Qore Conversations
Tune in for engaging conversations about the latest technology, innovations, and trending issues in the automotive industry. QoreAi's podcast delivers expert insights, practical tips and problem-solving solutions. Stay informed and inspired with every episode.
Qore Conversations
Driven By Data: AI's Role in Accelerating Car Dealership Performance
What happens when data and AI combine forces? Like chocolate and vanilla, they're good on their own but magical together. Join us as we promise to unravel the secrets of how data quality and AI can propel dealerships into a future of proactive decision-making and streamlined operations, setting the stage for groundbreaking changes over the next 18 to 24 months.
We tackle the intricacies of data curation, exposing potential pitfalls like biases and inaccuracies that can mislead businesses and inflate customer figures. Imagine marketing to customers who no longer exist—that's a reality when data hygiene is neglected. By maintaining an evergreen state of data, dealerships can leverage AI for cleaner insights and smarter strategies. Through real-world examples, we explore how centralized and protected data platforms are key to unlocking AI's full potential, allowing dealerships to make informed decisions and stay ahead of the competition.
Marketing strategies also take center stage as we explore a shift from outdated mass marketing to more precise, data-driven approaches. Discover how embracing AI can help dealerships understand customer buying cycles, focusing efforts on engaging the right customers at the right time. This calculated shift reduces wasteful spending and enhances customer engagement. By continuously updating customer profiles and adapting to the digital landscape, dealerships can maximize their data for a competitive edge in the market.
Get ready to see data and AI in a whole new light, and learn how harnessing the power of data can revolutionize the automotive industry, transforming dealerships into lean, mean, customer-focused machines.
For more information about QoreAI, visit our website: www.qoreai.com.
Follow Us:
www.LinkedIn.com/company/qoreai/
www.Facebook.com/Qoreai
www.x.com/qoreai
Podcast Directed and Produced by Hired Guns Agency: https://www.hiredgunsagency.com
Hello and welcome back to Core Conversations. It's Sean and Todd.
Speaker 2:Todd how are you Awesome man. I'm so happy to be here on this Friday today.
Speaker 1:We are recording on a Friday. This is true. This is true. Listen to the audience listening and or viewing the episode. We're back. We're talking again about data, but its relationship to ai. That's your favorite right.
Speaker 2:Well, data and ai yeah, I mean I feel one goes hand in hand, like they go hand in hand. It's like chocolate and vanilla you can have either, but you really need both together.
Speaker 1:So yeah, okay, I like that, I like it well maybe that's a good one again, I think, as think, as we kind of want the audience to have an idea of what we want to go after on this episode. The exploration of how transformative data is is important, but it's very important relative to all of these AI applications in automotive there is. I think by this point, people realize that you're, you are literally the leading voice. You, you are the, the thought leader in this space, who is not afraid to speak into the gray area and the contrarian, because there's a lot of people that and I'm not going to name names we don't do that on this podcast, but, um, maybe someday we will in the future.
Speaker 1:But it's not about that. It's it's about realizing that you do have to have some people like Todd who will, for just the purpose of speaking honestly and candidly and truthfully about this, so that, ultimately, the dealers are the ones that we want to see not just take the knowledge, but then execute on it in a way that they don't regret, like we've done so much of that for so many years. When new things come down the pathway, it usually ends up getting weaponized in some way. That makes dealers make decisions that are very, very difficult to unwind. So today we're talking about data specifically in AI applications. Once again, quality maintenance, enrichment, and I want to start this one Todd really around kind of setting the stage of importance. We've talked about it before, but I'm going to give you the floor to say just explain why data is essential in AI, especially in, obviously, the car business.
Speaker 2:Yeah well, thank you, and it's a great question. You know why is data really important in AI? Obviously, the car business. Yeah well, thank you, and it's a great question. You know why is data really important in AI? And if you start thinking about it, well, without data there is no AI. So in our world, you need training data to make AI work, and dealers produce a ton of data.
Speaker 2:The problem is a lot of it's all siloed, never gets connected. But, from an important standpoint, there is no AI without the data element, and it's not just having data, though. Data has to be formatted in a way that AI can consume it, so it is the critical backbone. So, for dealers who are thinking about AI, the first thing they would have to do is get control of their data. At no point can you depend on just third-party tools and data being held in their silos being able to be useful for you. If a dealership says I really want to be an AI driven organization to get the benefits of what AI is already beginning to produce and absolutely will be producing over the next 18, 24 months, they have to get put data at the center of that, and I think that's the number one thing.
Speaker 1:I think a lot of dealers right now, as they're exploring AI, when they're exploring the influence that it will have relative to data, I think they probably have questions around the quality of that data influence in the success of potential AI applications that they're looking at. How do you believe that that data quality influences success in these AI applications? Are there things they should be looking for?
Speaker 2:Absolutely. Look. The truth is saying and I think we've probably all heard this, sean, from our days in CRM, dms garbage in, garbage out. So if you put a lot of garbage and then try to train an AI model on it, you're going to get some, probably garbage outputs. So it's not only critical to put data in one place, but then you want to clean it, hygiene right, organize it to be able to be used, maybe even enhance it, put additional data against the data you already have to make it more valuable or to be able to glean deeper intelligence from.
Speaker 2:So let's say I have Sean Raines from a DMS and CRM perspective, I have. Sean Raines lives at this address, has this email, this phone number. Bought this car was in six months ago for an oil change and tire rotation. But maybe we can enhance that data by saying well, sean is married, has kids, makes this kind of money, has kids, makes this kind of money, likes outdoor activities, likes animals on seams, behaviorally buys a car every five to six years. Right now his household income is X, his household debt is Y and he's behaviorally signaling, based on other activities that can be collected, that he is currently in market. So that's where data tells an amazing story, and I think any dealer should be looking at a lens that the more quality data we can put into a system will be able to leverage AI to its fullest potential.
