Image source: The Motley Fool.
Carvana Co. (NYSE:CVNA)Q4 2018 Earnings Conference CallFeb. 27, 2019, 5:30 p.m. ET
Contents: Prepared Remarks Questions and Answers Call Participants Prepared Remarks:
Operator
Good afternoon and welcome to the Carvana Fourth Quarter 2018 Earnings Conference Call. All participants will be in listen-only mode. (Operator Instructions) After today's presentation, there will be an opportunity to ask questions. (Operator Instructions) Please note, this event is being recorded. I would now like to turn the conference over to Mike Levin, Vice President of Investor Relations. Please go ahead.
Michael Levin -- Vice President of Investor Relations
Thank you, Andrea, and good afternoon, ladies and gentlemen. Thank you for joining us on Carvana's fourth quarter 2018 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 fourth quarter shareholder letter is also posted on the 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 risks and uncertainties that may cause actual results to differ materially from those discussed here.
A detailed discussion of the material factors that cause actual results to differ from forward-looking statements can be found in the risk section of Carvana's most recent Form 10-K. 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. We will be discussing some non-GAAP financial measures, including gross profit, ex-Gift SG&A, ex-Gift EBITDA. All gross profit SG&A and EBITDA metrics mentioned by us on this call are on an ex-Gift basis, unless otherwise noted.
You can find reconciliations of those measures to the most directly comparable GAAP measure in the shareholder letter we published today posted on our website at investors.carvana.com. And now with that said, I'd like to turn over the call to Ernie Garcia, Ernie?
Ernest Garcia -- President and Chief Executive Officer
Thanks, Mike, and thanks everyone for joining the call. The completion of another year is always fine because it provides an opportunity to take a quick break from all the things we're doing to continue our progression forward and to instead pause and reflect on what we've built so far. We're excited to be reporting results on another incredible year of building the Company. We closed the year with our 20th straight quarter of triple-digit growth. There is an accomplishment that clearly demonstrates the power of our customer offering and the quality of our execution, and is the feat that are extremely proud of. We grew units by nearly 50,000 for the year. This growth makes us the fastest growing automotive retailer in the country. We grew revenue by 128%. This makes us the fastest organically growing US-listed public company at anywhere near our scale in either retail, consumer or technology.
We grew our GPU from 1,539 to 2,133, an increase of nearly $600 for the year and an increase of over 1,100 since going public in April of 2017, cementing our path to our mid-term goal of $3,000 and beyond. We continue to march toward profitability, showing 700 basis points of leverage for 2018 and over 1,300 basis points since going public. All of this progress is the result of two powerful long term forces. We're building a product our customers love and we have incredible people who love building and delivering it. If we do our jobs here, the same two forces will drive us to our goal of selling 2 million plus units per year.
That number may sound daunting to some, but looking backwards provides some helpful context. Getting to 2 million requires that we grow roughly 20x from here. Few companies scale like that. That said, we are 120th size we are today just 3.5 short years ago. The next 20x won't undoubtedly be harder than the last, but that is a pretty interesting statistic that speaks to the power of our customer offering as of the power of the growth it has generated. In the shareholder letter we published this afternoon, we include our market penetration and customer acquisition charts by cohort. These charts continue to get us very excited and show continued impressive penetration into this one trillion dollar market.
Our growth in customer acquisition cost leverage in each of our cohorts continues to progress very quickly and clearly likes the path to our long-term unit financial model goals. Led by Atlanta, our first market, which achieved a nearly 2% market penetration in the fourth quarter and a $462 CAC for the year. The trends that underlie at the cohorts progress have persisted in a consistent encouraging way as well. Newer markets are on average growing faster than older markets. In fact, 72 of 84 non-Atlanta markets are ramping faster than Atlanta at the same time in its life and Phoenix, our home market, made it to the 1% market penetration milestone in 16 months versus the 42 months it took Atlanta.
Smaller markets continue to ramp faster than larger markets. An example of this the Montgomery, Alabama and Charleston, South Carolina became our fastest markets to 0.5% of market penetration, both in that landmark in only their second quarters after launch. We continue to see markets that are nearby pre-existing markets ramp faster than those that aren't, which we think is exciting indication of the value of the brand we are building. As of today, we are in 100 markets that span different geographies, demographics and market sizes. The results we are seeing are consistent and compelling, cementing our belief that we have built a better way to buy a car that has mass appeal.
Part of the materials we published nearly included a deeper dive in our finance platform, which we hope you will find helpful. There are many takeaways from that information, but the most important in general I would like to make today is this. Fundamental improvements in our model, namely lower variable cost, incredible customer experiences and thoughtful vertical integration are powerful and have many expressions. One such expressions in our finance platform which enjoys many advantages when compared to traditional finance businesses. In the case of traditional automotive finance companies, they're part of a vertical chain that sits on top of brick and mortar dealers. Effectively brick and mortar dealers are suppliers to automotive finance companies and all the attributes of the brick and mortar dealers, including their cost and pricing structure, their vehicles and the customer experiences they generate are embedded in the loans originated by their finance company partners.
In our case, the innovation that exists in our retail platform flows into our finance platform, allowing us to generate higher quality loans and to create customers finance experiences that are as simple as the retail experiences we're known for. There's a lot more detail on our platform and on the performance of the loans we have generated over time in the 101 materials we posted today which we request you to review. Our strategy has always been to build an industry-changing business through a combination of great technology, world-class operations and a culture of delivering great customer experiences.
I'll now touch on the progress made in each of these areas in 2018. First, technology; there are two areas inside of technology I'd like to briefly hit. Customer-facing and internal-facing improvements. We made a ton of progress on customer-facing initiatives. This included redesigning each of our main pages, improving content, usability and speed, developing and launching our first mobile app that has all core functionality available on carvana.com and laying the foundation for what we hope will be an exciting SEO progress over time. We also continue to work integrating, integrating on and improving the technology we acquired from Carlypso, Car360 and Propel, each of which have exciting releases coming out in the first half of this year and broader product roadmaps in front of them stretching as far as the mind's eye can currently see.
