Carvana Co (CVNA) 2018 Q3 法說會逐字稿

Good afternoon, and welcome to the Carvana Third Quarter 2018 Earnings Conference Call.

(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.

Thank you, Austin.

Good afternoon, ladies and gentlemen, and thank you for joining us on Carvana's Third Quarter 2018 Earnings Conference Call.

Please note this call will be simultaneously webcast on the Investor Relations section of the company's corporate website.

The third quarter shareholder letter is also posted on the IR website.

I'd like to remind everyone that we will be webcasting our first Analyst Day later this month on November 29, where we plan to discuss our long-term outlook and technology foundation.

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 Factors section of Carvana's most recent Form 10-K and Form 10-Q.

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.

And now with that said, I'd like turn the call over to Ernie Garcia.

Ernie?

Thank you, Mike, and thanks, everyone, for joining the call.

The third quarter was great for us when evaluated through a number of important lenses.

We performed well against our guidance in all respects.

We exceeded well and relieve many of the operational pinch points of the business.

We made significant progress in our preparation for the volume we expect, heading into the seasonally stronger first half of 2019.

We continue to invest in technology that we believe will further differentiate the quality experience we're able to provide to our customers.

We made significant progress in our business to buying cars from our customers.

Most importantly, the sum of everything we saw in the quarter continues to increase our confidence in the size and economics of the long-term opportunities stand in front of us.

We're building a fundamentally better solution to Carvana for our customers.

These fundamentals are playing out in many specific ways and revealing many opportunities.

At this point, we strongly believe the biggest hurdle that stands between us and achieving market share that has never been seen in automotive retail before is our ability to execute our plan.

As a team, we couldn't ask for more and we are motivated to make that dream a reality.

We had our 19th straight quarter of triple-digit growth in units and revenue in the third quarter.

To illustrate the power of that compounding growth, we delivered more cars to our customers in the third quarter alone than we did in all of 2015 and 2016 combined.

During the third quarter, we continued to meaningfully expand our population coverage and importantly, we continue to see strong trends in market share penetration across our markets.

The demand we're observing and our execution to date has only resulted in even greater optimism about the future we see in front of us.

The progress we've made in GPU was also very exciting and lights the path to profitability more clearly than ever before.

In the third quarter, our GPU surpassed the SG&A per unit of many traditional used automotive retailers.

The implications of that are significant and it deserves a moment of contemplation.

It's a powerful step but it's just one of the steps along the way.

Our midterm goal of $3,000 is clearly in sight and we're beginning to get more concrete visibility beyond $3,000.

This will be among the topics we will hit during our Analyst Day on the 29th.

We also continued our march towards profitability in our expense line items, even with our efforts to alleviate demand-generated pinch points in the business and our investments in additional scalability to prepare for 2019.

SG&A per unit ex-gift decreased by $702 year-over-year with about $230 coming from leverage and customer acquisition costs and the remainder coming from non-advertising SG&A.

This allows us to achieve 8.3% EBITDA margin ex-gift for the quarter, reducing those losses by over 750 basis points year-over-year.

Given the quality demand we're seeing from our customers, our focus on the further scalability of our business continues.

We've taken many forms internally and is well underway.

One visible sign of this is the acquisition of Propel AI, which we completed yesterday.

Propel is an artificial intelligence-powered communication platform that delivers automated conversations with customers via SMS, e-mail and on-site chat.

We expect this technology to improve our customer experiences further and to make us more efficient over time as it rolls out and is integrated into our platform.

As part of this acquisition, we gained a highly experienced product and technology team that we are excited about and that will be joining Carvana full time.

One of these leaders is Tom Taira, co-Founder and ex-Chief Product Officer at TrueCar.

Tom immediately earned my respect several years ago when we first met in TrueCar's offices and I'm excited to finally get a chance to work with him every day.

Initially, he will focus his efforts on integrating Propel and on our business of buying cars from our customers.

As noted last quarter, we've been working hard to give our customers a better way to sell or trade in a car.

We continue to see exciting progress for the third quarter.

The total number of cars we bought from our customers, including both retail trade-ins and cars bought from customers that aren't buying from us grew 273% year-over-year in the quarter.