Speaker 1:That's a great answer, and I think for a lot of dealers who are trying to just get their arms, let alone their minds, around how AI can benefit their operation, I think it's really important for them to remember things like you just gave, which is it is very similar to DMS and CRM in that if what you're putting in is really low quality or shouldn't be there in the first place, then your expectations should be set accordingly, so that garbage in, garbage out is really an easy thing to remember. Like hey, this is exactly the same thought process when you consider data and AI. Well, so I want to move a little bit to the collection side. There have to be challenges. I'm curious to know, from your perspective, what are some of the challenges that dealers face in collecting data and, at the same time, why they should actually take that very seriously.
Speaker 2:Wow, that's a good one. So you know when you think about collecting data. Data is coming into your CRM Some of it's good, some of it's bad but data is coming in. There's no filtering right. There's no guard standing there and saying I need a last name you can put. My phone number is 212-555-1234. You can do it that easily in any system currently in automotive. So, as I see this, it's like putting the data in. Is you know?
Speaker 2:Step one and for many dealers I look at, what I really think is you have all these siloed data across systems. That hurts dealers, meaning being able to actually understand our consumers, and you know you have a glimpse of a consumer in a CRM who's now in, maybe activation mode. You have a glimpse in the D in a CRM who's now in, maybe activation mode. You have a glimpse in the DMS as at the transactional mode. Maybe you have a glimpse in web analytics as shopping mode. Maybe you're using some other data source for conquest or looking to find in-market stuff, and they're all separated. Any dealer today should be thinking like, okay, how do I get all that organized? Is it easy? No, is it worth it.
Speaker 2:Yes, and I think that begins that journey that I think the smart dealers today go. It's going to be a difficult road heading out, yeah, you know, but the gold is out there and we are going to head out and we're going to find it. And I think that the first thing I've realized in talking with dealers, dealer groups is that mindset that they want to really be able to be in control of their business, which to me starts with being in control of your data, Whether it's silent or not. Yes, is it the simplest thing to get it out of a lot of these systems? No, is it in a hundred different formats? Yes, is it clunky? And I'll give you a good example. We looked in one data system I'm going to say the guilty will remain anonymous of this dealership. There was 297 Luke Skywalkers and listen, I love Luke, but I thought he had the little Land Cruiser thing. I didn't think he was buying Toyotas.
Speaker 2:So I come to find out if the one service writer didn't know a name, he just threw him as Luke Skywalker and there's tons of trash data. So you see those things. But we have already experienced where, okay, we can get all that data out and now you have to work it through a pretty deep hygiene process. But that's like one step of like 20. Because if you think about it, data also decays. If you think about it, data also decays. So when you look at like we'll take email addresses just as a simple analogy. So email addresses degrade at 2.5% per month, so that's over 30% a year. So if you have not done anything to your CRM or DMS data in the last year, guaranteed 30 plus percent of all those email addresses are bad.
Speaker 2:So this now let's put this into a practical sense. If you're a dealer and it's what is it the 15th right we're recording? Today I'm halfway through the month and I'm like, oh crap, I need to sell more cars. So you go down to the BDC or your internet team or whoever and say, hey, we need to send out a email blast and look, we have some new car incentives and we have some used cars, let's blast them, bro. And you go in and let's say the guy punches in and goes okay, we have 10,000 people who bought new Toyotas. And you're like, yeah, there's got to be buyers in there. So you hit write this amazing email Come on in Best dealership to buy at Been in business for 100 years, we are easy to deal with, we'll get you in and out in a quick amount of time, painless purchase process.
Speaker 2:We send out the email with the coupon and then we wait and you're like where are all the people? What's going on? Why is our phone not ringing? And, interesting enough, if your delivery rate, meaning your bounce rate, as soon as you send that email campaign out, as soon as you hit a 10% bounce rate which you're absolutely going to do in every CRM because your data is not hygiene in every DMS, because your data is not hygiene, once you hit that, the ISP, your internet service provider, throttles all the way down, delivering more. So your 10,000 might realistically be 2,500 or 3,000. And then in your mind you sent 10,000 and you're like dude, why did I get a half a percent of people even open it? The problem is all our statistics fail when you don't have good data and then we're making decisions off of a bad data set or I think we talked about before show like or you start weaponizing the data Meaning inside groups. We'll see data being used to secure budget, data being used in a certain way to validate current market or thinking or validating my job.
Speaker 2:So it's interesting. I talked to some groups. I talked to Matt Lasher at Westar and I love Matt. Matt is a great marketer and we were talking and we were like, yeah, man, you see a lot of managers inside stores They've weaponized the data to build a mode of protection for their job.
Speaker 2:And you start thinking data is ones and zeros. Right, it should be exact, not curated, to paint my picture. And look, we can laugh at this, because when you think about weaponizing data, I always go to internet leads and you, I believe, have visited lots of stores, like I've visited during my lifetime in auto. And how many stores have you gone into and have you ever seen the anomaly where it's like we have a 62% closing ratio of internet leads? You've seen this right? Yeah, I wish we had like some of the internet guys on too, because everyone or there are 34%. Let's be more realistic.
Speaker 2:The 60% I've seen it, but I've seen it as high as 70%.
Speaker 2:And I'm like there is no way. But let's say they're showing 27% close rate and I'm like, okay, so let's dig into this. So first let's look at all these leads and I was like, well, what about these leads? Oh, if they don't respond to the first email, we don't count them. Or oh, they didn't have a phone number, we don't count those. Or oh, no, that comes from what is it? The classified ad service? We don't count those leads because they're. Or we don't count General Motors hand-raiser leads because nobody ever that's from a state fair yeah they're from a state fair, so we can't count those.
Speaker 2:And then you have this curation and now you've weaponized the data to create a story that you want to tell and believe. You've created a bias and now you're convincing other members of the organization that that's correct. And now you're convincing other members of the organization that that's correct. You are already on. You are going down the slippity slide with super dawn grease at the highest rate of speed into a self-fulfilling prophecy. And I don't know.
