CarGurus, Inc. ($CARG)

Jason, thank you so much for your time.

Thank you.

We're going to get into it. And so Jason, when we see the stock trading at these levels, it kind of pops out on the screen as one of the most attractive valuations across the Internet sector, not only within online autos, but all of Internet.

What do you think has been most misunderstood about the trajectory, whether it be your recent results, guidance or the broader AI narrative?

Thanks for the question, and we'll try and talk over our neighbors here. So hopefully, everyone can hear us. So it's hard to say what may be misunderstood. But I mean, if I take the different elements of what you just talked about, from a results perspective, we're very proud of our growth. Most recently, grew 15% in Q1 year-over-year, have had multiple years of double-digit growth now, and that's largely through product innovation and product expansion. Our EBITDA is growing just as nicely. We have really strong margins in the 30s. And I think maybe, actually, something that might be missed is just the strength of our EPS growth and free cash flow per share growth over the last several years because we bought back so many of our shares. So EPS from '23 to '25 grew at a CAGR of north of 50%. On the guide, we guided to, again, double-digit growth in '26. We also talked about a little bit of margin compression that is really intended to maintain our product innovation velocity, especially as we move into software and data serving dealer or dealer customers with software and data as well as marketplace, which that requires significant product builds, but that's also helping support our growth of wallet share with our customers. And then on AI, we think of it, and I'm sure we'll talk more about it today, but we think about it in a few different dimensions. One, we're positioning ourselves really well with the LLM. So we are winning in our category for representing well in the AI piece of the funnel. Two, the trust of our experience, shopping for a car is a multi-month experience. We have extraordinary proprietary data and integrations, which we'll talk more about. But we have the trust of dealers and consumers to go through that multi-month journey, which is important in AI. And then lastly, and maybe the most exciting is that we're building AI native tools ourselves for both consumers and dealers and AI products that are leading the industry. So we're not just sitting idly while AI advances quickly, we're actually, we believe, at the front of that curve.

And when you think about how AI changes the car buying experience over the next 3- to 5-year period, where does CarGurus fit into that change? And how does the consumer kind of change their experience?

Yes. We've defined the consumer shopping journey in 3 segments. There's research consideration and then purchase. So research is determining what type of car one might like. Consideration is which of those used cars, typically, could be a new car too though, which of those cars is best for you and then purchase is your experience with the dealer to actually complete the transaction. And so we're focused on advancing each of those. There are certainly threads that tie across all of them, but those tend to be 3 pretty discrete experiences. And we're transforming that experience into a full AI modal experience for our consumers. And so I think the 3 themes of things that are going to change most with that AI evolution for us is: Number one, it's moving from curated information to an expert guide that makes recommendations. That's really powerful and useful to the consumer, and we're seeing them adopt that and really embrace it and really engage with it. Number two is it will reduce the human effort. So shopping for a car will still take a couple of months. Today, on average, it takes 3 to 4 months. It's still going to take time because consumers need to line things up. They need to do their research, they need to feel confident. But the time that they invest during that 2- or 3-month period, we think will be less because agents can do a lot of that work. Agents can do the comparison. Agents can seek out alternative options for cars. Agents can start to engage with dealers in useful ways for both sides. And then the third theme is it will optimize the result for both sides. So there's a lot of information asymmetry and a lot of confusion in car shopping. AI and agents can help reduce that. So it gets to a better outcome. That is a positive outcome for both the dealer and the consumer. I don't think that will necessarily have people buying more cars than they otherwise would. But we do think that it will lead to better outcomes and better deals and transactions that are in total better for the car industry.

Got it. And you have an early ChatGPT app integration. Are there any learnings from that so far? We hear about higher conversion from these types of LLM experiences. Is that something you could speak to or any other learnings initially?