We also made a lot of less visible progress in our internal-facing initiatives that are aimed at further improving efficiency and scalability through proprietary software. This progress including building applications and supporting and little bit improvements to enable our buyers to buy cars our customers will have love more intelligently, enhancements to our logistics software and schedule that enables us to support more activity types, including buying cars from our customers at a more efficiently and intelligently scheduled deliveries, and applications used by our advocates that puts vehicle and transaction knowledge into a scalable tool that allows our advocates to spend more time doing what they do best, delivering exceptional customer experiences, a dynamic underwriting tool that enables us to complete underwriting tasks more quickly and with less effort and many other internal tools that we expect to enable us to deliver better customer experiences over time even more efficiently.
Now heading to ops. As we've said before, given the customer response we're seeing as well as the size of our market, we believe execution is the biggest single risk that stand between us and our goals. Accordingly, a value in our execution to date gives an important and positive indication of our team's ability to execute. In 2018, we opened more markets than we ever had before, and more vending machines and we did our entire cumulative operating history prior. Our purchasing inspection center, logistics and advocate teams have all done an amazing job keeping up with 20 straight quarters of triple-digit growth, while simultaneously investing in better, more efficient and more scalable processes. Achieving that growth and improvement in any company is difficult and extremely rare, but doing so in the company operationally as complex as ours that has reinvented the entire automotive retail supply chain is even more impressive.
Our team has done an outstanding job and they deserve a tremendous amount of credit for our success today. Finally, I want to touch on building a culture of delivering great customer experiences. We believe a great company culture has two fundamental ingredients, a better business model and great people. These two ingredients are very closely intertwined and we believe we are positive outlier in both. Great people want to work with other great people and they want to work on problems they find interesting and meaningful. These foundations are what enable our culture, all the progress we are reporting and all the potential we see in front of us.
I don't think I'll be able to adequately convey the comfort, confidence and pride I get whenever I walk into a meeting or into one of our locations in the field and see the quality of people, the focus and dedication and the belief in what we are building together. But I believe it is unique and it is something that I personally very grateful for and excited about. We are positioned to achieve our mission of change we feel by cars. We are positioned to change the industry. We have a lot of work in front of us and there will inevitably be bumps along the way, but we are more confident than we have ever been and where we're headed, and firmly believe we are in control of our destiny. If our team executes, we will achieve our goals.
Mark?
Mark Jenkins -- Chief Financial Officer
Thank you, Ernie, and thank you all for joining us today. Unless otherwise noted, all comparisons are on a year-over-year basis. We are pleased to report another quarter of triple-digit growth in both retail units and revenue. Retail units totaled 27,750 in Q4, an increase of 105%. Total revenue was $584.8 million in Q4, an increase of 121%. In 2018 we also completed our second year of triple-digit growth as a public company. Retail units sold totaled 94,108 in 2018, an increase of 113% and revenue totaled $1.96 billion, an increase of 128%. Underlying this Company growth was another year of strong growth within our cohorts. Atlanta grew 30% in 2018, reaching a quarterly market penetration in Q4 of 1.94%, up from 1.54% the previous year.
2014 cohort growth accelerated in Q4, reaching a market penetration of 1.11%. The 2015 cohort faced a harder comparison this year than last year, with replacement demand in Houston in 2017, but it remained on a strong trajectory and the 2016 cohort side second year of nearly triple-digit growth. Finally, the 2018 cohort launched with the fastest start in our history, exceeding the 2017 cohort, which ramped faster than any previous ones. We believe the strong and consistent growth in our cohorts demonstrates the significant consumer demand for our product and the replicateability and scalability of our model.
Total gross profit per unit in Q4 was $2,131. For the year, total gross profit per unit was $2,133, an increase of $594. Our growth in GPU in 2018 was broad based, including gains in average days to sale, financing and ancillary products. EBITDA margin was negative 10.8% in Q4, an improvement of 4.7%, reflecting significant gains in the quarter. We made substantial progress on operating leverage this year while also increasing our investment in scalability and expanding our logistics and delivery network.
Since becoming a public company nearly two years ago, we've made significant progress on our three main financial objectives. Since 2016, we have grown units and revenue more than five times while increasing gross profit per unit by more than $1,100 and reducing SG&A per unit by more than $1,350. This represents more than $2,450 of progress toward our goal of becoming the largest and most profitable auto retailer. We ended the year with $410 million in total liquidity resources, which includes cash and equivalents, available capacity in our floorplan line and available capacity under existing sale leaseback agreements.
We plan to report on liquidity resources going forward to provide investors with a holistic picture of the resources that underlie our financial strategy. As of December 31, we had $253.6 million of borrowing capacity on our floorplan line. Over time, we plan to borrow on the line to increase financial flexibility, while also paying it down selectively to optimize interest expense. This means we will maintain cash and equivalents that may be lower than in past periods, while simultaneously maintaining higher availability on the line than in past periods. Combining the two along with our real estate liquidity resources gives a more complete view of our financial picture.
On December 6, 2018, we converted the outstanding balance of convertible preferred stock that we had issued in December 2017 in the Class A common shares. You should use approximately 150 million shares on a fully exchange basis in the first half of 2019. This year, we have made the decision to move to annual guidance that we will update quarterly. This decision was primarily made because our management team evaluates and runs the business with a long-term perspective, focusing on providing the best customer experience while driving toward the goals outlined in our long-term financial model.
The goal of our guidance is to give the market the information necessary to evaluate our business, while more closely aligning our shareholders' perspective with our own. As we look toward 2019, we expect significant growth in retail units and revenue, increased GPU and improved EBITDA margin. Our outlook for retail units sold is 160,000 to 165,000, up from 94,108 in 2018. Our outlook for total revenue is $3.4 billion to $3.5 billion, an increase from $1.96 billion in 2018. We expect GPU to increase to $2,450 to $2,650, reflecting gains across all parts of the transaction as we move to our $3,000 mid-term goal and beyond.