Speaker 2:I look at this and, man, I'm like no, no, no, no Data. It has to be all data. You just want it normalized. Like a good example, we just did a group. They are like we have a data lake. And both Martin and I were like because a lot of data lakes, we go to man, the data going into them is not complete, it's all over the place. We find a lot of data going into them is only designed to write reports on the other side. So you get a lot of just strange data. But they're like we got 1.6 million customers, guys, and I'm like, okay, sounds good. So once we go through this process 600,000 real people they're like what. They're like no, we have more.
Speaker 2:I was like you don't have more Meaning. We have matched this to people who have credit files. These are living people and, by the way, 2% of the people in your database were dead, dead. You're still marketing their email addresses. You're still marketing direct mail to these people. They're dead and they're like no, I was like look, deceased, deceased, deceased, deceased. I gave them the deceased list and said here you go. And I said look, this happens all the time because life keeps moving all the time and there's no like.
Speaker 2:That's why I always feel like I just talked to a dealer the other day. He's like I just want hygiene data. I want to take it out of my DMS, clean it, put it back. Will you do that? I said no. And I said no, I believe, for a very valid reason. It's because I'm setting you up to fail, because the second you put it back in, it still starts to deteriorate and yeah, you're good for a day or a month or whatever it is, but you're going to get erosion and it's inefficient and it's expensive to just clean data once and try to repurpose.
Speaker 2:And I just wrote a post about this I think it's today's actually where I said you know, it's like washing clothes in dirty water. Yeah, you know it's like, but you're like. It's water, it's got soap in it, but it's dirty and you're letting more dirt just accumulate inside of it. And this is why our entire concept is like look, you got to get the data in something else that you own as a dealer that's protected and it's in what we call like an evergreen state. It's like hey, every month, look at email addresses, because that's critical Every, let's say, 90 days. Let's look at propensity, because it changes on a wave. Let's look at do they still own that car? Did these leads buy from competitors? Stop calling them and emailing them and understand why they bought from competitors. Maybe compare it to all your pencil versus the outcomes of the transactions, because you can see transactions outside your organization and you can see the financial structure. So I think there's a lot that dealers can do with data.
Speaker 2:They just but look, I think it still comes down to block and tackle. Get all your data in one place. It's painful, it's worth it and it's the only way you leverage AI.
Speaker 1:It's very interesting. I mean you shared several examples. I mean my gears were spinning. It's interesting that the digital age brought us, obviously, an explosion of data and from all types of areas that we never really even thought about until we had it, from digital marketing data.
Speaker 1:It's the same thing. Like you're talking about people wanting to weaponize or basically tell a story, set up a narrative, if they will. That it's a bias to basically get people to either agree or to believe in something or to buy something, and you know there used to be digital marketing firms that would. They would basically ask the dealer like well, what's most important to you, like the amount of traffic that we can send to your website, and if that's the most important thing, and the dealer's like, yeah, like we need website traffic really really bad. Okay, great, that's going to be one of the things that they make as a note, that they're going to report to you every single month, and they're going to chase that metric, not the conversion of that traffic, right, not the stuff that really matters downstream, from beyond. They visited the site and it goes like that, on and on and on. Through paid search campaigns hey, what's the click-through rate? Is that the most important metric to you, the click-through rate or the cost per click, or how many impressions you served.
Speaker 1:As soon as you get people to fixate on a thing that some person who's been either self-appointed or the industry thinks is the leader on something, this will relate to you like think of it as the surfer. Like you're just waiting. Like you, you read the sets until you know the one you want to ride like, and there are people that will do that and they'll tell those stories, unfortunately, leading dealers to to ask of you what one of these examples did, where it's like, well, we come in, rip it out, clean it all up and put it back in. It's like so, do you just clean your house like fucking once a year. That's all you do. Like the point of all of it now and I have, and I know you do too we kid about stuff. We have a lot of fun with this, but we're only doing all these things because we care about the industry, including dealers and really everything that's going on here.
Speaker 1:You're not really helping a dealer when, after all of these years of just technology, data stacked upon more data and so many that don't even understand how to utilize it, we've ushered out so many buzzwords, the big data and attribution modeling and all these things and yet we have not moved forward. It's kind of like you know people talk about like hey, let's do more cancer research, like clearly we're making a lot of money off of cancer research and treatment, but we have more cancer than ever before, which is true. We also still have tons of dealers maybe more than ever before, that have all of these different points of data. And now we have to think about, hey, what's its relationship to AI? And some of them still want a quick fix or something that they believe from something they learned from somebody that didn't know what they were talking about two years ago or seven years ago, or maybe right now today.
Speaker 1:And I'm glad that you're kind of the.
Speaker 1:You're not a bullhorn, you're much more, I think, kinder than that, so I won't say it that way, but you certainly are that person kind of on the street corner saying, hey, these are all really important things to be considering as you think about your data and what you want to do going forward.
Speaker 1:And you've made points even on previous episodes where, if you think about doing it right now and how much that begins to give you a stake in the ground of differentiation between all of your competitors, whether they're same brand, different brand, same category, different category but you're getting some distance between them and doing it right. I just I wanted to, I guess, say all that to also then say for the dealers that tune into this and are listening to this, it is critically important that, as you start to make some of these changes, that they're ones that are yeah, they're maybe not going to be the easiest change, but so worth it to your point a minute ago. You're not going to regret making changes like this, but it might take a little bit more elbow grease than you realize. Just know that it's worth it. And, by the way, if you don't do these things right now, it's going to be far more painful a year, five years from today.
Speaker 2:So I wanted to get the advantage. You'll never get the advantage of AI and like it's just gone, of ai and like it's just gone. And if you're a dealer, depending on your vendor to give you ai or you know, to produce some ai functionality for your dealership wrong meth, wrong thinking it. Just clearly the dealer needs to control his data and start training his AI child to understand his business. It's the only way we move from reactive to proactive and prescriptive.
