Inside Retail’s AI Transformation with Greg Buzek of IHL Services

00:00

I’m Eric Chemi, and this is Politely Pushy. Welcome to Politely Pushy. I’m your host, as always, Eric Chemi. Today, I’m joined by my friend

00:12

Greg Buzek from IHL. One of the most fascinating analysts that I know, especially in retail. Greg, you’ve been all over it. the amount of depth, detail, and thoroughness of how you’re looking at the major retailers in America and in the world today and

00:28

especially what’s happening with AI, right, in both its usage in retail and its usage as an analyst and in the technology space. I’m always impressed by you. We’ve been friends for a long time. So, I’m excited to actually have you here on the podcast for the first

00:40

time. Thanks for joining me. >> Yeah, my pleasure, Eric. It’s great to be with you. So, you know, first of all, when when you’re looking today, right, we’re here early October, fourth quarter’s started now. This is this is the quarter for retail, right? >> Right.

00:55

>> What are you focused on right now? Markets at all-time highs. Gold super, you know, at all-time highs. Bitcoin’s all all a lot of asset prices are at the top. How does that affect retail sales and what these retail companies are going to do about it?

01:10

>> Yeah. Well, we have a k curve when it comes to uh consumers right now. So, our focus is mostly on the consumers, not even the retailers per se uh in the sense that there are uh bottom 60% of consumers that are really struggling right now. Um and now we’re in the midst

01:27

of a government shutdown as well uh as we’re recording this, which adds even more pressure to those folks. The flip side of that is the top 40% and particularly the top 10% have seen their portfolios gain, their houses gain um their purchasing power grow. Um but

01:45

they’re under the threat of AI uh either as a benefit or or a curse. So right now we’re dealing with a lot of uncertainty as we go into the holidays. That uncertainty tends to cause people to freeze rather than heavily invest uh there. So I think the holidays are going

02:01

to be somewhat modest and gains are going to be muted as a result of that. It’s just because we don’t know we don’t know the fullest impact of of tariffs. Um most people though just to step back there are positives in the change in the tax code or the changes that

02:19

occurred that right now we’re feeling all the pain and next year you feel the the results of that. So when you think of the increase in the salt deductions etc like that that is also something that I’m not sure everybody is focused in on even though the higher uh

02:35

percentage or higher cortiles are the ones that are going to benefit the most from those uh tax deductions I’m not sure they factored that into their spending either. So the end result is just a little bit more muted holidays than we would have otherwise seen in the

02:49

last couple years. So talk about, you know, you starting IHL. How did someone even start their own? >> Yeah. >> Of course, thank you for that framework kind of what you’re seeing economically, but now I want I want to go widen that lens. Why did you start the firm?

03:04

>> And I feel like analysts are like my my alter ego. I feel like my my fundamental core identity is an analyst and if I wasn’t doing this, yeah, probably doing something like you’re doing. >> You would do a great job. >> I have so much smart stuff to say. I’m

03:16

going to start my own firm and someone’s going to pay me to say it. Yeah. Well, I I wish I could say I had this grand business plan and even if it was on a napkin. Um, literally I needed a job. I got fired uh uh from my previous job. Um, which was basically

03:32

trying to convince people convince scanner companies to put 18 pounds of copper wire with a magnetic field in a in a metal situation, a mirror spinning without impacting their performance. And I couldn’t convince them to do that at that time. It took Walmart to tell them

03:47

do it before it actually happened. But um so I needed a job. Um I had a friend at NCR who needed help with some product uh release stuff and I had done competitive analysis before and I took that on as a contract and then that I saw the writing on the wall there that I

04:06

couldn’t have just a single client and there was a need to expand that. And so we went to, hey, what if we did one a a single report that we could sell to multiple people instead of doing everything uh custom for each person that came in. And that really was the

04:21

launch. Um as well as some other folks who uh grew up, they got promoted, people that I knew that trusted me, and uh they they brought us in and next thing we know we’ve grown quite a bit and it’s been almost 30 years now. Has the mission evolved or changed since

04:38

then or has it been the same idea from the beginning? >> Yeah, it it’s changed quite a bit. The the most recent pivot would be obviously towards towards more towards AI um being experts in AI and retail adoption of AI, but we started off focused specifically

04:53

on point of sale um and point of sale technology. So there’s uh you know how many point of sales systems are there? How many stores are there? How many stores in the world? What segments are they in? We’re the guys that count that and we still do that today and that is

05:07

part of our core. And then the second iteration is we release a data service that tracks what retailers use, what what are the key installs and the key leaders within that organization. A third level then became IT forecasting um and then eventually turning into what

05:24

we see now with uh with AI. Um but there’s also the business focus as well as we got to a certain level of success. We also switched to a partial piece of our business was actually leading a charity um and and that we lead a charity that is good dogooders in the

05:42

retail industry to help orphans and vulnerable children. So it was not just about making money for us. It never was. It was like, you know, we want to make a living, but we want to build it, but we want to do good for the world. >> How many people are at IHL? >> We have a total of 12.