Yes. So just to briefly put it in context, as I said, we're doing well in positioning ourselves with the LLMs. We are the #1 traffic destination in the auto category from the LLMs. We are #1 in visibility on the LLM search engines in our category. And then we were the first, as Michael just mentioned, we're the first to have an app in the ChatGPT app marketplace. . The traffic that we get from that channel is still about 1% of our total traffic. So just to put that in perspective, it does though, as you said, convert well. It converts about 2x what our normal traffic does. A lot of that traffic comes into a product we have, which is our AI virtual assistant called Discover and the consumer engagement with Discover, while it's a small percent of our total audience, their engagement is deep. They have approaching double-digit prompts through a search process and recommendation process. That is now seamless between the research phase and actually recommending cars. The amount of information that they give us is extraordinary. They're not searching for a make-model trim. They're searching for what their family situation is and the weather and what type of car and how they want it to drive and we're then recommending the make-model trim out of that. And that's really important because we're then parlaying that information into packaging a better consumer profile for the dealer. So then when that consumer ends up connecting with the dealer through us and walks into the dealership, if that dealership has been able to get their salespeople to really leverage our platform as much as they can, they tap into something called shopper signals and they'll then see that Jason has come in, and he's looking for a particular car, but it's because Jason has a family with 3 kids and a dog and lives in New England and is looking for this type of performance, and they can use that to convert those leads much better to sales. .

And so conversion is double through LLM, which is a big leap. How are dealers reacting to learning about this and changing how they might maybe monetization or just how they're operating their ad spend.

So we have -- I talked about the 4 dealer pillars and one of them is conversion, converting leads into sales. And we have a number of features and products that we offer to dealers to help them convert better. So shopper signals, which I just mentioned is one. Another one is a product called Digital Deal, which allows consumers to do a number of elements of the transaction on our site before they walk into the dealership, they can get a trade-in value, they can put down a deposit. They can set up an appointment, they can buy other F&I products from the dealer. So then when they walk in, they're much further down funnel. Couple that with Shopper Signals, and you are much closer to a sold car and really understanding that customer. The challenge -- there's no shortage of data. The challenge is actually behavior at the dealership. It's dealerships facilitating and training their staff, which typically has high turnover to actually leverage these tools so that they can have a higher conversion rate. We have a group called Dealer Performance Partners, and they go in and work with hundreds of dealers a year for a day or 2 to make sure that they're getting the most out of our platform. They're using all the data and tools that we have to offer. They're leveraging best practices and they're understanding their competitive set better than they currently are. And it's not uncommon for that group to double the conversion rate of that dealer customer, which is extraordinary. So AI is a part of how dealers can do that, but there's actually it's not just AI, it's more around behavior change.

Yes. And we've been hearing -- we just had an e-commerce panel where we're hearing about different conversion-related data points and talking through categories. And it seems like the conversational search experience through LLM lends itself to more complex categories, less consumables, higher velocity, more kind of complex. So would you go so far to say that auto is probably one of the better use cases for conversational search across categories?

Auto is very high consideration, second typically to homes, and it's very complex because literally, no 2 used cars are alike. Make-model trim, mileage, options, condition, location, et cetera. And hence, that's why it's a multi-month process. the much higher upper funnel LLM experience to begin that process I can definitely understand why that would be a good use case for it. But I actually think that there probably comes a point I haven't thought about it this way where it becomes so complex that you actually need that period of research. You need that confidence building exercise that you have to go through. And so we don't believe that car shopping is going to be a zero-click experience. It just is far too important of a decision. And you saw this years ago when Google introduced what are called VLAs, vehicle listing ads. It went from just links to sites to a carousel of cars. And that has become a really wonderful marketing channel for us. We performed very well there, and consumers will sometimes select a car from the carousel, but then they come through to us to do the research. And we believe that LLM are similar that it's hard to imagine that a consumer will gain enough confidence in what has been a multi-month process to make a decision at that superficial level. And furthermore, the LLMs don't have access to all the data that we do.

Right. And that data advantage, CarGurus being, having the most data out of any platform online or off-line, how does that lend itself in an AI ecosystem? I know there's a lot of questions about data ownership. And if there's a intermediation risk there, web scraping and things like that, how do you protect your moat?

So yes, so data is certainly part of the moat answer, but it's not the only part. Data alone is not enough. But we do have proprietary data. So we -- again, just to set some context, there are about 42,000 to 45,000 dealers in the U.S., about 65,000 dealers total in the 3 countries in which we operate, U.S., U.K. and Canada. If I talk about the U.S. for a second, 42,000 to 45,000, we have about 26,000 who are paying us and we have over 30,000 on our site because we have a freemium model. So we have the most dealers, the most inventory, the most paying dealers. We also then have the largest consumer audience with by far the most sessions. And so on any given day, we're collecting about 0.5 billion data points around pricing, inventory and consumer demand. We also get feeds from all of those dealers, and we are integrated with many systems at those dealers. A lot of those feeds are unstructured data. They will describe a car. We then turn that into an ontology that helps us understand exactly what make model trim options are in that car and allow us to calibrate and compare that to other cars. We do pricing validation. We do deduplication. And so all of that plus all of the trends that occur over time, demand trends, a consumer profile that builds over time, those are all things that cannot be glean from scraping. And so that is all data. But it's data over time, some of which is proprietary, and it's what we do with that, that makes that a moat. You then build that into a trusted experience, and we think that is the moat actually, is the trusted experience that you need both sides, dealers and consumers to commit to in order for people to have confidence that that's where they want to transact.