We expect OpEx to increase in dollar terms as we expand, but to decrease as a percent of revenue. Combined with our GPU growth, we expect EBITDA margin to improve to negative 5.5% to negative 3.5%, representing meaningful progress on the path to profitability. In addition, we expect to open 50 to 60 markets in 2019, reflecting an acceleration of market openings relative to the 41 we opened in 2018. In our second year of as a public company, we made significant progress toward achieving our financial goals. As we look toward 2019, we intend to continue to expand our footprint, rapidly grow sales, increase GPU and demonstrate operating leverage, all while continuing to improve our customer experience.
Thank you for your attention. We'll now take questions.
Questions and Answers:
Operator
We will now begin the question-and-answer session. (Operator Instructions) Our first question will come from Zach Fadem of Wells Fargo. Please go ahead.
Zach Fadem -- Wells Fargo -- Analyst
Hey, good morning guys, sorry about that. When thinking about the drivers of the GPU improvement in 2019, could you walk us through some of the near-term buckets of improvement that you're expecting to contribute and how much do you expect the improvement to come from financing and other items?
Ernest Garcia -- President and Chief Executive Officer
Sure. So we've talked a lot about the different sources of GPU opportunity for us and I'll start with some of the big buckets that we talked about at Analyst Day and then I'll touch on some other items. So two of the big opportunities that we see are in buying more cars and customers and in expanding our finance monetization program. I think the buying more cars from customers is something where we obviously started to see more traction in 2018 as we devoted more resources to that business line. It's an area where we feel like we provide customers with a great experience and have some really, really nice aspects of our offering and we'll be looking to grow that business over time.
Of course, buying cars from customers is a contributor to GPU for two reasons, first, cars that you buy from customers in wholesale, you will earn wholesale gross profit that way and then cars that you source and customers are typically more profitable when you turn them into a retail transaction than cars that you source from auctions or other sources. So that's one place where we see opportunity. Obviously that's a new business and we expect to be doing a lots of testing and we're still we feel at a very early stage, but that's opportunity number one. Opportunity number two that we see is expanding our finance program. I think we presented a little bit of data today on the quality of our program to-date and I think that's something that we're very excited about.
We have a special model, our loans have performed very well over time and as a result are very valuable and we'll look to continue to monetize those loans over time by expanding our number of partners and bringing down our cost of funds. In terms of other areas, I do think we see opportunities across the other parts of the transaction as well. We've talked over time about continuing to leverage our existing infrastructure that we use to transport and recondition cars. We've talked over time about continuing to expand attachment on existing ancillary products as well as potentially over time adding new ancillary products to the checkout flow. There are many places where we see opportunity, but I've walked you through the couple of the big ones.
Mark Jenkins -- Chief Financial Officer
Yes, if I could just jump into Zach, I think most of the attention, most recently on the finance platform has been on how we're going to monetize that platform. I hope the data we presented today is helpful. We showed that our last experience has been great there relative to other public issuers that have data out there and that has been consistently improving over time, but I think it's a story lost, really important. We want to make sure we make a point on as well, which is as long as we're originating high quality loans, we're selling cars at low price to customers, we're putting them in a better position to succeed in their loans. They are also defaulting less which aside from creating more valuable loans we can monetize better is a better customer experience with customers that are less likely to default, who are more likely to become repeat customers. And so we think that's a really important area where our vertical integration is paying off in a way that is probably not yet totally understood. But we think it pays serious dividends over time.
Zach Fadem -- Wells Fargo -- Analyst
Got it, that's super helpful. And just taken this bigger picture, could you comment on just overall consumer adoption in the category. Any color on how online penetration has been trending in the used car industry in 2018? And how do you expect this adoption to continue in '19 and over the next 3 to 5 years?
Ernest Garcia -- President and Chief Executive Officer
So I think that's a hard question to answer precisely specifically because I think most consumers in the US still don't even know exactly what is meant, when you say you buy online. So I think if you even if you ask them the question, I'm not sure they have enough exposure to what our model looks like to know if that something is appealing to them. We continue to see very steady progress across all of our cohorts. Atlanta continues to grow very quickly. I think something that gets us really excited is that we don't feel like we necessarily even need significant additional gains in the number of customers who would be willing to buy cars online in order to continue to see a lot of gains in actual transactions in the market of Atlanta, for example.
And the reason that can happen is because as we grow inventory or as we add more inspection centers and shrink delivery time, we should get a higher conversion rate on the customers that we're already seeing today. And then, I think there's kind of this completely separate dimension of customers becoming aware of what we do and getting more comfortable with what we do. We haven't made public data on research that we do, trying to understand which customers are or what percentage of customers are really been aware of us. But the awareness levels nationwide and in Atlanta remain pretty low. And so we think there's a lot of opportunity there. And so, we think our job is to keep delivering great customer experiences one at a time and those customer experiences will tell the story and get more and more customers comfortable. And if we execute all the positive feedback and the model is going to show up, we should be in a really good spot.
Zach Fadem -- Wells Fargo -- Analyst
So if I were to ask that question slightly differently, I mean CarMax just launched an omnichannel platform in Atlanta. Just curious if you're seeing any impact there? And could you see that as a positive for driving brand -- consumer awareness in the category?
Ernest Garcia -- President and Chief Executive Officer
Yes,I think we would start out just kind of reiterating our general view on competition which is that this is an enormous market, it's enormously fragmented. The largest player has 1.8% market share, the largest 100 players have about 7% market share. So I think those dynamics are fairly unique across different retail verticals. It's also a business that is somewhat operationally intensive, making rapid scaling difficult. So we think that all those things, I mean the competition is relatively less important than it might be in other verticals. I think your point's a good one though. I think when you start competition, it's a little bit unclear which direction that it pushes because effectively two brands are being pushed when one of our competitors out there advertising selling cars online. They're pushing their own brand out there, right, which all as constant is negative for us. They're also pushing up the brand of buying cars in a new way.