05:58

>> 12. Okay. That’s pretty good size for a boutique because we’ve talked to analysts obviously are working at Gartner and Forester and then there’s people that run their own oneperson shop, but 12 is is big enough to have a lot of headaches. >> Yeah, we have five full-time and four of

06:13

those are analysts. And then we’ve got a number of part-time. >> Who’s the main customer? >> The main customer for us are the vendors, large vendors and and medium to largesiz vendors, hardware, software companies. Um we’ve worked with folks like Intel and NCR Voyagex and SAP,

06:31

Oracle, Microsoft, those kind of folks, those names that everybody recognizes. Uh Toshiba Global Services or Global Commerce, excuse me. um Fujitsu those are the guys that have primarily been our customers um about 90% and then about 10% of that comes from retailers

06:48

uh themselves and uh doing various things whether it’s benchmarking their solutions or now more than anything just AI where do I start how do I implement it you know >> usually I think of it as the opposite usually I feel like the vendors are trying to get in front of your customers

07:04

but in this case the vendor is the customer >> correct that’s how we pay our bills Um, you know, as a small boutique, it’s hard to get retailers uh to spend on analyst services. They they they buy Gartner, maybe Forester, but their budgets are eaten up by that, and it’s hard to break

07:22

into that piece of it. Um, we just happen to be in a very strong niche um that happens to be about 20% of the global G or global GDP. And as a result of that, um, we we have a strength that’s a depth that most others don’t. even those large guys. >> How does it work from an analyst

07:42

relations point of view? Right? We we know have a lot of firms that say, “I want to get in front of the analyst that they’re going to write about me so they’re their customers of research will see it and bring us on.” How do you fit into that ecosystem?

07:55

>> Yeah, it’s uh it depends on what their niche is specifically. I mean, you know, you think most most of the your crossplatform vendors, they think of retail or restaurants. They don’t get down to the 300 technologies that are actually make that work. We do. Um, and

08:12

so where others get into the forest level analysis, we get down to the bark level in some cases. We count how many point of sale systems were shipped this quarter. Um, and what processors were in those. um that level of knowledge and depth gives us a level of credibility

08:29

that um we’re not just a driveby in terms of the solutions that we’re going to offer them and the insight that we’re going to offer them. Uh the other thing that is uh impressive about our research the research side of that analyst relationship is that we don’t stop at

08:45

just the highle data. We just don’t say, you know, for instance, um if you if you if the R if it’s an RFID vendor and the story is about RFID, uh having a a data point that says 10% are using RFID today and another 17% plan to use it next year. Yeah. So what?

09:04

But if I came to you and I said, you know what, the fastest growing retailers are 160 times more likely to be using RFID technology. Those vendors love that because that is huge in their world. So a lot of our relationships and analyst relationships are are not just about

09:22

them selling us on their products um but also promoting them. And uh in most recently we’re uh we’ve released a number of of of research products that are just here is the market here is the ecosystem of players and it’s not a paytoplay. You don’t have to be an IHL

09:40

customer to be in the report. And so they like that independence uh piece of it and uh and just here’s here’s really what’s happening in the market, not just who has the money to pay the analyst firm. >> How do you feel about that? Because I I’ve talked to so many analysts and they

09:56

say, “Yeah, the big guys are all pay to play. I I don’t know what this is. It’s a game. You everyone the big guy the big customers pay then they get mentioned. It’s just an eos.” >> Yeah. It’s a good business if you can, you know. >> Yeah. if you can do it that way. But,

10:09

uh, it’s it’s a little dishonest because the little guys never really get a shot. Um, and that’s that’s the downside of of that approach. We’ve always taken a separate approach that our responsibility is what the market is. Um, now we we have data and we may

10:28

customize data to a sponsor if it’s a sponsored piece, but you don’t have to be a sponsor for us to present data out there and to have that data out there. So, when you’ve got companies pitching you to to, “Hey, mention me. Get me in the report. I I have something important

10:44

to say.” How do you want to deal with them? How how do you want to deal with them? How do you not want to deal with them? What What is over the line in terms of now you’re bugging me or you’re not big enough or you don’t matter enough. Stop reaching out. I’m not going to mention you.

10:57

>> No, we actually work with pretty much everybody as long as they’re willing to give us their information. Now, we’re going to run it through an objective process uh for things. Um we will we will sell reprints of the final results. Um and obviously the better you come out in

11:14

that is the more likely you’re going to purchase that reprint opportunity, but that really doesn’t impact what we cover or how we cover it. Um I’ve got, you know, one of my guys that does a lot of that work. He worked on the space shuttle uh you know on the engines you

11:30

know of the space shuttle. >> His stuff is very logical and very objective and it just spits it out and here it is here are the results uh for this and there’s very little wiggle room. Um if there’s wiggle room at all it’s you got to move dots you know so

11:46

they’re not overlapping each other. So you’re you’re comparing one to another just to move a dot for a visibility thing, but uh not not the core core piece of it. >> You mentioned the phrase if they give us their information, correct? What information do you want and what are you

12:03

doing with that information and who might not info? Yeah, usually we’ll have an RFI like right now we’re doing one for point of sale uh right now and I want to say we probably have 60 to 70% of the market in terms of market share of people that are participating in that

12:21

and we give them about 150 questions that go into uh market situation, market size, where are they installed, what segments are they installed in, what feature functions do they have, um are they using tools like AI? uh what’s their architecture? Um can they run on

12:38

the edge? Do they support an edge scenario or is it only cloud? Um you know, or is it traditional client server type thing? And then we score each one of those. Um and as it comes through, they get so many points and and once they answer that and take that serious,

12:55

we get that information. The people that are not in there, we still include, but we’re now guessing at what their their uh features are for things. And generally those folks tend to get lower not not because they uh they didn’t uh get the information. We just don’t know.