Speaker 2:And there's a very deep thing inside here where I think every dealer thinks in 30 days in current cycles, right, 30 days, 30 days, 30 days, right, I'm halfway through the month, right, that's all I'm not thinking oh my God, I got two and a half months left before the year, right, no, no, no. I'm thinking like boom, boom, boom or whatever. A month and a half until I'm done.
Speaker 1:Black Friday is almost here.
Speaker 2:It's coming, so cyber sale coming so cyber sale, and I I think I'd look at this and I'm like, okay, ai is very good at looking at data, large chunks of data. The more data it looks at, the more accurate it is. Hence, what? What is an llm? An llm is predicting the next, the next word that you are going to type, and that's what it's doing. It's amazingly good at that, and when you feed it enough training data, it's very good at predicting what will happen next.
Speaker 2:Good, at predicting what will happen next. But if you don't feed it the data because you don't have the data, because all your data is held in these vendors, you never benefit. You never get the Glenn Gary AI. You do not get. It stays on the shelf. You never get the AI. You will get the AI only provided by the vendor who's controlling it.
Speaker 2:It's not built on your store, meaning it's training on all stores which is far more generic than the individuality of your store and how your store behaves, the type of people who buy from you, the PMA you're in, the types of cars, the brands you sell. Do you sell more trucks than cars in use? All those small little things. Ai will explode and help you start to see absolutely within the next six to 12 months we will peer slightly around a corner. I will start to see 30, 60 days of what my sales will be, what I'll gross, how much service work I'll get. Ai will absolutely predict, within a very small finite rule of error, the accuracy of that, and then it'll start doing larger prediction, farther out, and then it'll start doing larger prediction farther out Today.
Speaker 1:Think about it we're going to end on the 31st right of November, and now I'm going to look backwards and I'm going to go what did I do? And now I'm trying to make decisions to correct my mistakes, or I have too much of this inventory. I need to reprice this. I'm looking backwards when AI is the only thing that's going to give you a glimpse into your future and I think it's such a critical will help dealerships with the gathering of the right type of high-quality, relevant data. Is that too early or are there some things that are starting to kind of pop up?
Speaker 2:no-transcript. And this is where I am just and look, I'm just not the lover of automotive CDPs and I say it in the nicest way. I just think that they're marketing companies and that's good. Listen, dealers love marketing companies and dealers want one throat to choke. And somehow in our industry, data and marketing have come together in this weird interconnected way, because if we go to any other industry, they're different, and if we look outside auto, it's church and it's state In auto, somehow the church and state are the same thing. And the problem with this to me? You choose a current automotive CDP, take your pick, let's say. They have some AI stuff, they have all your data, they're maximizing your marketing.
Speaker 2:They have a cool name. They have a cool name. 11 months from now, another CDP has some other widget marketing tool. This guy, your current vendor, doesn't, so you leave. Well, guess what happens? All that data is gone too. Even if you get a copy of it now, it's in a format that the new guy probably is going to take forever to ingest if he can ingest it, if he tags things that way, and most of the time they'll just won't do it so for a dealer.
Speaker 2:This is why a dealer has to be at the center of his data. And listen, I'm going to die on the mound telling dealers in this crusade that you have to own and be at the center of your data and not a vendor marketing company. You want to be the middleware between your data and the marketing company and you don't want it to go from DMS right to the vendor. And then you want to become the center. And then you want to become the center. And then you want to use AI to train your data on the nuances of your store, the behavior, even to a granular level. Think about it it's not very hard to train an agent AI agent to go. Wow, sean got 47 F-150 leads this year he closed three. Todd got 51 F-150 leads Closed 32. Where should F-150 leads go? Not to Sean.
Speaker 1:No, I'm getting fired.
Speaker 2:Well, you may be better at telling other things. The real reality is look, I think AI helps us identify strengths and weaknesses really fast in the data and then we can make better decisions. So, look, I may drive an F-150. I have a deep affinity. You drive an electric car and you're trying to sell F-150s. You don't speak truck dude. You don't speak redneck, even though you're from Texas. So you might speak a little redneck. But the real reality is like, yep, just poquito.
Speaker 2:So I look at it as like, hey, we have affinities, we all do, we have passions, data just, and AI instantly goes. I know who is doing what and how effectively. So I see AI right now in our business right being used for what I call the low hanging fruit. Right, it's like marketing spend or marketing optimization. Right Is where we see the most kind of usage right now. But the deeper thing is if dealers could get all their data in there, including employee data right, who works when? How efficient are they? All this data can be pumped into the dealer's data warehouse. Right, I'm not saying data warehouse, that's structured data. It can take in all the unstructured and structured data, but it formulates it into a way that AI can leverage. At that moment, when you have this running, you start thinking about operational efficiency then.
Speaker 2:So I see a three-step process just in AI in general. Ai is going to live as a co-pilot for us. It's going to work alongside humans, it's going to help humans do what they're already trying to do and it's going to make them effective. Think of AI as an like a was it the echo skeleton? Like exoskeleton, like to be able to go pick up a thousand pounds, right, you've seen the kind of military apparatus like alien movie where she's like lifting up the heavy thing, it's got the pinchers and all that Right. So co-pilot. Then you're going to see the beginning of autonomous agents.
Speaker 2:So when we do things like sell a car, selling a car is a series of small steps. It's not one step, it's a series of micro steps. Agents are going to start replacing pieces of those steps. It won't replace all of the process. It's going to create functionality across that process and do pieces of it. It won't replace all of the process. It's going to create functionality across that process and do pieces of it.
Speaker 2:And it's going to then shrink the transaction time because autonomous agents are doing some and humans are still doing others. And then, at a period in the future, the person just kind of evaporates from that piece or that process Because the AI agent will understand it enough. It does now not need the human to be the bridge between certain functionality. This is a very R and it works with us because we're linear thinkers, not exponential thinkers, and I think it's something that we can get behind, because, thinking for a dealership, a dealership's annual payroll is probably about 5.5 million on average. If I can make your store 10% more efficient by using AI, just that, that's half a million 600K to your bottom line.
