

Yes. I was basically saying the full year EBITDA loss guide increased by about $13 million. So does that -- all of that correlate to the 5 incremental new markets you're opening or planning to open this quarter?

So our change in EBITDA guidance involves 2 things. One is market openings, and that includes both more markets as well as launching markets earlier than initially anticipated. The other driver of our change in EBITDA margin guidance for the year is the change in ASP that we've seen over the last several months and what we also expect going forward. So those are the 2 big drivers.

And then I think you also asked about other ancillary GPU. I think the biggest new contributor in the quarter was -- as we said on the last call, we kind of rolled out and started to test GAP at the very end of Q2. In Q3, we continued to roll that out into more markets. We're very pleased with the results there. We're already kind of getting penetration levels that are in line with the industry benchmarks. So I think that's a success story and I think bigger than just a success story there. I think that's also suggestive of, one, our ability to continue to roll out these value-add products for our customers; and then, two, the reality that our customers are willing to buy them at similar rates to customers buying cars elsewhere. And so I think that, that bodes well for the remainder of our GPU walk.

The next question comes from Seth Basham with Wedbush Securities.

My question is around the guidance for the fourth quarter as it relates to the unit sales and maybe a guidance for the full year as it relates to unit sales. Considering the fact that you're opening 5 additional markets and your guidance for the full year hasn't really changed in terms of unit sales, should we attribute the relative decline therefore to the slowdown you saw in August? Or are there other factors?

Yes. I think I would say 2 things there. I think the softness in August is one, right. So I think that that's playing a role. Two, I would say these markets don't contribute meaningful sales early in their life, not at the volumes that we're kind of talking about. They start to contribute more meaningfully generally when they kind of head through their first Q1, Q2 is when we start to see bigger contribution. So I think it's those 2 things. As we said earlier, 2017 cohort is ramping faster than any cohort ever. And the other kind of trends that we've seen historically in market shares remain intact. So there's nothing concerning on that front.

Got it. So the 2017 cohort of markets, are they ramping faster than you expected them to?

They are. I mean, it's very early. Look, I want to be really careful about getting too excited about that because the average 2017 market is 2 months old or so. I mean, we have to -- I don't see the math on that, but they're not very old. But they are certainly, so far, ramping very well.

Got it, okay. My second question is around the other GPU line, that $125 million -- $125 per unit increase. How much of it was from actual financing as opposed to ancillary products?

So $53 of it was -- and I'm going to round here so this is going to add to $126. But $53 was financing and then $73 million was related to other products.

Got it. And that $53 is driven by improved underwriting? Or what are the drivers behind the improvement?

Yes. So pricing and scoring optimizations.

Next question comes from James Albertine with Consumer Edge.

Can you just -- I apologize if I missed it, but for housekeeping item here, what percentage of your customers finance the purchase? And of that percentage, how many did you help arrange financing for?

So approximately 70% of Carvana customers use our in-house Carvana financing product. Of the remaining 30% of customers, many buy cash and many use some third-party financing from a local credit union or bank.

Understood. And if I may, as it relates to sort of reserve accounting here, you've got a variety of different -- I think, certainly, the 7-day opportunity to return the vehicle. You've extended warranties and so on and so forth. To what degree have your reserve accounts sort of shifted from the second quarter? And is there anything worth calling out in terms of the trend lines there?

Nothing meaningful to call out there. We're using the same methodology and have seen very similar trends.

[Operator Instructions] The next question comes from Gary Prestopino with Barrington Research.

Most of the questions have been answered but, Ernie, did you give an inventory unit number for the end of the quarter? I didn't see it in any of the narratives. So...

We did not -- we did report the number of units that we had available for customers on the website. That was 6,689, but we did not report a total inventory unit number.

Okay. And then in terms of the warranty contracts and the GAP insurance, especially with the warranty contracts, I think, Mark, we talked about -- you said about 25% of the buyers took a warranty contract. Did that change? Did you see a movement up in the quarter? And then the other question I would have is that as you're starting to sell this GAP insurance, are you starting to see people, buyers take both GAP and warranty coverage in a big way?

Yes. So we haven't seen any kind of meaningful changes in the VSC attachment rate. We have seen, since we introduced GAP waiver protection into the checkout flow, buyers that elect to both have VSC and GAP, which I think is a good sign overall. I do think as we add more and more products to the checkout flow, having good data that shows customers are willing to attach multiple products when they're going through the checkout process, I think, is a very good thing.

Do you -- I mean, it's very early in the game here, but I mean, as you're looking at this, do you find that as you're growing your base of millennials or buyers, are they taking these kind of warranty contracts as well as GAP insurance as much as other age groups would or not?

So we don't plan to break out the specific attachment rates and metrics at the customer demographic level in general. We thought the millennials stat was interesting but don't plan to do that on an ongoing basis.

This concludes our question-and-answer session. I'd like to turn the conference back over to management for any closing remarks.

All right. Well, first of all, I want to thank everyone on team Carvana out there in telephone land. Thank you to each of you. It takes a lot of talent and passion and even more work to move as fast as we're moving, and I feel proud to come in every day and be part of this team. Thank you guys very much. Next, I'd like to thank everyone on the call for taking the time to join. We appreciate it and look forward to seeing you out at the conferences. See you soon.

The conference has now concluded. Thank you for attending today's presentation. You may now disconnect.