13:11

We just don’t know to answer it uh type of thing. So I mean we’re quite confident that you know 60 to 70% of the market that we cover. We’re we’ve had a good representation of the of the market but ch I mean we choose presidents for uh literally you know a thousand

13:28

registered voters. We choose presidents. So I think 60 to 70% we’re on pretty solid ground there. >> Yeah, that’s true. That’s true. Your pivot to AI, has that just been in the last three years since? >> Yes, it has. >> I studied neural networks in school 40

13:44

years ago and it was over my head. You know >> what grade did you get? Did you pass? >> I passed the class. Yeah, I passed the class. But it was it was one of those things where I couldn’t be a data engineer. I just did not um the programming the level of programming and

14:00

the level of things there uh was was such that it was just in I mean it was just outside my skill sets um and interest level at that time and like everybody else when I saw Chad GPT for the first time and then dug in a little bit on the neural network aspect of it

14:18

is it was like they did it they did this stuff we were talking about 40 years ago and this is going to change everything And at that point I was like I have to pivot our business and start looking at AI not only from a perspective of what it means for retail and how big that is.

14:35

We were the first ones to come out with a forecast for retail. Um we had $9.2 trillion over like the next seven years. Now that may be delayed a year or two um because we didn’t know what we know now uh in terms of what’s required culturally in an organization to deploy

14:53

this. Um but we we started using the tools internally and I think that’s the fascinating thing here. What we’ve learned since as of as of today uh for most large organizations this is like uh taking a battleship deconstructing it to make an aircraft carrier while you’re on

15:13

the water in a battle. And there are going to be some that are going to be able to do that. people like Tractor Supply and Walmart um there are way ahead of others because they’ve made that pivot and they’re moving that direction. And then um and then there

15:29

are others that are going to have a hard time doing it. And one thing I I really want to get across here is the success of AI is about culture, but it’s also about financial stability of the company currently. If you are growing as a company and profits

15:48

are improving and everything’s improving in a company, AI adoption is usually accepted by the working class there. People want to learn how to use the tools and they begin to leverage those tools. However, if you’re a company that is not growing, you’re struggling, the

16:04

end result is you’ll have people pushing back against it and the culture will eat that strategy change. And uh you’re going to have a really hard time to adopt it because they look at AI as that’s a replacement for me. Um and and that’s the piece that most companies

16:20

don’t think about is that if you’re struggling financially, you better overcommunicate overcommunicate that this is uh this is not to replace you. This is to augment what you’re doing. Um and there’s got to be trust there uh for for that that company with their

16:38

employees because otherwise it’s not. So when you look at the in fact I actually vibe coded a tool that evaluates the current Fortune 500 from an investment standpoint and predicted how well they could do um as a result. >> It’s interesting you mentioned Tractor

16:54

Supply and Walmart as your two like all right Walmart is the biggest retailer in the world and or you know one of and and Tractor Supply is is not that that big. Why are you putting those two in the same bucket? >> Believe it or believe it or not I was

17:07

shocked. I was shocked too and they’re right up the road here from where I live and in Tennessee. Um they were doing Agentic AI in January of this year. Um where most people were struggling to use uh you know AI and machine learning tools or start to use uh you know chat

17:26

GPT and other Gemini and Claude and other tools. These guys were already past that. They had cleaned their data. They had had done all those pieces and had built the tools already and they’re like they were already building agents seven eight months ago.

17:43

>> Why them like is it some visionary leadership or they >> Yeah, it’s leadership. I mean Glenn Allison there uh is a friend who uh has done a terrific job with what they’re doing in data intelligence and and strategic marketing. But they’ve got a they really have a terrific team there.

17:59

>> What would that look like for customers though? How are they going to see the difference? A lot of it’s hidden. Uh a lot of it’s hidden or it’s part of the app uh for which that customer is using. But a lot of it is associate level uh things that they’re building today that

18:12

that help the associates find the product to pre-stage the products for pickup, those sort of things. Um to be able to see they’re they’re using computer vision um to anticipate lines uh forming so they can get the right personnel in the right place uh for

18:29

things. Um, but they’ve got a lot of different tools. They’re individual pieces, not there’s not like this whole big system. They literally went to individual tasks and pain points and started doing it. Um and I think that’s one of the challenges that uh is most

18:45

concerning when we look at the data is there are a number of of uh struggling retailers which we would call lagards flat flat or negative sales who were looking at generative AI as an endaround to catch up and >> it’s not going to work. It’s like if

19:01

your business is is declining AI can’t AI can’t fix your business problems. >> Yeah. Your data. And I think the big mistake that everybody is making out there right now is they’re trying to add AI onto an existing process uh of what they’re doing. They are they’re