Speaker 1:In some cases it might end up being more than a 10% improvement, I mean.
Speaker 2:I say this at the most basic right, the real reality. I would say I could see a realistic 60 plus percent efficiency increase.
Speaker 1:Digital age brought us lots of great things. Lots of great things but also more complexity into an industry in a short period of time than it's ever seen, ever Right, exploding with a trajectory that's so fast that I get it. That's why I have a lot of sympathy for, you know, for the folks that are still in retail, that have been in it for a long, long time, because it's like gosh, there's so much to handle. But I've always been one of those people. I know you think this way too is if you could just actually build processes even if they need to be nimble processes that you tweak as you need right, maybe it needs to be tweaked every month, maybe it's twice a year, maybe it's only once a year. But for all the dealers that are still just kind of freewheeling the way they've always freewheeled and the process is, it's okay. But if you have a superstar salesperson that doesn't want to work the process, doesn't want to use the CRM the way you want them to, but they sell more cars than anybody, so they can they anybody, so that tail can wag the dog. You're not helping anybody ultimately. And now we're actually not imagining we are in the era where dealerships will be able to institute, build flexible processes that the humans won't buck the system they won't cherry pick their way into like, oh, I'll just take care of myself but not really think about the whole dealership. It's what you were just saying and that is going to have. I can't wait till we get real numbers on how much that positively changes. I think there's a lot of upside there in retail. I think there's more things actually to be really excited about in the future than I hear a lot of people talk about. It's why I love doing these episodes with you, because you're talking about the kinds of things that I think dealers should be like. How do I learn more about what Todd Smith's doing? What's Corey Aya doing? How are you guys approaching the market? Because it is so practical. It's not elementary by any stretch, but it is not like you're talking about rocket science in a way that people won't be able to approach it at all. It is completely doable and the timing could be better.
Speaker 1:Need for evergreen data keeping information current. You mentioned some things, too, that I want to ask you about, because there is an impact of when data breaks down and degrades. You mentioned at the top of the episodes a little bit. You were talking about email decay. That certainly happens in B2C. That happens in B2B marketing as well. It might even be bigger percentages of how much the breakdown happens on a monthly basis. But I don't think people talk about that quite a bit as much as they should. And so question for you you mentioned that email is degraded by I think you said about two and a half percent per month up to 3%. Can you give an explanation a little bit of the impact of that on dealers, of why the data update should be happening regularly?
Speaker 2:Yeah, to me it's you want as close. I mean monthly is kind of the cadence for some data, like email addresses are monthly. I will look at like addresses, home addresses, quarterly. Checking on the do they still own the vehicle quarterly or at least twice a year? Checking on the do they still own the vehicle quarterly or at least twice a year? Nothing worse than, like I'm still getting emails and direct mail pieces from Mercedes, which I sold my last CLS in 2016. And I'm like what? Why do they still market like I own it? And it's bizarre to me that nobody's cleaning their data. So people move right. So those are things you like to check quarterly and there's a host of little things in there.
Speaker 2:Phone numbers stay pretty consistent but still need to be checked, and to me, every time we come in contact with our customers, it's a good time things get audited there as well, but not as effectively. Because think about it this way if you come in, you buy a car, you come in for service, right, I don't ask to update your home address and sometimes I'll put you in as a whole separate person Todd Lear Smith. Todd L Smith. Todd Smith. And it's not uncommon that there's multiple right Because they don't have a identity graph verification process up front in any of these systems. So that's why it's super important for when the dealer has and owns this data, that it's constantly like hmm, 96%, sure that's the same, todd, and you can build AI and modeling to easily do that and go.
Speaker 2:Okay, identity graph. This is this person, or? Ah, this is interesting. This person was connected to this car, but this is a lady who lives at that same address Spouse or kid. Car like it's household now, car like it's household now. So AI is really good at spotting, like random forest, like making those pattern matches and correlations. Humans are just not nearly as fast or efficient at doing that.
Speaker 2:So, I think for dealers you know again harping on the value proposition get the data in one spot, get your arms around owning it, get everything streamed in and organized and put it into a structure and format that you can begin to leverage AI to learn about your customers. I think that the beginning of equity mining sounds really smart, and what I mean by that is and all the tools today still do this right. It's like oh, this loan is coming, like it's going to be paid off in three months. Let's call that guy and get him to buy another car. But what happens if that person takes five-year notes and, on average, buys cars every seven years?
Speaker 2:Now you're going to jam him with a bunch of marketing material that he's going to ignore. You're going to train him to ignore your marketing because you're not putting your head up to look around and go oh, I've now made a correlation. Yeah, he's going to pay off this thing, but we should start marketing him in year six, with nine or six years, nine months from now, with nine or six years, nine months from now, because that's when we know he's going to cycle in, most likely based on the last three transactions. Or maybe there's a little chaos in there, but that's why AI exists. It's good to say, oh, here's a safety ground. Go in this section, because a lot of times, dealers, right now, I think, man, they just throw a wave my surfer analogy they throw the wave of marketing spend and say hit them all.
Speaker 2:The buyer's in there somewhere and they're right. But the problem with doing that, you may get some of those buyers. You're also training a vast majority 97% of your audience to start to ignore you or put you in spam, put you on red, you're done. And this is where you're overspending. Currently in every dealer I've seen is absolutely overspending. Currently in every dealer I've seen is absolutely overspending their marketing and it's weird because it's like Linus or whoever had the little blanket in Snoopy, I think it was Linus right or was Linus the dirty one?
Speaker 1:I forget. No, that was Pigpen.
Speaker 2:Pigpen was the dirty one. Yeah, linus had the blanket.
Speaker 1:Linus had the blanket. Yeah, okay.
