Daniel Swan, Senior Partner, McKinsey & Company, Zach Thomann, Chief Operating Officer, Americas & APAC, GXO Logistics Moderator: Ruth Umoh, Next To Lead Editor, Fortune; Co-chair, Fortune COO Summit
Category
🤖
TechTranscript
00:00Let's kick off with AI, because we've barely discussed that this entire conference.
00:05As it pertains to supply chain management logistics, where do you see the strongest
00:09use cases and potential to really drive value, reduce cost operations?
00:15Let's kick things off with you.
00:16So, great question, didn't see that one coming.
00:18So, what I would say from an AI perspective is that from a supply chain
00:22perspective, thus far the focus has really been on transportation.
00:26So, as we think about supply chain, I generally think our customers,
00:30as well as what the industry's focused on,
00:32is really how do you deploy it in labor, transportation, as well as inventory.
00:36So, what we're seeing a lot in terms of what GXO is going to market with and
00:40helping our customers with is specifically how do we help them with inventory planning?
00:43Where should the inventory be?
00:45How do you get closer to your customer?
00:46And how do you dynamically think about re-slotting that and
00:49making it available from an omni-channel perspective for all of that?
00:53Selfishly, as a contract logistics provider,
00:55we're hyper-focused on AI as it relates to how we do labor planning.
00:59So, what type of predictive analytics can we use for demand planning
01:03to augment the forecast that we get from our customers and
01:05make us more efficient inside the facilities?
01:09Yeah, I mean, I would say a couple things that we see a lot.
01:12One is just the end-to-end planning processes, right?
01:14There's so many things that people touch today.
01:17And unfortunately, we do the work when we go to clients, and
01:20oftentimes we see that humans touching it actually make the answer worse, right?
01:26And so, nothing more frustrating than spending a lot of time and
01:29effort to actually make the answer worse and less productive.
01:32So I think that's one big thing.
01:34I think the second thing we see is, just on the labor side,
01:37there's lots of places where there's interactions in a warehouse with
01:41touch points of where do you slot things, what order do you pick, etc.
01:46And just, there's so
01:47many decisions that are made across the end-to-end operations where, just quite
01:51frankly, a machine can make that decision better than a human,
01:54given all the variables that go into it.
01:55Give me a for instance.
01:57So for instance, when you think about an order coming in,
01:59what's the sequence in how you pick those things, right?
02:02It used to be you'd put things in a rack and
02:05it would go, you'd go down the aisle in a specific order.
02:08Now it's dynamically put away, right, based on what are the orders you have coming in.
02:13And there's no reason to say, hey, this is the slot for this product every time.
02:17You can put that product in the next slot the next time, and
02:20the next slot after that, based on the orders you actually have coming in.
02:24There's no way the human brain can kind of keep up with that.
02:27I had separate prep sessions with the both of you, and
02:30you both really zeroed in on taking a problems first approach to leverage in AI.
02:35And you noted that most operations chiefs kind of do the inverse.
02:40So talk to me about what you mean by that, this problems first approach.
02:44Sure, so from my perspective, I think earlier in the session,
02:47someone said they're an optimist about AI.
02:49I'm actually a pessimist.
02:50I think all COOs are inherently skeptical about how things generate ROI.
02:55So what I would recommend, and really from how we look at technologies,
02:58be it AI or even automation, what we try and start with is what's the problem statement?
03:02So what are we trying to solve?
03:03How do we get an output metric that makes sense, and
03:05then really back into that use case?
03:08And really what I wanna make sure is that when we think about AI or
03:10any technology that we use, that it's not a solution in search of a problem, but
03:14actually a problem statement that we're trying to solve.
03:17And that's something that resonates with our board and
03:19certainly with our CFO in terms of how we deploy capital.
03:21So I tend to take a very pragmatic approach to how we leverage technology.
03:26So slotting's a great example because it's something you can look and
03:28point specifically in the warehouse and say,
03:30how do we get the most efficient pick path inside the facility?
03:34Least amount of footsteps is kind of a clear ROIC.
03:38I think demand planning is something that is inherent from an AI perspective.
03:41That's something you can fully outsource from that perspective to an AI
03:45predictive model.
03:45So it can tell you when you should expect ebbs and
03:48flows, cuz really operation's all about level loading a facility.
03:51So those are things that have a real tangible outcome for us.
03:55And when you start getting into fuzzier activities,
03:57you tend to see less of a business case for it, and
03:59more of an exploratory R&D approach.
04:01In which case, you have to be mindful about what you're doing and
04:04what the intended outcome is for that.
04:07Your thoughts?
04:07I mean, I walk into client meetings all the time as a consultant, and
04:11people say, tell me what should I be doing on AI, or
04:13tell me what should I be doing on generative AI, and I cringe.