19:20

literally saying this is the way we do things. How can AI make this better? Um what they’re not doing is rethinking the entire process from an AI perspective because that’s where the multiple uh percentage gains are. That’s where you double, triple, quadruple. We have some

19:36

tools where we see 27 to 55 times uh what we were doing um previously the speed I things that were taking four days that are taking five minutes uh now to do because we reorganized the entire process. >> It’s it’s a good point you mentioned. If you’re in a declining company then

19:56

workers think you’re trying to lay me off with AI. If you’re in a growing company it’s like okay well we needed to hire more people anyway. No one’s laying me off. Yeah, it can help me. But I see how that’s a totally different mental uh model. >> And I’ll tell you, we’ve got some

20:12

research coming out here soon, and I’ll give you a sneak peek here. One of the things that’s most fascinating is the delta between the executives and the uh the worker bees uh within the organization. >> I bet. I bet. Yeah. and and it’s um if you ask will there be more people or

20:32

less people working in these different functions and we looked at stores we looked at supply chain we looked at um headquarters and we looked at delivery uh personnel there in every single case it was upwards of 80 to 85% of the executives said there will be fewer

20:51

people there um when we asked and and the flip side of that is the worker bees thought in every one of those cases which I think other than maybe delivery they said there will be more people working so there is a massive disconnect on that but when you ask about what’s

21:10

the reason for that why why is that the worker bees say overwhelmingly AI is the cause of that issue and the executives say nope we’ve been dealing with trouble getting people for years and we’ve just optimized our processes to require fewer people in the store. So, one of

21:32

>> Who’s right? Who’s right? >> Huh? >> Who’s right? >> Well, I think the executives I think the executives are right because we’ve got some advancements that we’re seeing in AI robotics for inventory. Think of the number of of people in the stores that

21:45

are going around just checking inventory and trying to make sure the inventory is correct and filling in shelves, etc. Well, if you have a robot that’s doing that now, you’re you’re you don’t need all those people. Those people just get into the stocking uh side of things, um

22:02

which means you can do more with less. And that’s been the challenge for retailers for the last 5 years or so is they’ve not been able to get enough people to work in a retail environment. And that and that that extends to restaurants. I think it’s on average the

22:14

average kitchen has four fewer people working in it than it did prior to COVID. And you’ve got to figure out automation. Now that automation may be AI related, but it may be process related, how we do things. It could be improvements in like Sam’s is doing with

22:30

we’re removing selfch checkouts and and etc. And we’re moving checkout stations because we’re just going to let you scan scan and go on your phone. So consumers love it. >> Every time I pick something up, just scan it and be done. >> You scan it and go. It’s glorious. If

22:44

you haven’t done it, it’s it’s glorious I I I went in and out of Sam’s in three minutes at Christmas. Um because I could literally just go in, pick what I wanted, scanned it, walked right through a tunnel that looked at my items and said, “You’re good to go.”

22:58

>> What if you don’t have a phone? Like in the sense you don’t bring it with you or maybe I’m trying to not be addicted to my phone and I want to leave it at home and now I’m like, “Oh my god, I can’t even buy anything.” >> Well, you you could buy something.

23:08

You’re just going to stand in more lines. >> That’s how they get you. >> Yeah. I just read an article today about Gen Z. It was in the Wall Street Journal and it was about how they’re trying to there’s a, you know, movement of let’s get regular cameras, let’s go buy CDs,

23:22

let’s use flip phones, let’s be off of smartphone so we’re not doing everything on our phone. And I I want to be like them, but then you get caught in these things, right? Where it’s like, oh, you go to Whole Foods, hey, we’ll scan your Amazon app for the QR code to get the

23:36

discount. Like, well, I can’t do that with a dumb phone. >> Right. Right. There’s a big my my mother uh she lives in Cincinnati and Kroger and she feels that Kroger discriminates against uh old people because they weren’t printing out the coupons anymore. They finally relented and just

23:51

started bringing back print coupons because you needed to be able to know how to use the app >> to be able to get the coupon. And uh so their decision was we’re just going to stop shopping at Kroger as a result because we can’t figure out the the

24:04

thing. We’ll go to Aldi. And that forced Kroger to bring back paper coupons. Yeah, because a lot of these apps, same thing with CVS, trying to pick up my kids prescription. Oh, you know, you can use the app. And Bob’s like, I’m not using an app for CVS. Forget it. Like,

24:18

I’m just gonna I don’t want to >> just think how much paper you can save. >> The receipt six foot long for one item, you know, >> so bad. Like, I didn’t even ask for the coupons. I don’t want the coupons. I just want to pull up and they have like the drive-thru prescription pill. Just

24:35

pull up. I’m gonna give you a name and a birth date and you just give me the medicine. I don’t want to use an app with a code and and I don’t want any of this stuff. >> Yeah. Exactly. It’s It’s almost like you want the Little Caesars’s What is the Little Caesars pizza pickup thing? Give

24:48

me a QR code. Let me do that. And you spit it out, you know? >> Yeah. Then you mentioned Walmart too as as being able to kind of rebuild the battleship while they’re on the water, >> right? Yeah. Walmart hired more data centers than Google for about three to

25:02

five years. Um and the end result has been somewhat remarkable in the tools that they’ve been able to do. In fact, Doug McMillan came out last week uh mentioning that uh literally every every job will be impacted by AI, particularly in a Walmart environment. One of the

25:20

tools that’s having the greatest impact uh I call it Ask Sam. I’m not sure what the exact name is, but every associate has access in their native language to the the behaviors and the and the uh processes for Walmart. So, how do I change the printer paper? What is our

25:38

what is our uh return process for such and such? Um, where can I find this uh piece of information? Where’s our inventory for this? They can now ask in their native language rather than having English as their primary language. and and that works every all over the world.