Speaker 2:Dude, it's been a long. My kid's 14 now I'm just going back. So with Linus, with that safety blanket, dealers are constantly like no, no, no, we got to market to everybody. I'm like, but all these people aren't in market. Yeah, but they might buy, you don't know. I was like, actually, with data we kind of do know. Now we can see based on behavior, because this device is one of the most amazing things and just because they visited your website, it doesn't make them a buyer. Just because they visited carscom does not make them a buyer. There's tons, hundreds, hundreds and hundreds of signals that are all being weighed and measured to then go okay, look, ai is like this guy. And to bring back to the previous conversation, when I was talking about that, we got down to the 600,000 records. Out of the 600,000, fascinating enough we said 9,874 were in market. Cool, okay. Now as a good company, we're like hey, these are your Glenn Gary, like these guys are in market right now. Our modeling says these guys are in market. So the auto group, like any good auto group, we hand them this and the marketing director goes on vacation. So they do nothing, which now, in all transparency.
Speaker 2:I was very upset, but I also then reconditioned my brain to say, wow, this is a very good control group for us to look at. So we said these 9,000 are market. They don't do anything with it except normal business Marketing guys on vacation, you know, drinking pina coladas and doing all this stuff. He comes back two weeks later still doesn't do nothing. At week three, I go back in and I cause our, we take all the data and we put it on an LLM so I can just query Okay, I was like, okay, based on these, how many of these customers bought? And it was crazy, sean, in three weeks, 2,110 of the 9,874 bought cars from the group. Wow, wow, yeah, so whatever, it's 21 or 20.5% bought. And I was like, ah, I'm like I smell a good meal being cooked here and I'm about to run it again, which I don't have the data, otherwise I give it to you where I think, at the six-week mark, my belief is they're going to hit about 30% to 32% will buy from them. I think they're going to have a defection rate of 60-some percent defects and then we're going to track that because we have the the data.
Speaker 2:They bought from a competitor. Did they buy the same brand or did they swap brand. And did they swap brand that you also had as the group where you, we knew that they most likely were going to buy from this brand and you should have marketed to them, but again, you were on vacation, chose not to. They, you know, bought a competitive cross-brand. So I think, again, that data intelligence shows you can determine if people are in market. That's validated. I think that AI is only going to make that tighter and more and more accurate through just a short period of time where we'll know, you know, sean is going to buy, like in weeks, like we will see it in the data, like the matrix, we're going to be like him her, him, her buying Him her, him her.
Speaker 1:Fine you mentioned it a few minutes back the predictive capabilities. We have been in the descriptive world for so long of like we're going to describe what could happen, but it's mostly somebody's hypothesis, you know like. So let me think this it's um, that's describing. It is one thing. Prescribing even just the story that you're just sharing right now is that is emerging literally. I mean you're watching it unfold in real time. That's another reason to be like super excited.
Speaker 1:All of the things like not everything, but a lot of the dreams that dealers have had for decades of like. I wish I could know this. I mean, you're starting to talk about things where the day is going to come where I regarding yeah, I know I've got my marketing work. I have 50% of my marketing works. I just don't know which 50% works. I'm one of those people that believes my marketing works. I just didn't know which 50% works. I'm one of those people that believes all marketing works. So, even if it's not selling you a car or generating a lead or making your brand bigger or whatever, it's still telling you okay. Well, nobody likes that. But what you're sharing is the future of how those things you are going to be able to get even better at predicting these things and not just well, out of this large number, we know we're going to probably end up with this number of real opportunities and from that number, once you market to them hopefully before the pina colada vacation, but once you market to them we know that you're going to get a take rate in this range and you're going to have a defection rate in this range.
Speaker 1:Now you're talking about things that really matter to dealers. Obviously, how many vehicles are we going to sell? But you're also giving them some insights into things that are really important in terms of well, are we going to have retention loss for whatever? Are some of these people going to be moving to places where we won't have them as service business and we wouldn't consider them to be somebody who's going to buy six cars from us. There's just so much of that. That's exciting. I do want to talk about a couple of things before we get too close to the end. One around data enrichment. There's obviously all this dealership data. I'm curious to know if you have thoughts on additional data sources that enrich and enhance the understanding of the customer. Give me some of your thoughts on that. Are there some things that you think are kind of can't miss for dealers.
Speaker 2:Yeah, look, I think there's again. There's tons of auxiliary I call it data that surrounds the dealer, data that can do everything, from taking all your customers and putting them in cohorts. So, hey, I know these customers are like-minded, similar or these customers share like hundreds of attributes. So you know, okay, they're going to most likely buy these cars. They have this level of affordability. All that type of data is available today Marketplace data knowing, hey, these are the you know what car's average selling for, coupled that with demand. So being able to see and cross those things in data models. So I look at a few things like, okay, we're entering a time where we could take the dealer's data. We understand the inventory the dealer has. We understand, with enhanced auxiliary data, hey, these customers likely will buy these cars. So this type of information can help match cars I have to customers in market, cars I probably need.
Speaker 2:I should be buying to customers in market more effectively. I should be buying to customers in market more effectively. Because right now, how do dealers buy cars? I get trade-ins, I take anything I can get, but they are sourcing cars. But how do you know what cars to source? How do you know what you should pay for those cars. I think intelligence is going to open up new windows of opportunity for dealers to connect cars to people more succinctly. Right now, dealers buy cars and then they go hunting for the people and they spend a lot of money hunting for people Average $700 based on NADA.
Speaker 2:So what happens if you can create better alignment to hey, I need these cars or hey, I have some of these cars already with customers I have, I have new customers looking. Let's pull some of these customers who are in market and do matching. This is where data at scale, the big data you call dude. It's so good at that and we are already having lots of fun and interesting data correlations, I would say in our own work at Core AI and like even to like I'm going to break it down to like super simple ones, like I think they're simple, but I have found many groups struggle with which is they buy at one store, let's say a Honda store, and they service at a sister Honda store.