I want to go back to your recent Q1 print. How should investors think about the key drivers behind your Q2 and full year guidance that was updated. What are the main puts and takes from here?

We guided to the year in Q1 and remain -- or at the end of Q4, and we kept that -- we didn't change that guide. So we've guided to double-digit growth this year again. And we -- and the underpinnings of that, and I think you may -- we may be getting into sort of the elements of -- 2 of our important KPIs in what drive our revenue are number of paying dealers and then a metric called QARSD. QARSD Is quarterly average revenue per subscribing dealer, so how much dealers pay us. There are a handful of key drivers in QARSD that have long runways and are really cranking right now nicely. So the underpinnings of the revenue guide are double digit of primarily QARSD, but also rooftop growth. On the margin side, we guided to a little bit of margin compression that's a temporal, not a structural thing. It is a function of as we move into software and data and as we want to maintain our product innovation velocity that we've achieved, we've introduced more products in the last 18 months than we probably had in the last 3 or 4 years. So that velocity and expansion requires investment, and we're really leaning into that. We are definitely getting efficiency, workflow efficiency, a ton of engineering efficiency from AI and agents internally. We're parlaying that or we're focusing that though on productivity enhancements rather than on focusing on margin in the near term because we've seen, as those products -- as that product velocity accelerates, we're getting more engagement from dealers, and it's giving us the license to introduce products in these other pillars.

Can you go into some of the specifics of the products that you're investing more heavily into in the second half?

Yes. So -- on the dealer side, we continue to innovate in our marketing category, which has been our bread and butter. So one key example there is new car exposure, which allows dealers rather to market specific new cars in addition to or in more sophisticated ways than they have in the past. That's a very timely product right now because new car inventory is building up on dealers' lots. . New car affordability is an issue. It's a concern with consumers and dealers are trying to find ways to move that inventory more. So marketing continues to see innovation. Inventory is a category that I'm incredibly -- we're incredibly excited about. It's a big category. It's between $1 billion and $2 billion of spend for dealers, relative to a $3.5 billion U.S. marketplace spend. So it's significant. We had introduced some free products there in the past and dealers just absolutely devoured them. And so that gave us the positive signals we wanted to introduce a pricing product. We introduced in Q4 of last year, it's called Price Vantage. We shared that between PriceVantage and new car exposure, both of which launched in Q4. We expect those to grow 15x this year and be 8-figure revenue stream to the 2 of them. So that has hundreds of dealers and we're seeing really high engagement. On conversion, I talked about shopper signals, digital deal, there's huge opportunity there. We are still pre-standalone commercialized product. And then data, what's really nice about data and market intelligence is that makes every other product smarter. Dealers are very competitive. They need to be. It's a competitive arena. And so we are giving them insights and intelligence around their competition and around the market that they literally just can't get anywhere else. And so that is -- I think of that as sort of an umbrella or a layer over all the other products that we're introducing that help those products become more effective. .

Got it. Can you unpack the dealer ads portion of your growth? And the mix between kind of macro tailwinds versus more execution-driven gains? And where do we stand today in terms of penetration or the overall dealer TAM?

So in the -- as I mentioned, in the 3 countries in which we operate, there's about 65,000 dealers. We have about 35,000 paying dealers, so we're just over 50% penetrated. We're the market leader in the U.S. and we are the market share gaining #2 in both the U.K. and Canada. Canada is, I believe, at a tipping point where we are generating lead quality or lead volume rather, quantity that is, in many cases, on par with the incumbent AutoTrader Canada such that the largest dealer group in Canada called AutoCanada recently announced that they fully switched from Trader to us. So we're a little over 50% penetrated. We're gaining share in all 3 markets, both in rooftops and in spend. And we think that the momentum that we are building by focusing on productivity rather than short-term margin expansion is winning formula. In all 3 markets, we're considered the best ROI. That helps us all sleep very well at night because at the core before we even add any of these other features and products, we know that we're delivering significant value to them.

