Interview with Dave Anderson, CEO & Founder of EvenFlow AI

Bringing some muscle to the appointment debate 0:00

Steve Shaw: Welcome, welcome. What's going on, Dave Rogers? How you doing today, sir?

Dave Rogers: Doing fantastic there, Steve Shaw. First day of baseball season.

Steve Shaw: I bet you are. Baseball season, Cardinals playing today, and I'm sitting here recording a video missing my Cardinals. Definitely a little bummed out about that, but we'll get to see a replay at least. Hey, I see you brought a friend today. A couple of days ago we were talking about appointments or no appointments, and I was kind of in the no-appointment camp. I'm kind of like, bring it in right now. And you were a little more in the "let's make an appointment, but it's OK if you don't" camp. I see you brought a buddy to give you a little ammunition about making appointments.

Dave Rogers: Yeah, I had to bring some muscle into the conversation, with you running us ragged with no appointments and everything else. So we have Dave Anderson from EvenFlow AI, and he's put a little bit of science into the appointment world. Welcome aboard, David. Nice to have you on.

Steve Shaw: Before we get started, I've got to ask you a question. Do you specifically bring on people who are bald? Does that help support the argument?

Dave Rogers: It's pretty much one of the first or second criteria. They have to have no hair. We have to outnumber you.

Steve Shaw: OK, good. I just wanted to make sure we were clear on that. Sorry, Dave, I didn't mean to interrupt your start there. Tell us about you. Tell us about EvenFlow. Tell us about appointments. I'll let you guys drive this conversation.

David Anderson: Sure. I'll just say I'm biased, because I built a company that helps service departments schedule more intelligent appointments. So I'll put it out there: I'm for appointments.

Dave Rogers: We actually met briefly at NADA and had a nice little chat. Steve and I have talked about it before too, and I think there's room for everybody. You've got to leave some walk-in time or upsell time. I had a conversation with somebody today, and for some reason they still book right up to the edge of the day on Mondays. I'm like, you don't leave any room for emergencies from the weekend? So I don't think any of us are against walk-ins or emergency appointments. But to base your business on just willy-nilly, seeing what walks in the door, takes some of the business out of the business, doesn't it? I could have a hot dog truck and just hope people come to my stand during the day.

Control the inputs, like a factory 2:44

David Anderson: Yeah. Another analogy you could think about: would you run a restaurant and not know what food is being delivered in the morning? "I guess we didn't get any onions, so we'd better change our menu around." I think about it like any operation. You want to run it as well as possible as, quote-unquote, a factory, and part of that is controlling the inputs into the factory to maximize the output.

I get that in service it's really difficult, because you've got a lot of variability. Monday's busier than Tuesday, but maybe not every Monday. You've got a lot of unpredictability, as you mentioned, Dave, because you've got walk-ins, no-shows, people who are early, people who are late. What people schedule for is different from what ends up on the repair order. If I have 100,000 miles on my Odyssey and come in for an oil change, I'm probably going to get something more on my RO. All this variability and unpredictability makes it hard to control the inputs, but as best as possible you've got to try, to optimize the factory, in my mind.

Dave Rogers: So you have software. How does it work? Give the audience a little bit of an overview.

Emergency room or surgery center? 4:07

David Anderson: Before I go there, if you don't mind, Damon was talking about leveraging practices from other industries, and I was thinking about how to tie that to the whole appointment question. Then I can bring it back to EvenFlow. Maybe the first thing is an emergency room. Why does it take six hours when you go to the emergency room? It's an emergency. It should be fast. There are really two problems that drive it. First, and unfortunately, Steve, this one goes against you: it's 100% walk-ins. There's no appointment in an emergency room.

The second challenge is that everything's variable. Like you said in your other podcast, you get five Cherokees with a check engine light and you get five different jobs. Think about the emergency room: I have an ear infection, or I need stitches, or I'm having a stroke or a heart attack. The level and type of resources required is hugely variable. Then you've got the unpredictability of who's showing up. That's why an emergency room takes six hours when you'd think it would be fast. It's a factory with no inputs or controls. It just takes what it gets, and you have a suboptimal system.