This now represents 37% as many cars as we are selling to customers, up from 21% 1 year ago.

Many of these cars are certified in our inspection and reconditioning centers and are sold to our buying customers.

In the third quarter, 16% of the cars that we sold to customers were bought from other customers, up from 11% in the second quarter and about 6% in the first quarter, just 2 quarters ago.

The remainder of these cars are sold at auction and those cars flow through our wholesale line item.

This obviously very rapid progress speaks to several underlying drivers, specifically our ability to give our customers an actual value online, in minutes and then pick a car up from them on their time is resonating.

Our brand is attracting customers at scale that have an open mind to new offerings and trust us to deliver on those offerings.

Our team is composed of high-quality people who can rapidly execute when their energy is focused in a particular area and there's a lot of low-hanging fruit and latent opportunities in Carvana, due to the fact that we're still less than a 6 -year-old company and haven't yet had the time to focus our energy on all the important places.

We believe the business of buying cars from our customers is still in the early phases with lots of growth and optimization in front of it.

While our Q3 progress has raised our confidence and expectations in this area, we want to continue to manage expectations carefully over the shorter term.

We will remain focused on optimizing and understanding of how to best run this part of the business and this may lead to variability in both growth rate and margins.

But make no mistake, this is an enormous opportunity and we are focused on its massive long-term potential.

Looking forward, we are optimistic, confident ambitious and energized.

We have a product people love, we have nearly endless ideas on how to make it better.

The unit economics of the business continue to prove out.

We have clear visibility into GPU and SG&A gains.

We have executed for 19 straight quarters of triple-digit growth and the team that delivered that growth over the last 5 years is motivated to keep delivering for the next 5 and beyond.

Rarely in life do you outshoot your own ambitions.

We started this journey with big ambitions and the data we are seeing and the people we are surrounded by have only emboldened us.

We'll be working hard and having fun as we chase our goals and we'll continue to keep you informed along the way.

Mark?

Thank you, Ernie, and thank you all for joining us today.

Unless otherwise noted, all comparisons are on a year-over-year basis.

We're pleased to announce another quarter of triple-digit unit and revenue growth, increasing total GPU and improving EBITDA margin.

Retail units sold totaled 25,324 in Q3, an increase of 116%.

Total revenue was $534.9 million, an increase of 137%.

In September, in recognition of Carvana selling its 100,000th vehicle early in the year, Ernie committed to giving all employees shares of his personal stock.

We provided a lot of detail on this in our shareholder letter and to ensure investors have visibility into our key metrics, excluding Ernie's gift, we will refer to several non-GAAP measures, presented as ex-gift, with reconciliations available in our materials.

Unless otherwise noted, all gross profit SG&A and EBITDA metrics mentioned by Ernie or I on this call are on an ex-gift basis.

Our guidance in past quarters has corresponded to ex-gift results and we expect to provide guidance on ex-gift results going forward.

Because the gift is awarded after each recipient's 1-year anniversary date, we expect it to impact our GAAP financials through the first half of 2020.

While this means readers of our future financial statements and shareholder letters will have to get used to some new terminology, we believe this gift was a very positive event for our company and its shareholders and it very much reflects our commitment to one of our core values: We're all in this together.

Returning to our Q3 results.

Total GPU, including gift was $2,263 and total GPU ex-gift was $2,302, an increase of $560.

This increase in GPU was driven by a significant reduction in average days to sale to 63 days this quarter, down from 97 days a year ago, as well as gains in financing and ancillary product gross profit.

EBITDA margin, ex-gift, was negative 8.3%, an improvement of 760 basis points year-over-year.

In addition to GPU gains, we are showing significant operating leverage while simultaneously expanding our geographic footprint.

We opened 13 new markets in Q3 and have opened 4 more so far this quarter, bringing our total to 82.

We now expect to open 40 new markets in 2018, bringing our end of year total to 84.

We also made significant progress toward increasing our inspection and reconditioning center capacity in Q3.

Fifth IRC in Indianapolis remains on track to begin producing cars around the new year.

And we are currently evaluating sites for additional centers in 2019.