And can you unpack some of your QARSD growth algorithm recently, it's been very strong. Are the new products contributing to that? And what are the biggest components there?

So QARSD, as I mentioned, is effectively how much a dealer pays us. Our QARSD in the U.S. is around $7,500 a quarter. So dealers -- the average dealer pays is $2,500 a month. That's the sort of baseline way to think about it. That's been growing at high single digit, low double digits for many, many quarters now. The drivers of that, there are 2 top drivers. One is upselling to higher package tiers and the other is cross-selling. . So if you look over the last year, upselling is our #1 driver. If you look over the last quarter, cross-selling new products is our #1 driver. We love both those drivers because those are a function of our innovation and our product expansion. Upselling is typically now a function of we are adding more and more features and value to higher tiers, and that is causing dealers to upsell into higher package tiers. Cross-selling would be PriceVantage, Sell My Car, new car exposure, all of the ones that -- digital deal that we've talked about that are discrete, monetizable add-on products. So those are the top 2. Lead quality and quantity is a driver. So we deliver a lot of customers to dealers, and those are in the form of traditional leads, which are e-mail, phone call, text chat. We also are sending a lot of consumers to walk into the dealership. That proxies for that or that we're sending them clicks to their website. We're sending them people who have clicked on the map in directions. And then a recent very high-growth channel is in our app, we have something called dealership mode, which gives consumers a ton of value when they're in the dealer. It helps them compare cars. It helps them understand financing, gives them a lot of information in a particularly anxious part of the process. And the number of consumers who are checking in at the dealership in dealership mode has grown very quickly recently. And the most compelling stat is that about 80% of those who are checking in did not submit a lead. So that's showing this is another avenue of value that we're delivering to dealers that we have historically not been getting credit for. So the audience quantity that we're delivering and quality is paramount. And then we do have unit price to pull on. So we are considered the highest ROI. For most dealers, when you look at the survey data, we are the highest volume of leads, highest quality leads and highest ROI pricing in this industry is not standard. It depends on where the dealership is, how big the dealership is, the package [indiscernible] that they're on a variety of things. But there is a unit which is a unit of currency, which is cost per connection or cost per lead. And in many cases, we are still below on that metric, are smaller, less innovative competitors. And so unit prices have fueled only a couple of points of growth for us per year over the last handful of years. We don't want to get greedy. We think it's still there in the long term for us to pull if and when we choose, but that's not a focus for us.

When you think about dealership mode specifically, and you gave a really interesting stat there about the 80% of customers not submitting a lead. I guess you talked about that trust advantage that you bring for consumers. What are you unlocking for those consumers? And is there like a path to monetization of that over time a greater monetization?

It's a good question. So today, we don't monetize the consumer at all. And well, we don't monetize the consumer at all and we don't have near-term plans to monetize the consumer. The relationship that we have with the consumer is long term over the duration of their search. And it's increasingly as we build more of these AI features and elements and then pull them together into a more cohesive AI mode, we really are becoming their expert adviser, soup to nuts, start to finish. . And our Discover has memory, for instance, and that memory feeds into their sort order, and that feeds into dealership mode and that feeds into Search A that they have and Search B that they have with 2 different dealers. And so we are really becoming their guide through the whole process. And we don't think that we need to monetize that because if we're their guide for the whole process, then that becomes infinitely valuable to the dealer. And if we can win over the consumer, then we will win the dealer, has always been our philosophy.

Got it. And how do you assess your evolving competitive landscape with AI accelerating development. When we think about, especially increasing focus online from more traditional offline dealers that are investing there?

And is there a specific competitive angle that you're thinking about with that or competition?

I think -- when we look at where the consumer will be in 3 to 5 years with their shopping journey, I think everyone is investing towards that. How do you feel about your place in that evolution versus competitors?