On the flip side, at a surgery center, assuming it's not an emergency and there's no complication, it runs like clockwork. Why is that? Because they know what you're going in for. You don't go in for ACL surgery and say, "While you're talking to me, doctor, take the appendix out too." You're not upsold. You're there for your ACL, and that's it, so they can plan for it.

When you think about service, in my mind, you'd love to run it like a surgery center, but there's too much variability and unpredictability. That's where EvenFlow comes in: helping overcome those ER challenges to make it as much like a surgery center as you can, by controlling the inputs. Part of those inputs is that there are going to be walk-ins, there are going to be no-shows, people are going to be late, people are going to be early. But as best you can, understand those inputs to optimize the output.

Dave Rogers: It's funny you said that, because I used to use the dentist's office as an analogy. You make an appointment at the dentist for a one-hour teeth cleaning and X-ray, and if they find anything wrong, they reschedule you. They generally don't do things the same day. "Next Tuesday at one o'clock, you're coming in to get your teeth filled or a root canal." If you're in extreme pain, they might block out a period of time. One dentist I went to had 12 to one blocked out for emergencies.

I've always felt like, jeez, I wish I could do that. Bring your car in, do an oil change and the multi-point, then sit there and say, "OK, Steve, next Tuesday when you come in, your radio part will be in, and I'll put your radio in. And by the way, you need your 15K service. We did your oil change and tire rotation today, but we're going to do another two hours of work." But we're all afraid of doing that, because the customer is going to go somewhere else and shop us.

David Anderson: Or just never come back. The comeback is the hardest thing.

Dave Rogers: Right, never come back. But it always amazed me. You go to the dentist, get your teeth looked at, and they'll bring you back four times. Fillings on the right side, fillings on the left side, the root canal once, then a second time. They have no problem, and people go, "OK." Same with the orthodontist. He'll X-ray you once and then bring you back for something else. What do you say, Steve?

Steve clears the air: schedule, but keep express open 8:24

Steve Shaw: Well, listen, I like to kid around. I think having your regular service department scheduled is very, very important. What I was really leaning toward was express maintenance. I think you should be able to drive in at any time, from open to close, to dark-thirty, and the dealership should be prepared for you to get express maintenance at any given time.

One of the conversations we had when we met at NADA was about when I was running a big store in Southern California. We had eight groups, eight lateral support groups, about 50 technicians, and we were going from 6 a.m. to midnight, and then we had an overnight shift and so on. And we did this manually. Literally every group had its own appointment taker, and we had these spreadsheets that were eight miles long.

What I found fascinating about what you did is that you automated that with AI. I used to call them service administrative managers, SAMs. The ASMs, the assistant service managers, were the advisors, and the SAMs took appointments. Their job was to schedule and load their one little group. And what I thought was super cool was when you talked about how your software automates that and looks for the holes. Forgive me if I'm saying it incorrectly, but it helps plug those holes so the department stays full. So I think that's a great idea. I do love to be the antagonist sometimes, but I am a total fan of taking appointments. I just want to make sure that if you have an express department, I'd better be able to walk in. Anyway, now that I've explained myself.

Dave Rogers: You had to clear the air on that one. I like that.

Steve Shaw: I did. I was feeling like I'd get my ass kicked. Go ahead, I'm sorry.

Doctors who interrupt every 23 seconds 10:22

David Anderson: I'll lay out one last health care analogy, because I feel this one is really the most impactful, as an outsider looking in. I've worked in health care and still have ties to it, so I'm speaking with some knowledge here. I don't have to pick on them, but I will: primary care doctors. The way the economic model works, they've got to turn you in eight minutes. When they walk in the door, they've got to get out in eight minutes. So they're giving you the health care version of an oil change and rotation, because that's all the time they've got. They know they should give you the health care version of a coolant service or an alignment. They just don't have the time.