25:55

So they have 500,000 associates doing that. But uh just to just to give you just the opportunity, the sheer volume of opportunity when we put together our original forecast of $9.2 trillion, we were going from top down worldwide looking at things. Now uh Kathy Wood, I

26:13

think for the same area, she she benchmarked almost $14 trillion. I just couldn’t get myself going to that high of an impact on retail um there. So 9.2 worked until we started doing the bottoms up. And when you realize that um over over 75% of the trucks on the road

26:35

today are less than 25% full at any time. They’re that empty. >> Yes. There are tremendous inefficiencies across the board when it comes to cost of goods. that is in the transportation layer that can be optimized with AI. It’s too complex for humans to do, but

26:54

with AI, it’s it’s a pretty easy process. What we looked at was Walmart’s margins at the time were 24.2%. If we using AI, if Walmart’s margins increased to 25.8%. Across the board, our 9.2 trillion was too low of an impact worldwide. Um that is the kind of scalability that you have

27:17

when you have two three 4,000 stores where what a little improvement just scaled to that level is is seismic and um that’s that’s why things but Walmart is a is is really one of those companies that you prioritize where do we have clean data let’s focus on

27:37

getting the the data enablers as we call it which is our core pieces of data that will most be impacted and see the impact from AI if we clean it. Focusing there and then unleashing that those uh trials to thousands upon thousands of stores and they’re seeing tremendous impact as

27:55

a result of that. >> You mentioned trucking. Are we going to see AI fundamentally change trucking? It’s driverless trucks. They’re all at 100%. They’re >> that’s where most of the focus is is is on driverless uh trucks. And yes, there’s going to happen. We have a

28:11

tremendous shortage of of truck drivers that needs that needs to happen because of the shortage but it’s more about load factor and it’s miles driven and it might be electrical trucks versus you know diesel trucks. Um there are going to be a lot of innovations there but I

28:28

think the single biggest uh area is going to be load factor. Um, one of the things, >> what does that mean? Load factor for people >> as I mentioned those 75% of the trucks that are less than 25% full that that back haul route going back filling most

28:44

more of those up. Okay. So, that’s one of the things that’s uh rather fascinating. So, if you imagine how many empty cargo uh containers go back to China or go back east, what if those were full? I mean, that’s the easiest way to visualize it. But I’m not only

29:02

I’m not only dropping stuff off at this store, I’m picking stuff up. Or maybe I’m just going to a store across the parking lot and picking something up from another. It might be a competitor, but that pallet is now going somewhere else. That’s just not really done well today.

29:21

>> You But you think AI can solve it or make it much better? >> Yeah. thirdparty logistics providers, uh people within like specific um uh companies, you know, the size of a Walmart. It’s it’s their it’s their stuff. You see you see it happening to a

29:38

lesser degree on the last mile delivery where people are delivering uh different packages. So Walmart’s actually delivering Amazon packages and and and other things in some cases because of the last mile. And it’s like if I can optimize that delivery driver, that

29:55

delivery vehicle with multiple uh companies, multiple solutions now, now I’ve dramatically lowered the price for both. >> How will that affect consumers? Is it just, hey, retail inflation might be kept in check because their costs might not go up as much? Is that is that the

30:13

benefit? I think it’s just, you know, other than Yeah. Other than the It keeps the prices down. Uh certainly, but the logo of the truck may differ each time. You might you might get somebody delivering something in a car and it might be a van. It might be uh you know,

30:28

something else. Uh you know, it might be a full-on truck. It depends. I mean, there are organizations that optimize that. I remember um uh a conversation with a guy that has a last mile logistics company and they worked for Lowe’s and uh somebody bought a bathtub

30:45

and they were wondering what it would be like if if Lowe’s had a uh versus being a Lowe’s truck. It was coming in a minivan. They were getting a they were getting a tub in the minivan. Well, they dropped the cost of delivering by like 70% because it was a minivan versus a

31:01

box truck. Um those those are the kind of things we’re talking about where you just optimize what is the best approach, what is the best load factor uh for that item as a result of that. So >> minivan, you know, it never would occur to me like why how can a minivan and a

31:15

box truck be competing for the same. >> Exactly. Yeah. Exactly. So a washerd dryer type of thing, who knows that pickup truck with a washerdryer on it might be an actual delivery instead of a box truck that’s uh >> labeled. What what does it impact for

31:31

you? Looking at all these impacts, how does it affect what you might recommend to someone? I know it’s not your business, what retail stocks you might want to buy, right? When we’re at all-time highs here, it’s hard to put money in at these valuations, but I’m