As awareness of our customer offering grows, we believe scaled nationwide reconditioning operations will be a key driver of our customer experience while also bolstering our path to profitability.

Additional IRCs allow us to improve selection for the customer while also reducing delivery times and enhancing logistics network efficiency.

On November 2, we upsized and extended our forward flow agreement with Ally, increasing their future funding commitment to $1.6 billion.

We believe this transaction gives us significant flexibility to provide customers with our seamless online financing experience while also enhancing our ability to monetize the finance receivables that we originate over time.

Deep knowledge and expertise of automotive financing is one of the core strengths of our company and we believe we are just scratching the surface of the financing and monetization opportunities in our business.

Since our last call, we've also completed several transactions to maximize our financial flexibility.

In September, we completed an offering of senior unsecured notes, generating net proceeds of $342.5 million.

The notes have a 2-year no-call period and mature in October 2023.

In November, we upsized our floor plan and credit facility to $650 million from $350 million and extended the term by 2 years.

We believe these financings represent flexible and efficient sources of capital for us as we continue to execute our plan.

In terms of outlook, for the fourth quarter, we anticipate continued rapid growth in retail units and revenue.

We expect retail units sold of 27,500 to 30,000, an increase of 103% to 122% and revenue of $570 million to $630 million, an increase of 115% to 138%.

We are also raising our outlook for the full year to 93,858 units to 96,358 units and $1.94 billion to $2.0 billion in total revenue.

We expect total GPU ex-gift to be 2,000 to 2,250 in Q4.

This reflects a normal season decline in the fourth quarter, including the anticipated impacts of our annual Cyber Monday promotion.

We are also raising and narrowing our GPU ex-gift outlook for the full year to $2,100 to $2,175 versus $1,539 in 2017.

This reflects a strong performance we have seen across multiple parts of the transaction as we march toward our midterm goal of $3,000 and beyond.

We expect EBITDA margin ex-gift to be between negative 8.5% and 10.5% in Q4, an improvement from negative 15.5% last year.

This outlook reflects the impact of our recent acquisition, a higher share price on stock-based compensation, and increased investments in technology and logistics to prepare us for anticipated sales growth in the first half of 2019.

You should use approximately 149 million weighted average shares on a fully exchanged basis in Q4 and approximately 150 million in Q1 2019.

As we look forward to the first half of 2019, we expect continued improvement in total GPU and EBITDA margin and we are excited of our continued rapid growth and progress toward our financial goals.

Thank you for your attention.

We will now take questions.

(Operator Instructions) Our first question will come from Sharon Zackfia with William Blair.

I guess just a few questions.

On the acquisition of the AI business, could you kind of quantify what that meant for the fourth quarter versus your original outlook in terms of the incremental cost?

And then secondarily, I'm not sure if you have another refinancing of receivables in the fourth quarter.

I thought that might be the case.

And I'm curious just given the other refinancing if you were able to get any kind of more advantageous terms with Ally.

Yes, so first, Sharon, this is Ernie Garcia.

So on the Propel acquisition, we expect approximately $2 million to run through expenses in the fourth quarter related to the Propel acquisition.

Now I'll sort of take the opportunity to kind of touch on that acquisition for a second.

We're very excited about the acquisition as we said.

We've got a great team coming on board.

Their technology is very, very interesting.

They have the ability to parse any tax increase from customers and then translate that into intents that our computer can understand.

I think our vertical integration makes our ability to integrate that much, much more significant.

We can basically take those increase and put them all into the services we control around financing or trade-in, the purchase process itself, any of those different processes, we're going to be integrate with this technology in ways that will be very, very difficult for other companies that don't kind of own the entire stack of customer experience to do.

So we're excited.

We'll also be able to innovate that not just through SMS and text but also on-site and on our website itself.

So we're very excited about that and the team is great so we're super excited.

We did a refinancing acquisition in Q3 that did flow through the financials.

We expect to do another in Q4, which will also flow through the financials.

The terms with Ally, I would say twofold, very similar, would be first order in terms of how I'd characterize it in that I think there are some structural differences that make it a little easier for us to do more of these refinancing deals and deals that are similar to them.

That should enable us to continue to improve our finance monetization over time.