Speaker 2:Now, for the typical dealer, the data in the first store where they bought they are showing as a service defector. Right, they've never serviced there. Yet the other store they're as a service quality customer, but they've never bought there. So you have one store marketing them cheap service specials where they're now taking and using them in the other store. So they're eroding store two's customer pay RO value because this store thinks they're a service defector. We see this, it's crazy. And then you're like oh, but we can see it in their data and they're like stop sending cheap coupons to this customer. And what we realize is oh, but we can see it in their data and they're like stop sending cheap coupons to this customer. And what we realize is oh, this customer lives close to this store, but they work close to that store.
Speaker 1:Interesting. Yeah, all of that. So just a couple of things. I just I mean, the allocation consideration is huge. I won't name any names because it might make some dealers that sell these brands, you know, feel bad, but they'll know as soon as I say if you happen to be one of the brands that has like two or 300 day market day supply going on on a bunch of your models.
Speaker 1:Um, the manufacturers, instead of saying something that they would construe as me being, you know, harsh towards them, I'll just say it this way they should be opening their arms really, really wide, and I'm sure that many of them are. I hope they are in looking for how? Um, quality data, the right data, the things that we're talking about on this episode then the proper usage of these AI applications being able to help them align things like yeah, we're not going to be sending you another 50 of this particular product that is not moving and, by the way, for whatever reason, you and your little town and your little city have a really hard time moving those. So, instead of giving you more that are all going to have birthdays pretty soon let's use this data to actually plug in in a way that we haven't seen before, which is almost surprising that it hasn't gotten significantly better.
Speaker 2:Yeah, I think they're trying Look.
Speaker 2:I feel that OEMs and again, I don't probably talk to enough OEMs, as I should that I think ambition would always be put the right vehicles in the market. That is going to move those vehicles. And it's hard from the sense that an OEM is a national and they're chugging out cars at an alarming rate. And then what's happening is and think of it this way and a good way to look at this if a manufacturer makes cars and puts it in the markets and then has to subvert that by incentives and rebates, it has failed the marketplace process, because now I'm having to pay to sell those cars in those markets and I think AI will absolutely right that ship. And look, covid did a good job of it, right, because manufacturer got shrunk down and there were no incentives for a long period of time and supply demand outstripped supply dramatically and it allowed prices to rise and dealers got fat. Everyone was happy. Now we're collapsing back into reality.
Speaker 2:Manufacturers have redone what they said they wouldn't. We're going to produce more cars because of, maybe, union agreements and contracts and all these things that have to happen. But there is definitely there is algorithms to be built that will be far more accurate saying this marketplace needs these cars, this marketplace needs those models. And now we're still wrong. We're still going to do incentives, but my thinking is an OEM should be looking at this as their 12, 18, 24-month plan.
Speaker 2:How do I keep putting cards in the market while reducing my serventive spend? How do I reduce incentive rebate money? Because that should be a whole model. If I was hired by an OEM, that is literally all we would be looking at is okay, let's look at these national markets for this product and saying how many buyers are in market in all the big metro markets, all the markets, and now inventory should be moving to those markets at a pace that demand is there and I think with AI you can do that and there's nothing else you could do to get you there. And I feel like everyone has tried to do the data, but unfortunately they're looking in the past to make predictive choices or predictive decisions and it hasn't worked.
Speaker 2:Hence, rebates and incentives are creeping back up because supply is outpacing demand in markets and once you move a car to a market, you can't un-move it to the market. Yeah, it's too costly. So you're better to use data to start to build predictive modelings for localized demand. The problem is you know you need to do that at scale and again, this is what data and AI is literally built for. The problem is you have archaic data systems. They have to give that data to companies like us that we can then put it on a graph, vectorize it, turn it into AI usable and then I think dealers will be happy, oems will be happy, customers will be happy, oems will be happy, customers will be happy. You get the trifecta right of performance across all the parties. Yeah, that's where we have to go.
Speaker 1:I love that. If I were an OEM, you would never do this. Well, if you were smart, you wouldn't. Well, I guess it would depend on how much money they want to give you. But if I were an OEM, I would be trying to get you and your new company to be exclusive with me. I would want a corner. I'd want you know. But you think about it, right now it's wide open for dealers, Like they could jump in on this, which is good.
Speaker 2:But yeah, Look for some sense. But the person who's buying a Land Rover isn't buying a Honda Accord, Like in most cases. Right, it's just that's. There's very different buyers for different products, you know, you're affluent.
Speaker 2:Subaru buyer isn't buying a Silverado a time like this. But I think building unique algorithms and leveraging AI and building it for brands that they have you know it's learning for them is incredibly valuable and the same for a dealer Like look, two competing dealers. I'll give you another very interesting piece in data. And again, I feel like every month when we do these I come back with some learnings. So we looked across this other group and propensity to buy when a consumer would hit like 800, 850, 900, dude, they're buying. Like we know now Statistically they're in market.
Speaker 2:So you take Honda, toyota, chevy, ford a lot of behavior in there. And then I saw this data over here and it was Land Rover, and Land Rover it was like 550. And I'm like what, what, why is the propensity models broken? And I go back to our CTO and all the guys, our engineers. I'm like dude, what's up? Like it's 850, all the brands. They're like no people who buy this car. Look at the cohorts. They are Super high income paying cash. They don't go through the same process of all those buy signals. Some of these people are walking in and be like it's friday. Hey, honey, here, here's a new land rover for you. I just bought it. Oh, thank you, honey. Like there was no, I didn't go look at 14 other uh sdvs and compare them. I literally just went in and said, yeah, I'm buying it. And once you start understanding that you have to lean that and put the weights into those models and so I think for us.
Speaker 2:It's been. It's been fascinating to see um data paint the picture in front of you and then I like, fills it in like. So for us, I feel like data gives you the outline sketch. Ai adds all the vibrance, the colors, the shading, the depth and the detailed intelligence that we just were not able to do by creating pivot tables, cross table reports, like that just wasn't viable.