31:45

sure there are some stocks that you’d say, I’d still buy them here, but plenty of others that you’d say I wouldn’t. >> Right. Right. Uh, so we have a a a retail uh AI retail readiness index report that’s free out on our website right now. um where we literally uh

32:00

looked at every public retailer um in the US other than we took out the sea stores and we took out the uh the cellular stores because revenues are all mixed up and it’s hard to evaluate but u of of the rest there we ranked them and we ranked them on on AI readiness and

32:19

then we ranked them on the impact of the income statement. So we looked at increased sales opportunities, lowering cost of goods opportunity, and then knowledge work. Uh where you see most of the generative AI gains. Um generally that where everybody gets excited. It’s

32:33

like oh chat GPT at Walmart is going to be huge. Well, it’s only 29% of of the P&L. So it can have a big impact, but it’s a small piece of it. If you impact sales by increasing sales, now you’re hitting the full 100%. So this rank rank orders those folks. So

32:52

the obvious ones are the Amazons and the Walmarts at the top level. They are far and away ahead of others in terms of readiness. Um the one that’s most intriguing to me from a stock standpoint is Starbucks. That’s down right now. >> Oh really? >> Um because Brian Nickel, we rated

33:10

Chipotle the highest on AI readiness in restaurants and Brian went to Starbucks which was the company who had the most to benefit from AI use there. one person make that much of a difference cuz I feel like >> in a position like that from a leadership position. Yes.

33:26

>> Cuz he’s come in and and you’ll see some of these changes that they’re making at the store. You think well you and I could have come up with those ideas too. There’s a lot of smart people anyone come up with this idea. >> Yeah. Which really tells you how

33:39

inefficient some of that process was uh previously. um you know so but there is most of AI is done in the back end it’s the preparation it’s like you know when you equate it to a football game you only see what happens on Saturday you don’t see all the prep the preparedness

33:58

that goes into it >> it’s the data it’s the underlying fundamental data where these guys um there’s a group of of retailers that in 2018 when Walmart I’m sorry when Amazon broke out for the first time their retail revenues and showed profitability after 20 years of being unprofitable.

34:18

A light switch turned on a lot of retailers and said up to that time uh it spend growth was equivalent to revenue growth. they were they were tightly coupled right these guys decoupled that >> and these guys said we are in a technology race um here and so it’s kind

34:37

of what you see in the news right now between us and China you know from an AI perspective and they started looking at what do we need to spend in IT if we need to transform ourselves to compete with a profitable Amazon and so they started cleaning things and starting the

34:54

race and so now Um, I like to equate it to uh where people in retail there are there are retailers like uh Tractor Supply and Walmart and Target and Kroger and some others and you know um and Chipotle and others if the goal is to get to LaGuardia airport from downtown

35:15

New York. Okay, so I say Time Square, these guys are already through the tunnel, they through the toll booth, they’re on the expressway because they started that hard work in 2018, 2019. Um the rest of people are stuck on 42nd and between 8th and 9th on bumper-to-bumper traffic.

35:33

>> It’s crowded. >> Yeah. Yeah. Yeah. And they’re just like they get going and realize data can’t support this. Boom. you know, um, and that’s why it’s fascinating. And, and right now the news, I think the thing we’ve got to factor in is there’s two

35:49

sets of news. There’s the the news that hypes because it’s stockreated, and there’s the fear-based news that’s putting out negative stories and flipping negative stories about AI because they’re afraid of the impact on them personally. Um there was a uh there

36:08

was a study that was quoted by everybody about um MIT study that came out that and said but the headline came out in Fortune magazine that said I think it was Fortune or Forbes came out said 95% of AI projects fail. >> Yeah. Yeah. We were talking

36:26

>> that’s not what the study if you go back to the underlying it’s not the study that MIT did. They studied 52 companies and they had a six-month period to evaluate it and success was a million dollar plus uh return on investment in six months. There are very few projects in the

36:47

retail >> that get a million dollars plus. >> Yeah. >> In that time and and this was cross industry. But it wasn’t it wasn’t so much that they saw literally it was 25% had reached that ROI in six months. Um, and this was using a general purpose thing like chat GPT for for pricing per

37:07

se, okay? Or or a project is something they were doing. That’s not you don’t get those kind of results in that kind of environment that you can scale very often. You’ve got to have some AI machine learning stuff where you’ve got to get the answers exactly right where

37:24

you get that scaling. And that is really hard to do in six months, but it got picked up by everybody. Yeah. and they’re poo pooing it today and you know out there and everybody says, “Oh, this is evidence of the bubble.” And there’s just more evidence of a bubble and and

37:38

yet you talk to every one of these guys and they said, “We’ve got more functions. We’re just limited by compute. We just don’t have enough chips. We we don’t have enough data.” Um there. So, um yeah, I’m I’m I’m really really bullish because I’ve seen it. I’ve seen

37:54

the impact on our own on our own. talk about, you know, real quick because I I want to move to a different category, but talk about, you said Starbucks had the most to gain from AI while Chipotle was the most AI ready. So, what what does that mean? What do those