Although as we've said in the past, we're in relatively early stages of that and that could be somewhat chunky.

But we're very optimistic on where that's headed as well.

Your next question comes from Chris Bottiglieri with Wolfe Research.

Wanted to follow up on the retail GPUs, pretty strong there.

You seem to attribute it a lot to the lower days sales.

Just curious just given the pretty significant increases you saw in the customer trading rate, a lot of that did translate into, like, supporting higher retail GPU, I guess is the first part of the question.

Sure.

So retail GPU improved by about $325 year-over-year.

That corresponded to a production average day of sale from 97 last year to 63 days this year, which was really far and away, the biggest driver of that $325 incremental improvement.

I think there were some various puts and takes in retail GPU was well.

I do think we saw a benefit from buying more cars from customers.

We do make solidly more on cars that we source directly from consumers than cars that we source from auction in our retail business.

We also, though, in the quarter, saw slightly higher retail COGS as we took over 2 inspection centers in their entirety early in Q3 that are also ramping up production, looking towards the first half of 2019.

Yes, and I'm going to jump in there, too, really quick because I think it's a good opportunity to talk about how we're thinking about the business of buying cars from our customers and how that's likely to flow through the financials in the future.

So as I said in my prepared remarks, that flows through in 2 ways: one, cars that we bought from our customers and then sell at auction.

That's just the wholesale GPU and then cars, as you pointed out, cars we buy from our customers then sell to other customers flow through our retail GPU in a way that isn't really separately visible to you.

The reason that we're very optimistic about this business but the reason we want to manage expectations in the near term about how that's going to flow through our financials is because if you look at our wholesale GPU per unit for example this quarter, it's up $355.

If that number moves by $100, then you need to increase units by 30% just to offset it, right.

So there's still kind of a fair amount of sensitivity there, given the size of those margins.

And so we don't want to get to a place where there's an expectation that every single quarter, we're just going to be kind of ratcheting that quarter after quarter.

It may play out that way and it may not.

And we want the freedom to be able to price the appropriate ways and learn from all those different pricing experiments, et cetera.

So I just want to make sure we're kind of carefully managing that, but the opportunity is big and the progress looks significant.

Okay, so that makes sense.

And then from a Blue Sky perspective, if you think about getting your DSOs down to 30 days over some indefinite period in getting your sourcing rate up another, call it 30 points, and hold your value discount, what does that retail GPU look like, majority?

Is there a way to contextualize that?

I would just continue to point to the frame we've given in the past, which is every day that we reduce days of sale is worth approximately $10 of GPU, all at constant.

And that has some kind of seasonal fluctuations in it, but roughly speaking across the year, that's the right number.

I think our $3,000 midterm GPU goal has always had underneath it an implied turn-on goal of 30 to 60 days.

We're making a lot of progress against that.

We feel really, really positive about it.

I think as we continue to grow across the country and build out more inspection centers, we have for shorter inbound transit distances to those inspection centers, which goes into our days to sale.

I think there's a lot of opportunity to improve there.

I think the obvious areas of improvement is just we continue to grow sales against the fixed inventory.

We can push down website days and then kind of the other bucket is basically days of the inspection center where there are probably some gains as well.

So from a Blue Sky perspective, I think there's definitely upside potential but I think we remain focused on kind of midterm goal and we think 30 to 60 days is a good place for us to be aiming for now.

And the next question will come from Zack Fadem with Wells Fargo.

On the unit growth guidance, could you talk a little bit about some of the Q4 dynamics around seasonality and promo?

And specifically, do you expect the unit trends to be consistent throughout the quarter?

Or are you anticipating an uptick around the holiday in November and December?

Yes, so I would say our guidance reflects our expectations would be kind of the highest order of response to that question.

I think if you're asking us to conceptualize it versus last year, last year was a strange year.

We had hurricane replacement, we were moving trucks around the country to try to alleviate that, we had some noise around our Cyber Monday promotion.

We expect to do the Cyber Monday promotion again this year.

So I don't think we want to get into kind of inter-quarter guidance, but you can see what our overall expectations are.

Got it.

That makes sense.

And with the Indy IRC coming on in the end of the year, what should we keep in mind for modeling purposes as those new costs are layered in?