And we would like to think that we're kind of cemented in consumers' minds as the number one brand to go and buy a car online. And so we think there's a chance that ultimately will be positive. I think we haven't seen any material impacts in any of our markets where any of the different competitors out there doing any different things. In Atlanta in particular, we're very pleased with the growth we saw in the fourth quarter. It's extremely early in 2019. But we continue to see really good trends there. We've seen a slight acceleration of growth in the first couple of months in Atlanta. So far we just don't see anything that concerns us too much on that front and we continue to believe that we're the leader and as long we keep building, we're in a great spot.
Operator
Our next question comes from Sharon Zackfia of William Blair. Please go ahead.
Sharon Zackfia -- William Blair -- Analyst
Hi, good afternoon. A couple of questions, I guess it was really encouraging to hear about Phoenix and the ramp you saw there. I'm wondering if correspondingly you size similar decreasing in customer acquisition costs as you reach that 1%?
Ernest Garcia -- President and Chief Executive Officer
So we don't break out kind of our reporting at the individual market level. But I think in general, it's a very good first order assumptions that there is a inextricable link between customer acquisition cost and market share. So in general, that would be a good assumption. But we're going to continue to report at the cohort level.
Sharon Zackfia -- William Blair -- Analyst
Can you talk about anything you did specifically in Phoenix? Or anything unique there, other than you being headquartered there that you think drove the massive quick increase?
Ernest Garcia -- President and Chief Executive Officer
So I do think we had an advantage by virtue of being headquartered here, which I think made the early awareness easier to gain. So I think it probably benefited us. I also think the other advantage of Phoenix is it's very nearby an inspection center. And so there's on the order of 25% of our cars in the fourth quarter at least were parked just down the street from those customers in Phoenix and that meant their delivery times were fast and I think that probably hit in conversion as well.
Sharon Zackfia -- William Blair -- Analyst
Okay, great. And then on wholesale, I mean that was probably the only area where maybe it was a little bit weaker than we would have expected. Can you give some commentary around wholesale just given that that was part of the longer-term thought-process on GPU?
Ernest Garcia -- President and Chief Executive Officer
Sure, yes. I'm going to break that into units and then margins. Let me start with units. I think we are very pleased with the progress we made across the year. I think we grew by something in the order of 220% to 230% in total units that we bought from customers. Some of those flowed through wholesale, some of those flowed through retail sales and kind of in either case, there's a profit opportunity there. But it just depends kind of which line item is going to flow through. I think like I said, the process we made last year was exceptional and I think our expectation is that we'll make continued very significant progress in 2019.
I think when you look at the fourth quarter and the first quarter, there's some dynamics that are worth kind of keeping in mind there. We have some different customer selection effect in the fourth quarter due to the promotion. We have some different customer selection effect in the first quarter due to the tax season. The fourth quarter and the first quarter tend to be the quarters where we're growing inventory the most. And in general, kind of the easiest inventory to go out and grow is coming from non-direct consumer channels. So we're going to -- we tend to grow more in those channels, which means it puts downward pressure on the percentage of cars that we're selling that were acquired from customers. But the underlying trend in cars that we're buying from customers and the margin we expect we will make over time are all very, very positive.
And then on the margin front, in wholesale in particular, I think that probably moves around by 100 bucks or maybe 200 bucks something like that, not a super material number. We're still -- as we try to set expectations, we're growing this business very, very fast and we're learning very, very fast. We're changing the ways that we're going out and acquiring new customers, which you have selection effects and changes customers go through the funnel. We're constantly evolving the funnel and learning and changing conversion there. We're changing the way that we nurture these customers, we're working on pricing at the aggregate level and at individual car levels. So there are constantly lots and lots of tests going on. And I think that that business today is optimized for growth in learning, not for steady progress. And so I think if we look at it over longer time frames, I think we would be very pleased with the progress there. If we look at it quarter-to-quarter with all these different dynamics, I think it could be a little bit choppier and so I would build that in your expectations.
Sharon Zackfia -- William Blair -- Analyst
And one last question, I didn't see this in the shareholder letter, what was the 2018 national market share for Carvana?
Ernest Garcia -- President and Chief Executive Officer
The way we calculate that, the 94,000 units divided by $40 million, so that would put you just shy about 0.25%.
Operator
Our next question comes from Ron Josey of JMP Securities. Please go ahead.
Ron Josey -- JMP Securities -- Analyst
Great, thanks for taking the question. Maybe a quick follow on and then a broader one on just the IRC strategy. On the follow-up, Ernie, you're just talking about wholesale and buying cars from consumers or from users. Can you tell us what percentage of those cars are sold to retail? I think you said 16% in the past I think in 3Q. And then maybe a broader question, just the IRC strategy. You talked about adding a production line in Phoenix and in Indianapolis launching newer IRCs then with only two production lines, now with four. So if you can just tell us about maybe how you're thinking about IRCs and the build-out as you reach to 2 million units over time. And then here in the near term, just location of these IRCs being potentially closer to your markets, the importance of being closer to the consumer with the IRCs and maybe the impact on shipping times as you launch call it, four production lines and two production line IRCs? Thank you. And then of course, CapEx. Thanks.
Ernest Garcia -- President and Chief Executive Officer
So retail -- so cars that we're selling to customers or were purchased from customers, that went to 17% in the quarter versus 16%. That's the statistic, the kind of thesis the headwind of growing inventory, right. If we're growing inventory, it's generally coming from non-direct and consumer channels. That means that a smaller percentage of your inventories on the site any point in time was purchase directly from consumers and that creates a headwind for the likelihood that any given customer chooses a car that was bought from a customer. So I think we're pleased with that progress, especially in light of the headwind.
From inspection center perspective, I think we're really focused on being sure that we stay in front of all the growth that we see over the next several years that we basically try to calculate by extrapolating out the co-workers. And then we're trying to make sure that we're building flexibility. So I think we're excited to be building in this efficiency. It enables us to add additional lines in Phoenix and in Indy. We're excited to be adding additional locations in different parts of the country to kind of shorten delivery times overall. We expect this to be a constant push. We will be continually either acquiring new inspection centers or adding to existing inspection centers, in many cases it's probably both. I think something that we're really excited about, Phoenix is the first inspection center that we built from scratch and kind of laid out and set up specifically for Carvana processes and in just about every metric that's already our most efficient inspection center. Many of the learnings that we kind of got there, we're taking into the next several inspection centers that we're setting up and that's driving some of the subtle changes in CapEx. We certainly now estimate $10 million to $12 million per line and we think that thinking about CapEx on a per line basis is probably the right way to think about it and then we'll size individual inspection centers based on the needs that we see in any given geography at any point in time.