One of the craziest stats I've ever heard, I'll just throw it in here: they monitored primary care doctors and found that the average doctor interrupts their patient every 23 seconds, because they don't have time. If you start saying, "Well, my Aunt Martha had this thing," they yank you out: "OK, what about your hypertension?" They have the expertise, but they don't have the time.

Then I think about your teams in service centers. As an outsider, the word "upsell" was always weird to me. What is upsell about? At the end of the day, the team's expertise is recommending services and repairs that help with safety, or maintain and extend the life of your vehicle, or manage repair costs: pay $200 now to avoid a thousand-dollar bill in 12 months. Those are all things in the customer's interest. In the same way, I didn't think my doctor upsold me on a statin. He said, "Your cholesterol is too high. If you want to extend your life, you should probably get on a statin." I said, "OK, thanks, doctor. That's good advice."

I know there are a lot of reasons services aren't pulled through in the drive, but one for sure is time. If the team is backed up, they don't have the time to do the full inspection or talk to the customer. I go back to your example, Dave, of Philly Honda, where every RO was 0.9 because you were just jammed up. I'm sure there just wasn't time to talk to the customers. The economic model is such that if you build in the time, you can do it, unlike primary care doctors who just give you the proverbial oil change and rotation. I think about that a lot when I talk to the dealers we work with. We're trying to create time for them.

Automating the algorithm in the service manager's brain 13:22

Steve Shaw: So how do you do that? Give me an example of how you've been able to help a store drive traffic through appointments, because I know that's what everybody's looking for. We obviously have a capacity issue. We sell time every day, and we want to sell all of it, or be within 10 or 15% of our capacity, and maximize it. A technician's time, if he doesn't sell it, is lost. That's everybody's biggest fear: if they don't book enough appointments, they've got all this time left over, and if they book too many, the customer waits. So how do you go about maximizing technician time in the store? I think that's probably what our audience cares about most.

David Anderson: I think it goes back to two things you both said. First of all, Steve, our goal is to automate the algorithm in the service manager's brain. We're not coming in and saying, "Hey, I know better, this is the way it should be." It's: how do you want to run the shop? If you could schedule 100% of the appointments yourself, in all the different scenarios, what would you do? We're trying to extract that and then power it with software, math and automation. That's key.

Then, Dave, what you talked about were aspects of the variability and uncertainty. If there are a lot of walk-ins, we'd better bake that into the plan. We can't schedule up to 100% and assume zero walk-ins. Or, as you say, upsell. If we know the average Ford job booked is 0.7 hours, but on average the RO might end up at 1.3 hours, or it varies by mileage, 0.7 at 10,000 miles and 1.6 at 90,000 miles, then we should be baking that into the scheduling itself. That's the prediction we're layering in.

The other thing we're doing is tuning it to the team's capability. We adjust the clock time based on their efficiency. If it's 1.6 but you've got some really skilled guys who beat the clock, or your express is filled with apprentices and guys straight out of school, we'd better account for that. All of that factors into the core of how we bring the flow in.

The other aspect that's really key is how you flow it in based on the specialization of the job. It's like in an emergency room: you don't treat a stroke patient the same as someone who needs stitches, and you don't use the same people. So there are certain recalls it doesn't make sense to bring in at peak times, or you're on parts allocation, or it's a huge job that would chew up a technician all day long, so you only want one of those a day or the whole shop gets jammed up. Again, it's the service manager's vision of how they'd ideally do it, plus all the nuances and variability you see every day trying to execute against it.

Steve Shaw: I remember back in the day we did tire recalls at night. We did the Firestone tire recall at night, and I had a whole crew of people. We'd rotate our groups through the evenings and have two or three groups doing tires every night, because we didn't want to clog up the shop with them. And more recently everybody had the airbag recall, the Takata recall, and we didn't want to clog up the shop doing airbag recalls all day long. So if I'm hearing you right, your software can take those different types of repairs, plot them out and help the dealer maximize efficiency. Is that what I'm hearing?