And how should we think about the impact to volumes and days to sale as the facility is up fully and running?

Yes, sure.

So in terms of average days of sale, I don't think you should think of Indy as making kind of a meaningful difference.

I think we have a global inventory, and by global I mean nationwide, where all of our customers are drawing on all of our IRCs.

So I think in terms of moving on average days of sale, I don't think there's much to expect there.

In terms of the impact on sort of retail COGS, I think there could be a small impact there as we get things up and running and are a little bit less efficient early on, including we often start new IRCs with a little more outsourced kind of local services that we then in-source over time.

And so that can be a small headwind as we're getting them up and running, but I wouldn't think of that as being particularly large given that we got 5 IRCs up and going right now.

Yes, and I would add, in kind of a more medium-term perspective, I think you add that inspection and running, that means we have more cars located closer to all of our customers in the middle of the country, which just means that we're able to deliver cars faster to customers, which has conversion rate impacts and they have to travel lower amount of months, which means that we have lower costs.

So I think that's an exciting development for sure through the medium term and in the short-term, there will be the ramp-up costs associated with it.

And our next question comes from Mark May with Citi.

This is Nick on for Mark.

Just 2 questions.

The first on retail ASP, I notice that's kind of been creeping up year-on-year.

Can you give an update on where you are compared to Kelley Blue Book there?

And then second question is on population coverage.

You've called out around 200 markets or 200,000 people or more.

Kind of my quick math shows that's about 78% population coverage, is that kind of the long-term bogey?

And would we expect you guys to kind of aggressively go after getting that coverage?

Or will you reassess at the end of this year after you're about 57%, which is about 3/4 of the way there?

Sure.

Yes, so our changes in ASP year-over-year are almost entirely mixed.

We talked a little bit about this in Q2, where, going from Q1 to Q2, we saw a meaningful increase in ASP to the mix.

And we typically see an increase quarter-over-quarter, going from Q1 to Q2, as newer cars and lower mileage cars sort of start hitting the used market.

That has more or less maintained in Q3.

I think ASP was down $200 quarter-over-quarter as we started to normalize our mix a bit.

But and we do expect further normalization index over time.

But I think the ASP, by far, the largest driver of our ASP is mix and that, in turn, is driven in large part by what's happening in the market and also what our customers are demanding on the site.

Yes, let me just comment on your population coverage question, I think your math is right.

If you take the 200 largest DMAs and add them up, 78% population coverage is approximately right.

I think it's important to also note that in general, we have delivery radiuses around each of our markets that are larger than kind of the line drawn around the DMA.

So with that number of markets, we'll be able to cover more of the population and that, but I think that, that's a good framework to be starting with and that's probably a topic that we'll touch on at the Analyst Day on the 29th as well.

And our next question comes from James Albertine with Consumer Edge.

Wanted to ask, with your new IRC up and running, can you give any or can you help us delineate kind of what your aggregate capacity is in terms of processing vehicles in a perfect run.

Now you're going to be building out over time, there's some -- there's a slope to that curve, but if you were to sort of put everything in through your facilities today, how much could you handle with that new IRC up and running?

Sure, so the Indy IRC is similarly sized to our previous 4 IRCs, which means it has about 50,000 units of annual production capacity at full utilization.

And so combining that with the 4 that we are already operating from, that's 250,000 units of total annual capacity at full utilization.

Understood.

And I think related to that, as we've seen you grow out across the top DMAs across the country, SG&A is still somewhat higher I think than we had modeled in terms of the linearity of the quarter, nevertheless, you're showing good leverage.

I'm just wondering, how quickly can that start to step down?

And what is the right -- I'm not sure if you've ever shared this publicly, but what's the right long-term SG&A-to-units metric that we should be thinking about from a modeling perspective?

Yes, I would say we're pleased with our growth.

We're definitely showing a lot of leverage.

We expect to continue to show a lot of leverage.

We're obviously investing in getting prepared for next year and kind of all the growth beyond that from a technology perspective, from an IRC perspective, from a logistics perspective.

So there is investment in that, that we will continue to make because we've got a much longer-term gains that we're planned here.