38:10

differences mean and what is Brian Nichols switch from Chipotle to Starbucks going to mean about that going forward? >> Yeah, I think it’s uh you know, with Starbucks more than anything else, it’s focusing on their loyalty plan, upsale opportunities, and speed of service. If

38:26

you ask, you know, everybody’s, you know, they’re concerned about the cost. Yeah, they complain about the cost of the cup of coffee, but more than anything else. When I show up, I just want it to be ready so I can get it and get out. Um, I don’t want to stand in

38:37

that huge line uh there. and the ability to speed up that production process to simplify the menu um that they’re doing to to automate more of that process uh there to um have in advance and almost almost pre- ready make sure you you know who your people are that are coming

38:57

every day. It’s not that you pre-make the drink, but you’re prepared for that crush of volume that is coming because you have the data and the analytics behind that that that Eric’s going to show up at this time to get his coffee a certain way and making sure you’re

39:12

you’re staffed properly. But if you can automate that that make I mean what does the barista actually add to the coffee experience? It’s a relationship thing. you know, they may be able to make the the design on top a little bit better, but brewing the coffee, you know, mixing

39:32

the ingredients, all of that stuff can be completely automated. And uh and you can do that. I mean, you look at McDonald’s, for instance, as an example. When McDonald’s went to the automated drink machines there, it didn’t change the quality of the drink. It didn’t

39:48

change. It just made everything much much faster. And so, he can bring that kind of thing. that’s the most visible, but it’s the backend backend things that are um in place. And and that’s not to say that Starbucks was wrong. One of the one of the studies we did that was

40:03

fascinating is we were looking at cash management and uh Starbucks has a local bank account for every one of their stores. They don’t use the large banks. >> That’s weird. >> The reason why is the uh cash transit piece of it. When you have a you’re

40:24

working with a local bank, everything’s included. You don’t pay any extra. You need change, no problem. We give it to you, etc. But the large banks, if you do like Bank of America or those guys, you need change, that’s going to cost you. >> Oh, I see.

40:38

>> You want to have an armored truck come to, you know, pick it up, we’re going to charge you for that. So, it’s saving them like thousands of dollars per store because they are doing the uh the local uh local account. Well, on the back end, that’s a headache because now you’re not

40:55

dealing with one bank, you’re dealing with thousands of banks and uh so automating that piece of it and doing a better job of making those connections, making sure those are secure, dealing with that on the back end for them is an opportunity for them on the back end.

41:10

>> So, how do you figure out when you’re ranking? How do you figure out, oh, they have these opportunities to improve? How how do you quantify people standing in line? How do you quantify the impact of all these banks? Well, believe it or not, it was that shift that we talked about

41:23

in the beginning, the different uh phases of the business. Um, years ago when we started creating our Sophia data service, we started tracking what companies used and so we built an algorithm based on it was publicly available data in most cases. Uh, what

41:38

do they use? Um, so we had things like data maturity in there, business analytics maturity, uh, size of the company. One of the things when it gets to scaling of of these guys is you can’t get away from size. The same change in a 10 store company versus a Walmart,

41:56

the scale of difference if you get a 10% improvement, >> right? >> It just goes up. So, we’ve got that. We’ve got profitability in there. We’ve got um are they aligned with the key players that are doing AI things in their platforms and were leading like or

42:12

were they aligned with a Salesforce or an Oracle those kind of folks with these systems? Um were they doing things like computerated ordering for food uh type things like a company like Relex uh and and Upshop? Were they doing those things and was that expanding into vendor

42:31

managed inventory? uh those are things that were in the algorithm that spit out the the rating for them and then it was just a matter of of where was that benefit in the uh in the uh income statement and then scale and that’s how it all came about.

42:47

>> I see. And then so you know you talked about how much AI has changed your job as an analyst and I know we’ve talked about that offline >> in terms of what it can do with analysis and writing and editing and you know charts and you know deep data like like

43:02

what what’s going to happen to analysts going forward if AI can do so much we hear about a lot of junior people coming out of college they can’t get a software job they can’t get an analyst job like I don’t need to hire you I already got I have this for very few dollars you know

43:16

>> and that is the big crisis right now we are doing with new hires what we did to manufacturing 25 years ago um and that we’re paying the price for now and trying to bring everything back and that’s the genesis of all these tariffs um we are gutting that early experience you

43:36

see it on uh not only um in uh in New York with the investment banks uh happening there but you’re seeing it in all facets because what AI does off the shelf is books smarts and and so I think what AI is going to do for analyst is it’s going to democratize um basic

43:58

analysis but looking at the anomalies of the data and building in the experience of people that have been doing it for years that are using AI tools to augment what they’re doing the the real nuggets the real insights so it’s uh that’s where the real benefit is so to to use a

44:17

metaphor. It’s like um you know, Bill Belichick type type defense or like Matt Patricia is doing at Ohio State right now with their with their defense. It’s at a whole other level than all the competitors are because they’ve got this knowledge and these tools that they’re

44:33

leveraging into things rather than just be the basic stuff. The AI is going to give you the basic stuff. The AI is going to overwhelm you with the basic stuff. Um, and uh, your job becomes editing it down to find the biggest nuggets. The AI will look at it and say,