David Anderson: Yeah, it's an optimization for each type of request or specialized skill set. The traditional way is main shop and express, but you could carve it up any way you want. Another example: there's some OEM recall work where you get paid 0.3 and there's no way humanly to do it in under an hour. These are nuisance one-line ROs, especially if they show up in the drive at 8 a.m. But if they show up at two or three o'clock, off-peak, when everyone has time, it becomes an opportunity. You see some CP work start getting added. It's all about optimizing the business, the factory, but the byproduct is a better customer experience too. If you're the customer with the nuisance RO, you feel it. Everyone wants you out of there, and everything takes longer.

Dave Rogers: Really? You guys can tell?

Steve Shaw: Is that like me driving my Wagoneer into the dealership and saying I have an electrical problem, and they all run and hide?

David Anderson: So that's what's been key for us. The underlying framework is the prediction on top of what we think the job is, translated to clock time based on efficiency. But carving it up by specialized request, or where you want it with specific resources, is where we get a lot of the magnitude of the benefit.

Training the AI and predicting job time 19:33

Dave Rogers: One of the things I'm discovering with AI, and I know it's a broad term and there are different types out there, is that a lot of folks just say, "I'm going to get this AI technology, plug it in, start it and forget it." To me that's ridiculous. I look at AI like a child or a new employee. You've got to teach it, train it and coach it. So I've got to assume that's what you have to do when you're starting out in a store, getting the service manager's feel for how things work, and every brand is different.

You were talking about recalls. There are connecting rod recalls on Hondas and Acuras where you've got to drop the pan and measure the bearings, then order the parts, and now the car is dead on the lift. You're right. You don't want to do that at eight o'clock in the morning. You want to do it at three o'clock in the afternoon, so when the guy leaves, he's got a car to come in and put the bearings in the next morning. So how do you work around that? Is that all input? How do the variables work for you?

David Anderson: First, I'd reiterate that I think about AI exactly the way you do. I think of it as a digital employee. In that context, would you ever hire an employee and say, "I'll check back with you in 12 months to see how it's going"?

Steve Shaw: We do that every day. That's what we do every day in the car business.

Dave Rogers: Yeah, we do. It's called the Earn As You Learn program. That's the core business, man. Get out there and keep going.

David Anderson: Well, that creates training opportunities, then. You've got to work with AI. The AI is only going to know what you train it on. It's a feedback system. And the way our product works, I always think we're 50% software, 50% people, meaning we have to earn the trust of the service manager or fixed ops director, whoever it is, that we're the tool or enabler of their vision. That's a people thing. So we launch, configure and deploy, but then we're the co-pilot or the caddy along the way, tuning and refining. And even as it's learning, your business is so dynamic. If two techs quit, whatever we thought is now different. If I just got three people straight out of school, that's different. If we just got hit with a major recall, that's different. Even if we think we've got it right, we've got to stay alongside and evolve as the operation evolves for every service manager.

Steve Shaw: You made a comment about prediction that stuck in my head, and I wrote it down. I think I heard you say that if a 2026 comes in with 10,000 miles on it, you'll allocate X amount of time, but if a 2017 comes in with 100,000 miles, you might allocate a different amount of time. Is that what you're saying?

David Anderson: Yeah. I also preface it with this: I'm 100% wrong in every prediction. We've never gotten one right. It's tough, because I don't know if you're going to decline all your services, or if this engine's got corrosion in it. How would we ever know that? Anyone who claims they can, I think they're just full of it. But it doesn't necessarily matter, because we don't have bias in our errors. If I'm a half hour too high on car A and a half hour too low on car B, it blends itself out, and we still get the flow in through the day. I hold no crystal ball to know, "Hey Steve, your 63,000-mile vehicle is going to have a coolant service."

Steve Shaw: Right, but you're using an algorithm. If I understand correctly, and I haven't used the software, you're saying that based on the history of your particular store, you can predict X, which is going to be a lot better than what most people do, which is block out a half hour or an hour for every appointment. That's it. And we hope we fill the shop by scheduling 25 or 40 appointments today, or whatever.