Ron Josey -- JMP Securities -- Analyst
Great, thank you.
Operator
Our next question comes from Chris Bottiglieri of Wolfe Research. Please go ahead.
Chris Bottiglieri -- Wolfe Research -- Analyst
Hey guys, thanks for taking my question. A couple of financing questions. So the financing gain on retail unit really pop this quarter even if you back out the refinancing transaction. So it was already a noise on timing of receivables or maybe you can talk about what drove that significant improvement year-over-year or even relative to the last quarter?
Ernest Garcia -- President and Chief Executive Officer
So one factor that influenced Q4 gain on loan sale, looking at on a per retail unit sold basis is our Q4 refinancing transaction. And so the Q4 refinancing transaction had one major difference from the Q3 refinancing transaction and that is in Q4 we sold some loans directly from our balance sheet into the transaction along with loans that were refinanced. And so, because those loans were sold directly with much better financing than our typical forward flow financing, we earned a higher premium on those loans. And so part of the increase that you're seeing quarter-over-quarter and also year-over-year is due to that direct to sale to the purchasers with much better financing.
Chris Bottiglieri -- Wolfe Research -- Analyst
Got you. Okay. And then sort of giving through the financing primer while -- well on the calls are certainly fully yet, but I guess my first question is, it's obviously very impressive improvement in CNLs. So my first question is, has this improvement in the CNL been pretty consistent across your three scoring buckets, are you seeing like one bucket versus the other driving more significant outperformance. And then the follow-up to that would be, for the 10% of loans that are in the riskiest bucket, are you still receiving a positive fee on these gain on sale type transactions or are you looking at that payer in that 10%? Thank you.
Ernest Garcia -- President and Chief Executive Officer
So first, what I'd say is, I think you can kind of look at the outperformance of any given band of loans. Based on the data we provided you can look at kind of how the credit score or the lost curve look over time and then you can compare that to other issuers in the market. We've tried to provide that data in there. So I think that'll help you to try to size the relative outperformance by credit band. I think if you're asking about the improvements over time which credit band those came from, in general, they have come from several credit bands, but the kind of higher loss credit bands are always that have the biggest impact on your overall losses just because the way the math works out. So in general improvements there are always kind of most powerful to your entire curve.
And then, we haven't really broke out the premiums that we get on any given credit band. But I think in general we have very high quality loans originating and you can kind of look at the lot differences by credit band and then you can do a little math to figure out how much more valuable those loans may be. And I think across credit bands we're originating very high quality loans and then we're also obviously functionally doing something different. We're kind of like the combination of a retailer and a finance company in the sense that we're doing the credit scoring, pricing, structuring, all of that work that's associated with verifications work, all work associated with originating those loans and that also puts in a good spot to monetize. So, in general we're monetizing well across the spectrum, but we haven't broken out exactly how we're doing it in each bucket.
Chris Bottiglieri -- Wolfe Research -- Analyst
Okay, great. then just one quick one, very quickly on -- wanted to talk about the transportation bucket. If you are growing so exponentially market growth and I guess you opened another RC, is that like having a quantifiable impact on your inbound transportation cost, it may be temporarily depressed in the retail GPU, just wasn't sure if there was anything kind of drove some of the Q4 step down, anything you could quantify there would be helpful. Thank you.
Ernest Garcia -- President and Chief Executive Officer
Sure. So, let me first hit the, I guess, sequential change in retail GPU because I think that'll help the thinking on that. So, there's a couple things that happened in the fourth quarter that in fact retail GPU, the first is our Cyber Monday promotion, which is $1,000 offer, customers to take advantage of the promotion that obviously impacts retail GPU. Secondly, Q4 is the period in the year where used car depreciation rates are highest. And so while you're holding inventory between acquisition and sale, those cars depreciate faster and that also has an impact on retail GPU. In terms of inbound logistics costs, there's nothing that I would call out there in particular. I think it's mainly the two big factors that I listed upfront.
Mark Jenkins -- Chief Financial Officer
Yeah, I'd add to that. I think the effect you're talking about is probably present to some degree, but probably it isn't a huge number inbound. I think that effect is meaningful outbound as we build out the network and kind of spread out across the country with fewer IRCs, supporting an entire nationwide network, that definitely does provide pressure because we're moving cars longer distances on average. And then as we fill in the IRC network, you're starting to ship cars lower distances on average. So I think the bigger effect that over time will be on outbound, which shows up in SG&A, but there probably will be something of an effect inbound as well, to your point.
Chris Bottiglieri -- Wolfe Research -- Analyst
Thanks guys, appreciate it.
Operator
Our next question comes from Nat Schindler of Bank of America Merrill Lynch. Please go ahead.
Nat Schindler -- Bank of America Merrill Lynch -- Analyst
Yeah, hi guys, a couple of things. One, if I look at just some basic math on the marketing per unit in Atlanta, it looks like you're running, call it, 2.5% right now with a target on your advertising spend being somewhere on the order of 1%, 1.5%. What gets you there from here to get way beyond where you are, even in Atlanta when you're hitting roughly 2% market share and what do you need to do from here to get there? Secondly, just a weird thing, I just was wondering if you could help me understand why do you seem to have particularly in Atlanta. I can't really tell the other charts, why do you have a downturn in share in Q3 sequentially in the last several years? Is there something that occur seasonally that makes offline guys do better.