Speaker 1:Yeah, I, I, you do come back with new learnings on every single episode and I'm glad that I get to be the host of your podcast because I get to learn as well. And, and I mean, our little business plays with lots of different AI, you know, in kind of different realm, but it there is so much crossover that you know it's it's it's it's nice to hear your voice of leadership in this area Because, again, I think every episode you're dropping things that I think not rocket science, also not elementary school. It's all doable, it's all important. And the last thing I just want to ask you to comment on before we close this episode down is really, why do you think it's important and I've asked you this before, but why is it important that now is the time for dealers to start building these data strategies and these AI strategies, versus later?
Speaker 2:Yeah, okay, so we can look at this and hopefully my answer is very clear that there are many dealers who look at AI as if it's a SaaS product and they are sitting waiting thinking that when the AI is perfected, I will jump in. I will jump in and when you start at a wrong point, everything past that becomes a mistake, and I'll do it to flying. So when you're flying, and if you're off one degree, it's 92 feet per mile that you're flying. So if you're flying 100 miles, you're 92 feet times 100, it's a lot. You're overshooting runways. So little mistake over long time, incorrectable disaster. And when you look at AI as it's a SaaS tool, you're already in the mistake. So I look at this as a AI is a river. It's a constantly moving thing and if you don't jump in, you wait your competitor who jumps in today. His AI is gaining data, training data. It's learning, it's improving. You're going to jump in and you're going to have to cover the same length of river he already covered. The problem is he has already covered it, which means he's already operating at more efficiency. You will not catch him in the river because the river flows at the same speed and even if you put paddles in, you're still not going to catch him, because AI is one of these things that it takes time to learn, to train, to improve, and your experience with AI will be different than a competitor's in the sense of training, volume of learnings and how much data you can put into your cloud. And I think there is no other time than right now to be involved with data and building that data strategy and beginning to leverage the power of AI. But it begins by organizing data, getting all your data streaming in and put into a format that AI can be used against it, and I think the idea of looking at it again as it's a SaaS, it's a product you already made the mistake and it's the same idea.
Speaker 2:We thought about the internet. We thought about the internet as a thing, and I still say this, sean, all the time I need to go get on the internet, and my kid looks at me like I'm a freaking idiot. He's like get on the internet. What are you talking about? It's like dude. It's on the tv, it's. It's right here in my hand. It's on this ipad, dude, it's literally you. You're always on it. Why, why? Where do you get on it? It's like. It's like a subway station in my mind, because that's how we learn and yeah, and we're now taking that same mentality and saying, oh the ai.
Speaker 2:It's like AI is going to be pervasive through everything we touch, everything. There's nothing that won't have AI components and processes going on inside of it. So waiting is crazy and I'm going to leave you with this. This is you'll love this and let me grab it, because I think it's pretty uh funny for our conversation here.
Speaker 1:So sorry, disappearing into the uh might even get a little show and tell.
Speaker 2:Todd is rummaging through his treasure chest my cabinet there we go I wrote this manual and back here in 1997 hold on, you're gonna love this dude arrow baby. Look at todd with more hair and a handsome todd with gochi wow I wrote this in 1997 and it was automotive retailing.
Speaker 2:Look on the internet yes and it's a perception and dude you'll laugh. Dude, this was all about. Uh, let me get the. So I had like new customer. Learn how to sell a new generation of customers that use the internet as a tool against the automotive dealer website. Online services. How to use auto responders, email newsletters uh, competition. How to drive traffic? That means get on Ask Jeeves and Yahoo.
Speaker 1:There was no Google, no Google at that point.
Speaker 2:This was awesome, dude. So I wrote this whole giant manual. Man, I sold a ton of them, but that was in 97. I look, though, and you learn a lesson.
Speaker 1:No Google, no Facebook.
Speaker 2:Dude, there was no Google, there was no Facebook, there was nothing, there was no.
Speaker 1:MySpaceTube. There was none of this.
Speaker 2:So back in 97, it was Ask Jeeves, it was AltaVista, it was Yahoo, which you had to pay $200 to get in the directory.
Speaker 1:Like a hot bot. Yeah was Yahoo, which you had to pay $200 to get into the directory Like a hot bot.
Speaker 2:Oh man, there was Hotmail. Right, I think Hotmail had come out. So I look at it and I think we have to change our frame. And this is the most important thing we have to change our frame and the realization of what AI is and how it's going to impact us. And it begins with data, and dealers have to get control of their data. And I will leave you with one extra ounce of this, which is I don't want us to see us repeat what happened from 97 through the 2000s, where we spun up a series of siloed applications to do things on the internet. We have an AI email tool, we have an AI phone process, and the problem is they're pulling from all different data reservoirs. They're all being trained different, which means they're going to deliver completely different experiences for the consumer, and we're doing nothing more than confusing. The end. Consumer versus dealer owns all the data in one thing Connect these tools to a service, the data that the dealer Excellent, excellent, great place to park this episode.
Speaker 1:I will say to the audience our industry just has been evolving faster in the past 20 years since the dawn of the internet, but the last couple of years we've hit another gear and these are the things that we talk about right now on Core Conversations. It's really critical, it really is critical that you embrace what you probably consider to be really advanced technology If you want to stay ahead of the curve, if you want to make sure that you're in the river flowing now versus trying to catch up to the people that you might compete with. So for the listeners, I always say this make sure that you jump onto LinkedIn. If you're not already following and connected with Todd, make sure you do that, because he supplements what happens on these podcast episodes with a lot of material, articles and pieces that he posts there, and I don't think there's anybody.
Speaker 1:Yeah sure, I'm biased and I'm a little selfish in this. There is no one else. I'm also somebody who doesn't lie into the industry for my own benefit. There isn't anyone else who's going to be the candid kind of truth teller around this that you're going to find than Todd. So make sure you go get connected with him on LinkedIn. Go check out coreaicom and find out what his company he really never even selfishly just dishes about what he's doing with this cool company on the podcast but probably something you want to check out, um, so you can get even more information. Thanks for everybody for joining us. We always appreciate it and we'll see you again right here on the core conversations podcast.