So I think you'll continue to see investments.

We haven't given a specific SG&A target as a percentage of revenue.

But I think that's also something that we will likely discuss to some degree at the Analyst Day.

And then kind of on the other side of the economics on the gross profit side, I would just say that we continued a lot of progress toward our $3,000 midterm goal.

We're very proud of that progress.

And our expectations have only gotten stronger there in terms of what we think we will be able to ultimately achieve.

So we're excited about that as well.

Our next question comes from Sameet Sinha with B. Riley.

A couple of questions.

First, considering that demand for used cars seems to be unseasonally strong as per industry data, but it seems like you're still trying to have that Cyber Monday sale.

What's the value of having that, considering that on one hand, you're stockpiling for a strong -- seasonally strong period just 2 to 3 months away?

Seems kind of counterintuitive.

Or maybe the strategy is to kind of sell -- just so you get to sell some of those slow-moving vehicles.

And the second question is, Mark, probably this is for you, CapEx came in slightly below expectations.

Has there been sort of a push-out or some sort of increased efficiency there?

So I'll start with the Cyber Monday and then Mark you can come in on CapEx, so I think your framework for Cyber Monday, I think is right if we're optimizing for now or the next 6 to 12 months.

We're heading into 2019, expecting to see lots of growth there.

We want to have cars available for our customers there.

We want to make sure that we are smart about managing our GPU so I think.

If we are optimizing from an immediate term, that's right.

I think the value that we find in Cyber Monday is that kind of the one time per year where we can do promotion that cements our brand.

I think in general, we like to be very careful about promotions because we think that promotions have been heavily associated over time in consumers' minds with less than transparent, simple transactions.

And so we stay away from that outside of this one event where we feel like we're able to really kind of bang the drum of, we're an online automotive retailer and we're able to tell our story through this promotion.

And so I would think of it as a brand-building effort and accordingly, we're going to continue to do it.

That's something that's playing out over a long period of time in the next couple quarters.

Sure.

And then, Sameet, on CapEx, our expectations for CapEx have not changed.

I think there can be some fluctuations quarter-to-quarter.

For example, in Q2, we had a large outlay for the initial purchase of our Indy IRC.

And so that -- things like that can cause some fluctuations but overall, our expectations for CapEx have not changed.

If I can sneak in one question.

So in the shareholder letter, you're indicating you're staying close to existing markets as you opened the new markets, staying close to the equipment.

Is that a strategy that you plan to adopt even if you continue to head towards that kind of 200 market target?

So I think as we continue to launch markets, I think it'll be a mix of network expansion markets and network density markets.

I think we are largely expanded coast-to-coast now.

So there's still a few large network expansion market left.

But as we look forward, after a couple of major network expansion markets, there will be a lot of network density markets as we're marching toward continued market openings over time.

And our next question comes from Colin Sebastian with Robert W. Baird.

It's Ben on for Colin.

First, I know you guys have provided average penetration rates across the markets before.

Is there any update to provide there?

Or would it be accurate to say that the markets of rolled out in 2018 are performing better at the same vintage or at the same stage as markets rolled out in 2017?

And then I have a follow-up after that.

Yes, so we've historically provided that data after the fourth quarter.

And we haven't provided it in between but what we said in our prepared remarks, I believe, in shareholder letter, we continue to see consistent uptick in consistent patterns in the way those curves are unfolding.

One of those patterns we called out in the past that newer markets are generally ramping faster than older markets.

And nothing significantly different there.

Got it.

And then secondly, on GPU.

You guys have been coming at the high end of that range over the last couple of quarters.

I know on the last call, you mentioned some potential GPU headwinds in the scaling of the Dallas and Philadelphia facilities.

Should we expect that to be more a fourth quarter impact?

Or is that kind of lumped in with the Indy scaling as well?

I think the way we talked about that on the last call and are still thinking about it is that's going to be a back half impact to a degree as we move into those 2 facilities, as we start get Indy up and running, as we discussed.

There's some potential headwinds there on retail GPU that are built into our guidance.

But overall, we're feeling very good about the way we're scaling these IRCs.

We're looking towards the first half of '19, making sure we're prepared to ramp up production, increase selection and be where we need to be there.