And finally, just one final question, and nothing is related, so you are going to have to remember them all. But there was a competitor made a commentary in the last couple of days that you guys are doing great but selling 100,000 units is one thing, when you start selling 400,000 units or something to that effect, similar to what you're suggesting your IRCs will be able to handle after you do this next land that you added in, it doesn't scale as well. Do you need to buy retail units, consumer units, do you need to buy from consumers in order to be able to get enough cars to meet -- to handle 400,000 units in the year?
Ernest Garcia -- President and Chief Executive Officer
Okay. There's a lot in there. So let's start with customer acquisition costs. So I think there's many, many ways that we think we can drive down customer acquisition cost. We talked a little bit earlier about kind of being in the early innings of awareness generation and I think that's true. We think that when we're advertising today, we're not only acquiring customers in real-time, we're also building a brand. We think there's a lot of evidence of that we can see in the cohort factor, right, they are downward sloping for reasons that we're building brand over time. And so we think there's some natural momentum that will probably continue to show up there.
I think the most compelling and kind of simplest way to understand how that -- we should be able to push that out over time, is just simply to think about conversion. So, if we take that market share of approximately 2% in Atlanta and we extrapolate that out nationwide, that suggests around 800,000 car sales. If we had on the order of 800,000 car sales and we had about a 30-day onsite turn time, we have about 70,000 cars on the site at any point in time, that's a very significant increase from where we are today and we think that alone with the exact same marketing dollars would lead to higher conversion and therefore push down our customer acquisition cost quite a bit.
If you had 800,000 cars per year you'd also have on the order of 16 inspection centers and so we'd see on average across the network faster delivery times, that would also push that down. So those to me are the simplest way to extrapolate out to kind of much lower customer acquisition costs, and so we feel really, really good about our path there. We're not going to rush it. We don't think that that's a particularly important target right now, right. We're already at a level where the variable economics make a ton of sense. And so if we can keep building brand and growing share, that's still a pretty interesting prospect. But we think it's pretty clear, the evidence is there that will be able to push down over time.
The downturn in Q3, that's a little bit of a quirk. And so I think the way that we kind of define our market penetration is we look at historical data, we use our historical seasonality to try to figure out what we think -- we use historical data basically to understand the total units sold in any given market. And then we've historically use our seasonality to spread those units across months or quarters. The indices, we're reporting in quarters, and that's because many of the public databases that we can look at for registration etcetera are either lagged or a little bit difficult to rely on.
And so we basically had embedded in our market share shape our historical seasonality at the time that we locked in the IPO. We talked last year about, in August, of the lack of an August bump. We saw in previous years a really big increase in sales in August and that was kind of locked into our expectations in our curve going forward. And so we think it's probably somewhat likely that that seasonality isn't right now that we have kind of more data and so we may continue to see Q3 in Atlanta looks like it's going down and popping back up.
I think if you look at any the other cohorts, there's many, many markets in those cohorts. And so all of those different seasonality assumptions are basically blended together and it's a much, much simpler curve to look at. But that's what you're seeing in Q3. And I think that's more what's driving than any interesting nuance with us versus brick and mortar dealers in the third quarter.
And then do we need to buy cars, more cars from customers to get the bigger scale. I think the answer is no, there's many, many cars go through auction. I think it's something on the order of 14 million cars and like that they do auction every year. So there is a lot of volume that goes through auction. Now, do we want to buy more cars from customers, the answer to that is absolutely. I think the way that we try to think about this entire market is basically that used car transactions are as consumers switching cars with each other and that switch happens through this extremely complex mechanism of a trading at a dealer and then transport through an auction and then an auction fee and then transports to another dealer and then sales to the customer.
And if you look at all that, there's a lot of cost in that chain. And I think the way that we try to think about the whole model is, how do we remove as much cost from the chain as we possibly can so we keep getting customers, very simple experiences, very fair prices. And then as a result, having that roll into the loans that they ultimately used to finance the purchase with us. So they've got great performance and are less likely to default and more likely to become repeat customers. So I think we definitely are going to continue to invest in buying more cars from our customers, but we would not agree that is accurate, that that is necessarily a necessity.
Nat Schindler -- Bank of America Merrill Lynch -- Analyst
Okay. If you look at the auction of that 14 million cars, roughly what percentage are in your kind of sweet spot of the three to five-year-old vehicle with, you know, the pretty prime used car?
Ernest Garcia -- President and Chief Executive Officer
I don't know the exact answer for you there, I believe and so I want to caveat, this was -- I believe about 50% of car sales in the US are less than five year old cars and so my guess is that cars going through auction would we somewhat proportionate to that distribution, but I'm not certain of that.
Operator
Our next question comes from Steve Dyer of Craig-Hallum. Please go ahead.
Steve Dyer -- Craig-Hallum -- Analyst
Thanks for taking my question. As it relates to your GPU guidance for this year, was there any assumption around another receivable sale in there or would that all be sort of incremental and additional in the guidance?
Mark Jenkins -- Chief Financial Officer
Yes, sure. So I can hit that. I think when we first announced the first refinancing transaction, we provided some commentary that we expected to do an additional fixed pool financing in Q4 and that we would endeavor to make it part of our ongoing finance monetization program. And so I think that still holds true today. Obviously, I think with the loan performance that we've seen and the quality of our platform, we think we'll have an ability to continue to use those types of transactions. We think those types of transactions make a lot of sense for everyone involved and we'll look to continue to do them going forward.
Steve Dyer -- Craig-Hallum -- Analyst
From that $400 or so dollar per unit GPU improvement this year, I guess what is your general thinking as to what role these transactions may play in that?
Mark Jenkins -- Chief Financial Officer
Sure. So consistent with past years, I think the way that we're looking at GPU and particularly our guidance on GPU is in terms of total GPU. We don't plan to break down specific line items. But we did to -- we've done two of these refinancing transaction so far, those give you a sense of the additional economics available when you transition from forward flow financing, which is less efficient to fixed pool financing which is more efficient. And like I said, we will look to do continued fixed pool financings over time to realize better effective cost of funds, which have a positive impact on finance gross profit.