David Anderson: Yeah, and the law of averages plays out well. I've heard you say this before, Dave: "We've got 100 hours, it's two hours an RO." If you look out 90 days or half a year, that's probably spot on. But on any given day it could shift wildly because we got five more diags than we thought. And Steve, to elaborate, we tune all the models to each dealer's data. It's not some theoretical "the factory says it should be this." We look at the actual behavior of the dealership. It could be that these advisors and technicians are amazing at inspections and working with customers to pull through services, or they're horrible at it. We build it on where they're at, so we plan for it. And it's a learning system, so if they get better at pulling through more flag time than we thought, we update the models accordingly.

Dave Rogers: So you're taking a baseline. The way I look at it, like you said, 100 hours at two hours an RO, and yes, some days you had too many cars with low mileage and other days you struck out and couldn't upsell anything. But I've noticed differences between a lot of my stores when I look at ASR percentages, the number of recommendations we make per car. Obviously it's driven by mileage in a lot of cases, and it could be brand too. And then you look at the advisors' closing ratio, and it varies. I have some stores that can't crack 30%, and other stores at 50, 55%.

When I made appointments for my own store, I could breeze through them and kind of know what day I was going to have. But as the day went on, I'm nowhere near as smart as AI or a computer at figuring this stuff out. I think you're taking it to the next level, which I think is great, because there are a lot of variables in what we do, and a computer that can figure this out with a good data feed from the store is going to make a difference. You can't make one policy that goes across, in my case, 34 stores. It just doesn't work. They're all different.

David Anderson: Yeah, 100%.

Record months and a 430-hour day 26:48

Steve Shaw: And we didn't bring Dave on as a sponsorship to say, "Hey, you've got to buy Dave's stuff." We really were interested, because the conversation the other day about appointments or no appointments was fun. But I love the concept of what you're doing and how you're plugging those gaps. Everybody is out there struggling, because, like you said, I can make an appointment every half hour, and if somebody isn't paying attention and all I make is oil changes, I end up with half my shop filled. Or if I plug in all these appointments and they all turn out to be check engine lights on a Cherokee, as Dave would say, I'm way overbooked, I can't get anything done, and I end up with a bunch of unhappy customers. So share with us some success stories. You don't have to name a particular dealership. How have you been able to help stores with this tool?

David Anderson: I'll preface it like this: most of the great ideas I've heard, I'm writing down in notes. The service manager tells me, "Hey, this is something I'd like to do. Can you help enable it?" We execute it, and then we go to the next dealership and I say, "Hey, I've got a great idea." I'm not creating these great ideas. I'm just taking good ones from people.

Steve Shaw: That's what I do.

David Anderson: What we've seen is that we've increased dealerships' revenue by 20, 30%. We've broken monthly service gross records in the first month, then broken that again the next month, and again the third month. We've got many case studies where we're breaking all-time-high service gross records for them. A lot of it comes down to pacing the time to the service manager's vision. Like I said with the recalls: these nuisance one-line ROs, we push them to the afternoon, and now we're getting $150 in CP on top of them, and we've got better flow in the morning. Or it's load balancing if you've got a big shop with tech teams. We may not be doing real-time dispatch, but if we have team A and team B, we make sure they're showing up at equal balance. Dave, you don't have ten diags while Steve has ten brake fluid services. We're balancing it from that standpoint.

Steve Shaw: What was that video game we used to play back in the '80s? Tetris?

David Anderson: I use that example, honestly. We look at technician capacity, when they roll on and off the clock, and we try to pack it in like a perfect Tetris board. If we get that right, then you've got time. I feel like we enable the operation to work with the customer to pull through all the services the vehicle needs. It doesn't mean you're going to execute, but we at least set it up to be possible.

Dave Rogers: It's amazing how many times you'd get one of those days where everything went smooth. You made the most money, you made the most hours, everything just clicked perfectly, and you go back and look: how did we do that? It's the work mix, it's the flow. But the one thing I wrote down and circled: I assume the AI factors in no-shows. I was looking at that the other day with one of my service managers, and their no-show percentage was 12 and a half percent. I'm looking at it like, is that a good number or a bad number? Is that normal? How do you factor no-shows into the equation?