(Operator Instructions) Our next question comes from Nat Schindler with Bank of America Merrill Lynch.

Can you help me out on what's the average FICO score of your buyers that are using financing on your site?

And how has that trended since you doubled since IPO or actually call it, massively more than doubled?

Also can you -- it doesn't look like Ally is now taking all of your loans based on my calculations.

Are you selling to others now as well?

And if so, who and what type of loans are you selling to other people than Ally?

Yes, so first of all, FICO question, I would say our FICO continues to look a lot like the broader used car market.

So if you look at other leading retailers or just kind of used car salesman in general, we've got the very similar FICO distribution to any of those retailers and that's been very stable across time.

Nothing to call out there.

What I would say is Ally is buying many of our loans and then they are also providing financing to other buyers that we've been able to refinance in refinancing transactions.

That, I would say, is in concept, somewhat similar to the way securitization markets kind of works.

And we'll probably continue to develop more of those financial buyers over time.

And those are some of the structural changes that we're talking about that we believe we'll have access to over the next several quarters.

We continue to bring more people in.

But Ally remains our biggest partner by a long way.

And our next question comes from Gary Prestopino with Barrington Research.

Could you maybe just talk a little bit about the profile of the car that you're retailing from consumers?

I mean they tend to be older cars, something like that in mileage, things like that.

In terms of the cars that we're acquiring from customers that we then retail?

Right.

Yes, so on average, they do tend to be a little bit older and a little bit higher mileage than cars that we sourced through our main auction channel.

And I think that's fairly -- our understanding at least is that's fairly typical in the industry.

Yes.

And what I would think -- Gary, just to add to that, I think one of these are exciting is that opportunity is we saw an enormous GPU opportunity.

It's another way for us to touch more of our customers and give them a great experience, but also it's a way for us to get access the fairly unique inventory that is hard to find.

So for example, on average, these cars that we buy from customers do sell faster on our site than cars that we buy from auction because they are more unique relative to the rest of our inventory.

And so we think that's another kind of hidden benefit of growth in this business.

And our last question today comes from Ron Josey with JMP securities.

I just had 2 real quick.

Last quarter, Ernie, you talked about automating a lot of what are still very manual processes, just -- and that's understandable given the relatively early days of the business.

But just wanted to understand if you could talk a little bit more about where you are in the automation?

I'm sure it's still early days but trying to understand sort of the progress here, maybe you can give us in innings or something and then an apology -- apologize if you've talked about this already, but can you just give us more content on the Ralph Breaks the Internet campaign that you're doing.

I think that's really interesting.

Sure.

What inning are we in for automation, that's a hard question.

Very early, I would say.

I think as we've talked about in the past, our kind of framework for prioritization has always been customer experience first, and then our financial goals, one, units; two, GPU; and then three, operating leverage which is a unique part of operating leverage.

It's basically operating expenses.

So I think that is always been kind of our lowest priority area of investment and so it has not gotten as much attention as many other parts of the business.

We started to invest more there over the last several quarters.

And I think that the motivation there is partially financial.

I think more than that, it's more about just getting the business to be even more scalable than it is as we look at kind of big growth in units in percentage terms in absolute terms, looking forward.

So I don't want to quantify what inning it is, but I would say very early.

And I would also say Propel is a very interesting technology solution that is going to help us to organize a lot of our data and present it to customers in ways that are more self-service and also in ways that helps to automate things that we're doing our own side.

So I think that fits into that general theme.

As it relates to Ralph Breaks the Internet, we're very excited about that collaboration.

Obviously, Disney's an enormous company that we couldn't be more excited to be associated with and the movie itself is great.

My kids are excited about it.

So I can't wait to go watch it with them and have our logo show up in there.

We think that that's really fun.

We also think this is potentially a broader movement of -- that may be kind of thematic, not just for us, but marketing in general where there's more and more marketing built into content that people want to watch.

And so this our first foray into that, that we're doing it in a pretty big way.

And we're very excited about it.

We have no idea whatsoever what to expect but we'll be watching it closely and hopefully, something that we do again in the future.

And we do actually have one final question today from Seth Basham with Wedbush Securities.