I've said the word a lot today, so maybe I'm beating the drum too much, but it comes down to trust and confidence. And we feel that if we can by creating the smartest, most efficient, most trustworthy, most data-intensive experience with the most inventory, the most dealers, the most seamless connectivity that, that is increasingly what consumers are expecting. I think you're -- I'm sure we're all reading a lot of the same content that is talking about how. When the novelty of AI starts to wear off, the scrutiny of what can I actually trust starts to set in. And we're seeing that with dealers and consumers. But with dealers, when we make a recommendation on how they should price a car, increasingly, they're starting to ask, well, why? How did you come to that, that recommendation. And I think consumers are starting to ask those questions as well. And so by having the ability to start from a position of trust because we have a contractual relationship with the dealer or all of the dealers, most of the dealers, and then reinforcing and substantiating why we're making the recommendations that we are to consumers, we're going to continue to earn that trust. The early adopters of AI are incredibly informative test cases. They are still very early, and it is still very small. And so for all of the AI features that I've talked about here and we've seen this in a lot of our e-commerce peers, their fantastic use cases, and there's really deep engagement, but it's on a very small percent of our total user base. The average user for us is still using the drop-down menus to search for make model trim. So we're trying to push that because we do think it's a better shopping experience, but it is still the tip of the spear.

Are there any questions from the audience for Jason? I want to ask quickly as we wrap up about capital allocation, you've repurchased a large portion of your share base since 2022. How do you think about it from here as we evolve into this more AI-driven ecosystem? And just how you allocate whether or not you invest in product development or elsewhere?

Our 3-tier approach is always how much should we invest in the business largely for innovation, and we've talked about that a little bit. But I would say we continue to invest intelligently, but we think heavily there, and we're not easing up on that despite all the efficiency gains we're getting in the business. We like how we're expanding into different pillars. We like how we're really evolving or revolutionizing the consumer experience on our site. So we're going to stay with our foot on the pedal there. Next is M&A. We have acquired a few companies in our tenure. We hope to acquire more. So that's the second use. And then the third use is returning capital to shareholders. And in the last, I think, 3.5 or 4 years -- 3 to 4 years, we've bought back almost 30% of the company, almost $900 million worth of shares. The way we think about it is on a free cash flow yield. And when we were at 7% free cash flow yield, we thought that was attractive, and now I think we're closer to 10% free cash flow yield. So we think in these times when we might be misunderstood as you said, or we think we have a long runway of solid durable growth, and we think it's a good price, then we're going to get more aggressive. So in Q1, we bought back $175 million worth of shares. We have a $250 million share repurchase approval for this calendar year.

All right. Any other questions for Jason.

The inventory systems, the dealers they manage the inventory themselves, one dominant, I hear that, they provide the inventory management business.

There's not one dominant. There is there's 2 layers. There are true inventory management systems some that one that skews toward price, which is one of the larger ones is called vAuto, which is owned by Cox. So they are inventory management systems, and then there are syndication systems that pull from hundreds, literally of small inventory management systems and then aggregate and syndicate those out. But I would say vAuto is probably the largest in that category and then there are a dozen-or-so other sizable ones and then there's a long tail of a couple of hundred.

And do you have all the data, all of the [indiscernible]

We interact. Yes. We have, I believe, it's 85-or-so percent of the inventory in the U.S. on our site, which is the largest of anyone. And that's by working with virtually all of those hundreds of feeds. And the information that comes in from those feeds, as I think I mentioned before, is very unstructured. The way you described the same car that he may -- if you 2 had the same car in theory, you describe them very differently. We take that unstructured information and turn it into something that can then be compared to other cars for the consumer, which drives our deal ratings and our instant market value and all of those things.

If the dealer more willing to deliver that data to you [indiscernible] is there any push back about, no?

no, that's the underpinning of how they market their cars. And and they know that we do a good job reflecting their cars. We give them advice then on how to merchandise them better, how to market them better, how to think about pricing them. We really try to empower them to be as successful as they can with that inventory that we're showing for them.

And the 15% that you because of large automobile deliver [indiscernible]

No. We have all the largest dealers. No, it's the -- it's probably the opposite. It's the very small ones, or it might be dealers who have a very high-priced strategy because they might be in a rural location and they know they're high priced. But by putting their inventory on our site, we have a deal rating, and we would say that's an overpriced car. And so their mentality is why would I have you say my cars over priced, I'd rather just not have you say anything about my cars. So that's -- there's also a segment of buy here, pay here where we think the price they're showing is not an actual fair validated price, and so we don't accept those. So there's a few segments like that.

All right. We are out of time. Thank you so much, Jason. Appreciate it.

Thanks, everyone.