David Anderson: To me, no-shows and walk-ins are forecasts. We can never predict them on any given day, but we can look at them by day of week, by service book or stop, and by time of day. There are patterns, and we factor those forecasts into the planning. Monday morning in the main shop, we're going to have a few tow-ins or whatever. You can look at it, so that's planned for. We're going to have more walk-ins on Monday and Friday and fewer on Tuesday, and obviously they drop off throughout the day. So it's not one size fits all. We base it on the data, monitor it and adjust accordingly. We're not perfect. Every calculation has an error in it. What we're trying to do is be more precise than where you are today with the law of averages, and that gets you closer to where you need to be.

And Dave, you brought up those days. Anecdotally, what our service managers usually say when we get it tuned right is, "The days are easier, and I'm setting records." I remember one service manager we had just launched with, and I asked him how the day felt, because I'm looking for that qualitative feedback. He said, "It felt pretty easy and smooth. It's early in the launch, so if we got 300 hours today, I'll take that as a win. That's still a decent day." I said, "Check it out. How many hours did you get?" He said, "Well, I haven't closed everything up, so take it with a grain of salt." Then, "Whoa, we just did 430." They had only done 430 once in the last 12 months, and it didn't feel that way.

Where to find EvenFlow, and the GM tip 33:24

Steve Shaw: That's awesome. That's a great place to tie this all together, because we're going a little long. But I love it, man. So many people would say, "300 hours? I'll take 300 every day." That's a pretty good-sized store. That's the stuff that warms my heart. I love these big stores doing three, four, five, 600 hours a day. To wrap this up, Dave, tell us a little bit about where people can find you and find out about EvenFlow. Then I'm going to turn it over to Dave Rogers to give us a GM tip of the day.

David Anderson: OK, cool. The easiest way to get hold of us is through the website, evenflow.ai. Just put in a contact-us message. You can also email me at DaveA@evenflow.ai. People reach out all the time, and I'd love to talk to you if you're interested.

Steve Shaw: We'll put the link in the show notes when it goes live, so everybody can click on it from there too.

David Anderson: Cool, I appreciate that.

Steve Shaw: It's great. We had fun talking to you. When we met at the convention, I thought back to my days running that big store, where we had eight people for eight teams trying to do everything. It sounds like that's something you can pretty much do with one digital employee, with some coaching. That would be freaking amazing. And you've got to keep coaching that guy. I loved it. I've never seen the software, but the concept you're describing definitely warms my heart. If a store can do a couple hundred more hours in a day, hey, it's worth giving it a shot. What do you think, Dave? Want to wrap us up?

Dave Rogers: Yeah. My head's spinning, because I'm like, all right, I've got to see how this thing works. GM tip of the day: I try to relate things to sales for all the GMs out there. You get 100 leads, and you're looking for how many appointments. It's a numbers game. That's how they look at it with the internet department, the CRM or cold calling. I think you have to look at service the same way. You have 10 technicians. We know the numbers. We've got to figure out how we're going to get more people in there, and we can't leave it to chance.

Now they have AI emailing customers and trying to book appointments from internet leads, and obviously phone calls. You have to look at service the same way. You're trying to stack the shop the way you stack the sales department, to maximize your success. It's not about anything other than putting bodies in front of bodies, in the right way. That's really what you have to do.

Steve Shaw: In the right way. Awesome. Well, Dave Anderson from evenflow.ai, thank you so much. Dave Rogers, thank you for bringing your big guns with you today. Let's get out of here, because we're way over, man. It's the longest episode we ever did, so hopefully everybody out there got a little extra today from Dave Anderson. Let's keep in touch, and let us know how people are doing with your software. Appreciate it, Dave and Dave.

David Anderson: Cool. Yeah, thanks, guys.

Dave Rogers: Thanks. Have a good one, man.