44:50

“Well, this is the highest number. That’s the most important thing.” Well, no, it’s not because you don’t know the nuance of of of what’s going on. Um, and those are the those are the things there. So, you become as an analyst, you become more of an an editor and getting

45:09

the the data into the best bite-sized chunks for your audience. Um, because at the end of the day with our data is I’ve got to be better than 80% correct so that person can make a decision um for what they’re doing. We’re on our side, we’re not curing cancer. uh we’re trying

45:29

to get we want to get them to a confidence rating where they know hey I make this move I’m confident that this is going to work because I’ve got the data to back it up and uh the speed at which we can get to that is much faster. So I used to print out I mean literally

45:45

a year ago I was printing out a book that was about I don’t know 2 in 2 in thick in 8 point font going through things looking for deltas and colored charts to find out okay what is it then I got to think through what is that story well I built I built vibe coded a tool that

46:05

does all that work in five minutes and then now I’ve got to look at and say instead of building from zero to 25 good stories from the data. How do I take the 800 that are here and find the most the 125 that are best? And so that process of speed enhancement uh makes it so much

46:25

better because you get uh you get a lot more data to choose from but now you’re sorting through what are the things that really move the needle which are the things that uh you know I’ll give you an example. If we said it, they may come back and say, you know what, the people

46:42

that increase their sales the most are are building the most stores. Yeah, of course they are. So what? You know, um but then you like you said, you go, >> but that doesn’t tell me should I build a store or not? Like >> Exactly. But if you went if you went in

46:58

there and and you looked at it and you said, “Well, look, the people that are using computer vision are eight times more likely to have grown their sales 10% or more.” There’s some causality and correlation there that’s really really strong. And it’s it’s if you have all of that and all

47:16

you’re doing is sorting through that now rather than having to find it in the first place and build it up, the speed improvements and the productivity improvements are phenomenal. >> Yeah, that that is fascinating because as you talk about that entry level

47:31

worker and what AI can do, it makes you wonder how would you advise someone who’s got kids, right? Let’s say someone like me, what should they be studying, right? Like you talk kids, grandkids, what should they be studying for the next 5, 10, 15 years so that when

47:45

they’re 22, someone doesn’t want to hire them. >> So, you know, college likes to call it critical thinking, but at the end of the day, it’s judgment, good judgment. How to how to read things and judge for a response. Um, your kids are young enough that I think we’ll get through this

48:02

phase. We will be at another phase. I’ve got two looking for jobs and just graduated right now. Oh, they’re right now like 22. They just >> 22 and 24. >> Yeah. >> Yeah. And uh two years I told my I said I pretty much talked my son into getting an MBA because I thought that would be

48:19

the best thing for him. And it may be long term, but it’s actually a hindrance right now getting a job in his field because he’s overqualified for the jobs that are open in his field, which is airport operations. And uh and some of the people look at as an MBA. Oh, he

48:36

doesn’t want to stay here, so we don’t want to hire him. >> Right. We don’t want to hire someone who’s he should just leave the MBA off his resume then. Just >> Yeah. Which is how how crazy is that? >> Yeah. >> Um there’s a there’s a I think it’s Sengage did a study that is really

48:49

really good that looks at it, but only 30% of new grads are now working in the field of which they graduated. >> Crazy. >> Um >> it’s crazy. And so, but at the same time, we’ve got this bubble where we’ve 800% increase in college expenses. So, if I was somebody I would, the advice

49:07

would be different. If I was in high school, I would go into trades, electrician in particular, uh electrician, plumbing, those sort of things. A desperate need uh for that. um if if I was younger and this is going to play this is going to play out and those

49:22

jobs are being are being created but the unfortunately the jobs are being eliminated faster than the jobs are being created and if I was anybody uh right now I said look at being an entrepreneur because that entrepreneurial experience can be used anywhere and uh building building a

49:40

business whether it’s building an app and just going with that or or just having a service and I would look at focusing on those higher income folks, if you were doing it as a manual service, what what do the higher income folks want to spend money on? It’s

49:54

anything that reduces time. What can you offer them that reduces the amount of time they have to spend on things and then build a business around that? >> That’ll be that may include college or may not. >> Yeah, that’ll be that’ll be a whole other podcast as we think now with the

50:09

little ones like should we even be saving for college? Like how if imagine what it’s going to be 15 years from now? I mean it’s it so but you’re right I don’t know it’s going to be a different phase altogether. >> Yeah. >> Fascinating Greg. I I appreciate the

50:23

time is super interesting. >> It’s really cool what you’re doing. It’s cool to hear what >> an analyst in in really looking at you said bottoms up when you’re looking at every specific that bark level analysis not the force level analysis. Yeah. You I could talk to you all day. I don’t

50:37

know if our audience can we we’re 50 minutes in so maybe we leave it there for the audience. Uh Greg thanks so much for joining us. >> No my pleasure. Just remember, culture is the big thing when it comes to all this stuff. You have a good culture, you

50:49

can be really successful with this. >> And if you have a bad culture, you’re probably going to fail. >> Yeah, exactly. >> Yeah. >> Yeah. So, >> see you, Greg. >> All right. Take care. >> Thank you to my guest and thanks for listening. Subscribe to get the latest

51:03

episodes each week and we’ll see you next time.