04:16And the reason I cringe is because I think the way you should look at it is,
04:20how do I solve my inventory problem, or how do I solve my planning problem, or
04:24my transportation problem?
04:26And by the way, all those problems you should solve with a combination of AI,
04:31generative AI, automation, good old fashioned process improvement,
04:36capability building of your employees.
04:38And so, if you look at it like that, I think you have a much better chance of
04:42actually solving the problem and having impact,
04:44versus trying to say across the entire chessboard,
04:47where can you use generative AI or AI?
04:49Yeah, when we talk about supply chain, we use kind of these nebulous phrases,
04:53like supply chain resiliency, future-proofing the supply chain.
04:57What does that really mean in practice, and how do you determine risk points?
05:01Yeah, I mean, I'll give you a great example.
05:03When we went into the pandemic,
05:05fewer than 2% of companies had visibility into their N-2 supply base, right?
05:11And I was on with a client two weeks after the pandemic started, and
05:14they had just gotten out of a meeting from the board as the CPO.
05:17And he said, we're in great shape.
05:19I know we're all our suppliers, and they all have plenty of capacity.
05:23We're in good shape.
05:25Two weeks later, I got a call that you could tell when I answered,
05:29he wasn't feeling like he was in such good shape.
05:31And what he figured out is they're kind of,
05:33this is a company that makes consumer products.
05:36All their tier three flex packaging was in the same region in China, and
05:41they didn't realize it until it started flowing through.
05:44And the reality is, resiliency, as we think about supply chain resiliency,
05:47is all about understanding where are your risk points and
05:50having more than one option for how you solve those risk points.
05:54Another great example was everybody figured out in the first part of 2022 that
05:58more than a third of the world's sunflower oil came from the Ukraine.
06:03I didn't know that beforehand, I'm not sure anybody did.
06:06But so many people that look at the Bill of Materials and
06:08say sunflower oil was the 214th most important thing on the Bill of Materials,
06:14all of a sudden it became the most important thing.
06:16And people just need to think very differently about where their risk points
06:19are and where they solve those.
06:21And it's not the traditional, here's my biggest components, and
06:25therefore, I need to solve those first.
06:27But oftentimes, those are the easiest to solve.
06:28But hindsight's 20-20.
06:29I mean, it seems as though you have a million and one risk points for
06:31any company.
06:32So where do you really zero in initially?
06:34Yeah, I mean, actually, hindsight is 20-20.
06:37It's actually very analytically possible to go out there and say,
06:41if I look at my supply base,
06:43how am I making decisions about which suppliers am I using?
06:46And am I single sourcing or am I dual sourcing?
06:49Is my manufacturing capacity closer to my customers or
06:53further away from my customers?
06:54All these decisions either introduce more or less risk into your supply chain.
06:58And they talked about regionalization.
07:01Globalism isn't dead, right?
07:03But there is a lot more regionalization going on.
07:06And people thinking about having more than one option for
07:08getting any specific product to their customer is a really big deal.
07:11I'm gonna circle back on localization.
07:13First, I wanna hear your thoughts.
07:14So I always joke around with my team and say the F word in fulfillment is
07:18flexibility.
07:19So you think about resiliency is how do you rethink what the whole supply chain
07:23is, say, what are my offsheets and contingency plans?
07:25So obviously, taking a data-centric approach,
07:28you can see where you have opportunities.
07:30But moreover, I'd say, what are you gonna do in case of?
07:32So what are the most important tabletop exercises that you can do inside your
07:36supply chain?
07:37So to the extent there's geopolitical risk with China, and
07:40you start thinking about what does that do for maybe a 321 program?
07:43How do you react to that, and how do you build in a resiliency for that?
07:47The other thing I think a lot about is how you place inventory.
07:49And we've seen over really since the pandemic,
07:52a push to omni-channel fulfillment.
07:54So not just B2B, but B2B to C transactions are really becoming paramount.
07:59And when you think about your inventory footprint,
08:01you really need to think about how do I turn that into an omni-channel footprint?
08:04So that it's not just a purpose-built facility for
08:06a B2B transaction or a case pick, but how do I do each picks?
08:10And then frankly, how do I leverage a transportation network that can leverage
08:13all types of different carrier options, so not just be holding to one?
08:17I see that as some of the table stakes that you really have to do to build in
08:20the contingency, and then you have to take a next cut to say,
08:23what's under the covers that we don't see?
08:25And frankly, that's where you take an analytical approach to it.
08:27How far in advance are you doing that risk assessment?
08:30What's the impact in the immediate short term?
08:32Is it three years?
08:33Is it five years down the line?
08:35I think there's some things you can do today in your supply chain.
08:38So it's really just how are you sourcing it?
08:40So take transportation, for example.
08:43If you go back to the pandemic and really think about it through an e-commerce lens,
08:47you saw a lot of brands actually go to market with a single national carrier.