Steve Dyer -- Craig-Hallum -- Analyst
Okay, so presumably there is something in it for this year. Just a question on inventory, I noticed your inventory was up 22% sequentially, floorplan debt -- would own 44% sequentially. Is that sort of as you hit a little bit you're using your floorplan line almost more like a revolver now, I mean taking equity in your vehicles or help me think about that?
Ernest Garcia -- President and Chief Executive Officer
Sure. So let me hit the inventory growth first and then talk the floorplan lines. So we typically grow inventory in Q4 in preparation for the first half of the next year. This year was no different. We started building inventory in advance of the positive demand seasonality in the first half of 2019. As it relates to the floorplan line, we expect looking forward to maintain more availability on the floorplan line than we have in the past. We think that's marked from a financial management strategy, the floorplan line is available to us to draw upon using our inventory as assets when we would like to, but from an interest expense management perspective, it makes sense to keep that line available as much as possible to minimize interest expense.
Steve Dyer -- Craig-Hallum -- Analyst
Okay. And then just one more from me. As it relates to CAC or more specifically advertising, how do you allocate those expenses by cohort when a lot of them are national? I guess I was always of the impression that it was really difficult to give any kind of profitability per cohort metric just because so many of the costs are national or corporate level? Thanks.
Ernest Garcia -- President and Chief Executive Officer
Sure. Yes, so on the order, half of our advertising is national and on the order half is local. We've talked about that before. That's because TV is our main national channel. And then in terms of allocating, so all the local channels are obviously very straightforward. For national cable, the way we allocate is based on population within a cohort. And so we take our national TV spend allocated out of the cohorts based on their population.
Operator
In the interest of time, we ask that you please limit yourself to one question and one follow-up. Our next question will come from Armintas Sinkevicius of Morgan Stanley. Please go ahead.
Armintas Sinkevicius -- Morgan Stanley -- Analyst
Great, thank you for taking the question. When I look at the free cash flow burn this quarter and the anticipation for further free cash flow burn in the quarters ahead, how do you think about financing? I know you have $400 million plus of availability, but at some point presumably you would need to raise either equity or debt and what's your thought process there?
Ernest Garcia -- President and Chief Executive Officer
Sure. So, we don't plan to raise equity or debt other than our standard asset based financing which we've accessed many times over the past few years. I think a couple of key points there, one, our inventory is financeable with our floorplan line and we haven't fully drawn of that floorplan line at this time as I was mentioning, with respect to the earlier question, but obviously have availability there.
The second key point I would make about sort of use of cash is a very large portion of our CapEx is hard assets that we've historically financed. Those include real estate assets as well as hauler assets, which we finance with some sale leaseback financing or other types of leases or equipment financing in the case of haulers. And so we will endeavor do that on a go-forward basis. At the end of Q4 we had approximately $130 million of unpledged real estate sitting on our balance sheet. We had a $75 million master sale leaseback agreement in place as well as some other smaller sale leaseback commitments. And so I think the combination of floorplan line when you're looking at working capital for inventory and then our historical access to capital for the hard assets on our balance sheet, I think we feel good about our liquidity position.
Armintas Sinkevicius -- Morgan Stanley -- Analyst
Okay, thank you for taking the question.
Ernest Garcia -- President and Chief Executive Officer
Thank you.
Operator
Our next question will come from Rick Nelson of Stephens. Please go ahead.
Rick Nelson -- Stephens -- Analyst
Thanks for taking my question as well. So, there is mentioned in the shareholder letter about a delay in tax refunds. To the extent in January '18, you comment on the first quarter and how confident are you that these delays every time as opposed -- some of the data suggesting refunds are coming in lower on a per person basis, so how that might impact the business.
Ernest Garcia -- President and Chief Executive Officer
Sure. So I think there's a lot going probably this tax season. I think there's probably more uncertainty about all the specifics today than there normally would be at this time of the year. So let me try to walk through that. I think based on the best research that we have seen, it seems like probably aggregate refund dollars overall will be lower. And our understanding is that that's going to be primarily driven by essentially 1099 filers and some higher income people who have kind of the ability to change their withholdings more easily. And as a result of the tax law change, the recent tax law change, that they increased their withholdings and therefore they will get lower refunds. As part of the tax law changes well though, there's an increase in something called the child tax credit.
Historically, if you look at kind of tax dollars that are released and then the kind of relationship that those tax laws have to you ultimate purchases the vehicles, the tax credits that kind of correlate best are the child tax credit and the earned income tax credit. Our belief today and based on research that we've seen from a number of banks is that those will be similar and potentially even higher. So we kind of think that the overall refunds will probably be lower with low certainty. We think that probably the refunds that are most highly correlated with the taxing response will be higher. We also feel that with relatively low certainty.
And then it is undoubtedly been delayed to some degree, the government puts date out there so we can kind of like watch when the drops are happening and we can see what year-over-year total dollars look like. So it's undoubtedly been delayed a bit. There was actually somewhat sizable lease drop today. So we're still kind of learning exactly how much it's going to be delayed and I don't think we know, but our guess would be with seasonally high confidence it's delayed, with low confidence it's probably going to be similar to maybe even a little bit better overall, but I think there's more uncertainty than there usually is in tax refund season this year.
Rick Nelson -- Stephens -- Analyst
And do you think effects the subprime kind of buyer more so than the prime buyer?
Ernest Garcia -- President and Chief Executive Officer
I think historically the tax refunds that have correlated most with kind of first quarter car purchases are the earned income tax credit and the child tax credit, which on average are associated with lower incomes. So I think it does probably impact the lower income population more than higher income population in terms of the immediate response to the tax refund.
Operator
Our next question comes from Nick Jones of Citi. Please go ahead.
Nick Jones -- Citi -- Analyst
Hi, thanks for taking my question. I want to focus back on IRCs, can you dive a little deeper on kind of the shift work, are there 2 shifts a day, 5 days a week or 7 days a week. And is it possible to add a third shift. So I guess I'm asking is there kind of like at 33% upside to what your footprint is going to be even at 400,000 unit capacity?