My question is around the cost.

If you could elaborate for us, your implied EBITDA guidance for 2018 went from negative $152 million to negative $189 million at the midpoint with GPUs not changing much.

I presume that the biggest issue here is cost.

You called out $2 million extra cost for your acquisition in the fourth quarter.

What are the other large cost buckets driving that decline in EBITDA guidance?

Sure, so I think there's a number of things to call out there.

So I think at the midpoint of our current EBITDA guidance versus our previous EBITDA range, it's about $10 million difference that we're building in versus that previous range.

That basically boiled down to a handful of things.

So Ernie mentioned $2 million for Propel.

We also had about $2 million in increased stock-based compensation expense from the run-up in stock price.

We had about $2 million versus our kind of previous expectations on logistics network expense and also engineering and technology hiring.

And then another $2 million that was a smattering of other things.

And so I think those were the major things to call out since our last call.

And I think obviously, we feel like all those things are obviously intelligent things to be doing looking toward 2019.

I think we're feeling really good overall about all the progress we've made this year.

Obviously, levering SG&A.

I think at the midpoint of our EBITDA guidance for the year, it's about 740 basis points of EBITDA improvement year-over-year from '17 to '18.

That's actually higher than the amount of improvement we had going from 2016 to 2017.

I think we feel great about continued margin improvement.

I think as we're looking to 2019, we see EBITDA losses declining meaningfully, looking to next year.

And so I think we're feeling really great about all the progress we're making while doubling markets, investing in technology for the long term and still showing substantial leverage while doing all of that.

Yes, and this is Ernie.

I want to add to that, I think it's useful context for the fact the last 3 years.

We've only levered from Q3 to Q4 in 1 year and that was last year.

The previous 2 years, we delevered.

Because the normal seasonal factors, you have higher depreciation rates with impact GPU, we have the Cyber Monday promotion.

Generally speaking across the industry, it's a little bit of a softer period.

And then we're generally investing in preparation for all the sales that we see coming in, the first half of next year.

We, in our previous guidance, had implied that we would lever again this year and we wanted to make sure that we did that while simultaneously preparing for next year.

I think as we headed through the quarter and we made this acquisition and then we saw some weird quirks happen with the stock-based compensation, that impacted that a little bit.

And then we have to make choices in real time about how the scale and how to prepare for scale going into next year.

And we felt like a handful of millions of dollars was not a difficult decision there, given what we expect going into next year.

So that's what drove that decision to change the guidance.

Got it.

Helpful context.

And then my follow-up is on other GPU this quarter.

Ex the $158 per unit that you earned from the $4 million refinancing fee, other GPU was flat with the second quarter and also we saw more limited gains year-over-year than the second quarter as well.

What's going on there?

Any change in the financing penetration rate?

Or other factors impacting that other GPU this quarter?

So yes, I think the other GPU, as you said, was roughly flat sequentially after kind of adjusting for the finance monetization fee.

I think there's some small puts and takes there.

Interest rate continue to rise, which has a little bit of an impact on finance premium.

And then we've made some gains actually in attachment of vehicle service contracts to some of the initiatives that we've undertaken.

Started to see some of those gains in the third quarter and believe that'll be a positive for us going forward.

And Seth, I'm going to jump on top again, I apologize.

The last point that I do think is important to make here is and I said in my prepared remarks but I want to make sure that it's heard.

I think our GPU this quarter, being above the SG&A of many scaled used automotive retailers is a really interesting data point.

I think that carries a lot of implications.

And we still honestly believe there's lots of upside to our current GPU and we believe that our cost structure is going to be meaningfully lower over the long run than other traditional automotive retailers.

So we think that's a great signal and we're on that path and we're working as fast as we can to get all these different line items.

And this will conclude our question-and-answer session.

I would like to turn the conference back to management for any closing remarks.

Thank you guys very much.

Thanks everyone on team Carvana.

Great quarter.

We still got a lot of work in front of us.

We did a great job this last quarter.

And then to investors and analysts, we look forward to speaking to you at Analyst Day.

It should be a fun day.

Thank you very much.

The conference has now concluded.

Thank you for attending today's presentation.

You may now disconnect.