08:51So they went with a FedEx or a UPS or what have you.
08:54And then what happened in the pandemic is capacity was constrained and
08:57they couldn't move their product.
08:58So what we're doing today and you see people doing is just one, for instance,
09:01is how do I have a disparate number of carriers so that I can level load into
09:06the appropriate carrier and de-risk the total supply chain?
09:10And frankly, I can de-risk things like surcharges at peak and
09:12things like that to actually have a more balanced and nuanced supply chain.
09:16That's just one example, but I take the same approach against all the different
09:19pinch points and say, where do you have redundancy and
09:22optionality when events unfold?
09:24Anything to add to that?
09:25I'm going to ask one more question, then I'll turn to the audience.
09:28So start thinking about that.
09:30Zach, you told me you're seeing brands increasingly outsourcing their
09:34supply chain management.
09:35Why is that?
09:36So it's something that I think has been a trend for a while.
09:39So as you think about sticking to your knitting and
09:42what we're trying to accomplish, supply chain's hard.
09:44So anybody that's in this space will recognize there's a lot of unforeseen
09:48challenges.
09:48There's strikes at ports.
09:50There's all types of things that cause disruption.
09:53And frankly, when you think about investment and
09:55how a brand should think about their supply chain, there's inherent things that
09:59you need to be mindful about over a long-term horizon.
10:01So what are the choice investments?
10:03What's the network need to look like?
10:05And how do I admin flow that network for today and for tomorrow?
10:09So if you take the pandemic lens to now, what we saw really was in 2020,
10:14there was constrained capacity in the US.
10:15So what a lot of brands did is they went out and took down space or
10:19started to outsource specifically to fill that capacity gap.
10:23And then what we're seeing now is that inventories are slowly coming back down
10:27because you're seeing more of a normalized flow into the US.
10:29There's excess capacity for a lot of these brands.
10:32So you think about outsourcing, outsourcing actually solves that problem.
10:35So you can go to a provider like GXL or one of our competitors, and
10:40you can take advantage of their network and say,
10:41how do I actually buy meaningful capacity in spots where I don't want to necessarily
10:46have long-term investment, or I can go build and grow through a growth curve?
10:51That's really where outsourcing is intended.
10:53Take advantage of the leverage that a company like GXL creates and
10:57actually reap the benefits for it as we go forward, so.
10:59The only thing I was going to ask, there's like a 25-year trend of the frequency and
11:04magnitude of supply chain disruptions increasing, right?
11:06People want to talk about the pandemic, but
11:08in the last two weeks we've had a port strike and two hurricanes, right?
11:12When you look at that, what we see is building resilience in your supply chain
11:16is both sometimes expensive and oftentimes not a capability that a lot of companies have.
11:22So you either got to build that muscle yourself or
11:24you got to work with people that will help you build that muscle.
11:27Absolutely.
11:27Question from the audience, raise your hand and
11:29we'll have a mic handler come to you, anyone?
11:34All right, I'll turn back to the audience again momentarily.
11:38As we've seen time and time again, geopolitical tension and
11:41fragmentation can severely disrupt supply chains.
11:44There's been a lot of talk about decoupling.
11:47I mean, how feasible is it really and
11:49how are you guys thinking about it from an operations standpoint,
11:51from a supply chain standpoint?
11:53I mean, I'm happy to start.
11:54We've been doing an annual supply chain survey for
11:57the last five years since the beginning of the pandemic.
11:59And basically, there's been three things that people have done to try and
12:02build more resiliency.
12:03One is add inventory, which is kind of the knee-jerk reaction.
12:06The second is to try and get multiple sources of supply, right?
12:09Multiple suppliers, etc.
12:11And the third is to try and do more regionalization of their network.
12:14I think a few things with the regionalization.
12:16First of all, it takes a lot longer, right?
12:18You can add inventory next week.
12:19It takes months and years to build facilities.
12:23The other thing I think we see a lot is just because you move a manufacturing
12:26facility, it actually doesn't make your supply chain that much more resilient if
12:29all your suppliers are still based back in Asia.
12:33And so what we've seen is people really not,
12:35it's great to think about where your distribution network is and
12:38where your manufacturing facilities are.
12:40What's actually even more important is thinking about the ecosystem of your
12:42entire supply chain.
12:43Where are your suppliers based?
12:45Where is that product coming from?
12:46And where does that risk look like?
12:48And so what we've seen is a very slow movement.
12:50Because somebody moves, and
12:52then all of a sudden they start putting people in business around them.
12:55Whether it's suppliers from back in Asia opening plants or
12:59facilities here, or it's new suppliers being put into business.
13:01So I think it's a trend that we're seeing, but it's still very early days,
13:06just because of the lead time and
13:08the fact that there's multiple steps to do this properly.
13:11I couldn't agree more.