Ernest Garcia -- President and Chief Executive Officer
Sure. The way we're planning today is based on 2 shifts per line and those those shifts running 5 days per week. We do believe over time that there is an opportunity to expand beyond that and that could include weekend production, as you mentioned, so transitioning from 5 days per week to 6 or even 7. And then it also can include expanding shifts within any given day to 2.5 or 3. Those aren't in our immediate plans, but we do think given the infrastructure that we're building, that option is on the table at some point in the future and which I think is great, but we don't have any immediate plans for that today. We're focused on 2 shifts per line, 5 days per week.
Nick Jones -- Citi -- Analyst
Okay. One quick follow-up, how did the Black Friday, Cyber Monday weekend go this year?
Ernest Garcia -- President and Chief Executive Officer
I would say it went fine. It went a little bit better than 2017 and it didn't go as well as '15 and '16. I think when we guided for Q4, that guidance was basically -- the largest probably driver of the range was uncertainty around exactly how the promotion would play out. And I think it went a little bit better than '17 but not nearly as well as '15 and '16. Our kind of guidance range was basically for it to go somewhere between '15 and '16 or '17. So I think it went fine. I don't think any of our major thoughts about running that promotion have changed. We think about that promotion as much more of a brand building effort than kind of a real-time sales driver and the reason we like it is because it associates us with online buying, and so we think that's valuable. And so I think we will continue to evaluate that at long length and through kind of the shorter length how many sales it drove this year, it was fine but it wasn't great.
Operator
Our next question comes from Colin Sebastian of Robert W Baird. Please go ahead.
Ben -- Robert W. Baird -- Analyst
Hi, it's Ben on for Colin. Just two questions, first, given the advertising leverage you're seeing across your cohorts, does that imply that your older cohorts are profitable on a market contribution basis or have some the logistics investments can offset those. And then secondly, as it relates to the 2019 guide, how are you guys prioritizing unit growth versus GPU improvement? Is there a bias toward one versus the other?
Ernest Garcia -- President and Chief Executive Officer
Sure. I would say, I mean in general, I think our priorities are to drive unit growth and deliver customer experiences and then GPU and then to reduce SG&A. So I think in general, that's how we think about it. I think it's probably a fair way to kind of think about the business that our expenses to complete a transaction are somewhat similar across markets. So when you look at our Atlanta CAC for example and compare that nationwide, it's probably fair to then say OK, if the rest of country gets to where Atlanta is, how much that improves the economics and the answer there is very, very significant.
And then the question about profitability is always the question about how you're allocating expenses because obviously at the Company level, we are not profitable, but with many reasonable allocations our older markets are have been profitable and that has us feeling very confident about where we're going.
Ben -- Robert W. Baird -- Analyst
Thanks.
Operator
Our next question comes from Seth Basham of Wedbush Securities. Please go ahead.
Seth Basham -- Wedbush Securities -- Analyst
Thanks a lot and good evening. My question first is on the quarter, trying to understand a little bit better the bridge to your results relative to your expectations. It seems like Black Friday wasn't quite as good as you expected. But those are relatively low incremental margin units, highly discounted. What else drove the miss in EBITDA relative to your expectations. Can you please give a little more color around investments etc?
Ernest Garcia -- President and Chief Executive Officer
Sure. Yes. So the primary driver of our EBITDA margin relative to expectations was the lower units and also the lower revenue that was associated with those lower units. In addition, we made some incremental investments similar to the ones we've talked about on previous calls, and that includes investing additionally in our technology team here at home office as well as our logistics network.
Seth Basham -- Wedbush Securities -- Analyst
Got it. And when you think about that rate of investments going forward, do you expect a much slower growth rate in 2019 relative to '18?
Ernest Garcia -- President and Chief Executive Officer
Sure, so all of all of our expectations about future investments are incorporated into our 2019 guidance. I think overall I think we're very excited about the leverage that we've shown. I think Ernie alluded to, we made 700 basis points of EBITDA margin improvement year-over-year, a significant portion of that was driven by expense leverage year-over-year in 2018, which actually accelerated relative to 2017. I think as we look forward to 2019, we expect obviously further GPU gains but also further operating leverage gains as we march toward profitability.
Operator
Our next question comes from Sameet Sinha of B Riley, FBR. Please go ahead.
Sameet Sinha -- B Riley, FBR -- Analyst
Yes, thank you very much. So Ernie, you've previously spoken about as you look at network expansion part of it is sometimes increasing density, part of it is expansion. It seems like this year it's going to be about increasing your density and that will be your capital efficiency is going to be higher. Now if you start to think about next year, are you kind of -- would you say that next year you'll probably look to do more of the expansion part? And we just need just to kind of have our expectations adjusted accordingly. That's one question.
Secondly, the question is, if I'm just looking on a year-over-year basis on your retail business, units increased by about 14,000 year-over-year, gross profit by about $15,000. So incremental unit gross profit was about $1,000 let's say, maybe a little more. But that's down year-over-year. That's down like 20%. Is there anything specific in the composition or is it maybe to Cyber Monday, anything that you can point to?
Ernest Garcia -- President and Chief Executive Officer
Okay. So, let's start with the first question. So I think you're right, most of the markets that we're going to open or many of the markets that we are still going to open this year are going to be added basically on top of the existing logistics network which will be a little bit more capital efficient. Now I think that the CapEx required to open markets in general is pretty low, so I don't know that that's a super material difference when we kind of zooming out and looking at the Company. But I do think directionally we will have a lower CapEx per market that we're opening this year.
I think looking forward, our logistics network is getting pretty expansive now and it is covering most of the country. There's still some room for us to go up to the Pacific Northwest and a little bit more room for us to go up farther North in the middle of the country. But we don't have a lot to add there. So I do think that after these 60 markets probably in 2020 we will be opening more markets that are kind of are a little bit heavier in CapEx. But I don't know if that will be even a necessarily noticeable change but I think directionally that
probably is going to happen.
I think the most important kind of density that we can build into the system over time is basically just having all of these markets continue to ramp up their market penetration and then that drives up sales. Driving up sales allows us to carry more inventory, that increases conversion and then driving up sales also allows us to