13:12I think the lead time for
13:13actually de-risking the upstream supply chain is a multi-year process.
13:17And it's something where you have to really think to what the industry's
13:21moving towards.
13:22So what's the risk associated with Asia Pacific?
13:24Where are the future industrial economies that you should be investing in?
13:28And frankly, you're starting to see a lot of that.
13:30I've seen a lot more investment in Mexico in terms of de-risking that,
13:33getting closer to the United States and de-risking the China impact.
13:36But those are multi-year projects.
13:37And frankly, the band-aids around that, I think, are just to your point.
13:41How do you do inventory?
13:42How do you go diversify your network with different carriers?
13:45Those are the action steps being taken as a bridge solution right now.
13:48Yeah, what are the pain points you're hearing from your clients,
13:50from your customers?
13:52I think the number one pain point I have from customers right now is
13:56demand planning.
13:57So there really is, you look at where we sit today with the economy,
14:00the uncertainty around the election results, and how that will impact sales.
14:04The number one question facing them is, how should I think about inventories?
14:07How should we build up the inventory?
14:08And what's our in-case event for really our Q4?
14:11So it's really here and now, what's the top line gonna do at this moment in time?
14:15And how can our supply chain react to that?
14:18Particularly given that this is one of the shortest holiday shopping seasons that
14:21you'll see with Thanksgiving to Cyber Monday being
14:25the shortest period you'll see in every seven years.
14:27So that's the top of mind question today.
14:29The larger question, I think, is really largely around how you de-risk
14:33the Asia-packed geopolitical tension.
14:35And what does that do from a supply chain perspective?
14:37I think those are the push and pull of the here and now versus the long-term
14:40strategy.
14:42Yeah, I mean, I'm 100% with the demand planning.
14:44The two things I might add is one is,
14:46I think the productivity imperative has never been higher.
14:48People- What do you mean by that?
14:49Just people talk about, well, inflation is moderating to 2 or 3%.
14:53Well, maybe, but that's on top of 7% and 7%, so costs are still going up, right?
14:58And so if you look at the historical model of a lot of companies trying to drive
15:024% gross productivity, and then you have 2% inflation, so
15:06you kind of have 2% net improvement.
15:08When you have multiple years of 5% to 7% to 10% inflation,
15:11you're upside down on that.
15:13And so how you actually really drive productivity, and that's where things like
15:16automation and labor upskilling and all that stuff come in really important.
15:21And the second, in addition to kind of productivity,
15:25is then how people maintain the importance of resilience.
15:29So I think supply chain resilience was a board topic through the pandemic.
15:34And I think a lot of supply chain leaders are worried that it's not going to stay
15:37there, right?
15:38And it's going to go back to the good old days of it's great,
15:40you can build flexibility for x.
15:42But if you can make it without flexibility for x-30%, that's the answer I want.
15:47And so building that muscle in your organization around how you actually get
15:52people to understand the risks you have, like you asked about earlier, and
15:55how much that costs you, is a really important muscle, we think, for
15:59companies to build so that we don't get in this situation again.
16:02Let's go back to the AI and automation talk.
16:04How are you upskilling and reskilling your workers, especially in warehouses?
16:09So if you look at really what you see inside the facilities,
16:11it's really the tools that you deploy in conjunction with the workers.
16:15So I kind of look at the lens of large-scale automation and
16:19more pragmatic automation, and I'll put AI in that same category.
16:22So you have a purpose-built project to say,
16:24let's go do grounds up greenfield for automation.
16:27And that's a heavy investment in things that take multi-years of planning, and
16:30frankly, one of the reasons that you'll see a number of companies work with
16:34outsourced logistics providers.
16:36But even after that, you think about a CI focus.
16:38And back to your point about productivity,
16:40how do you iteratively add value into the network?
16:43So you're gonna see what I'd call adaptive technology inside the facility.
16:47So how do you leverage co-bots and things like that with the worker?
16:50So what introduces into the existing environment?
16:54And frankly, that's where we're also seeing tools from AI that are outcome-based.
16:57So how do I augment for the demand planning?
16:59How do I help them with the slotting tools?
17:01How do I help them with labor planning inside the day in, day out,
17:03to really marry all three of those things together?
17:06That's where you're seeing tools that are really driven for that in the facility.
17:10For instance, is obviously slotting, but moreover,
17:13the top of mind thing that comes to me is really just the way of
17:17management planning inside of a facility.
17:19If you think about how we push work through a building,
17:22the team has to really sit down there and say, when do you release work to a floor?
17:25Who's gonna do it?
17:25What's the optimal pick path?
17:27Those are things that are right for AI and where we're seeing specific use cases
17:30that we can use as a tool for the individuals inside of a facility.
17:34Are you tracking the cost reduction?
17:36Any numbers?
17:36Absolutely.
17:37Yeah, what are you seeing?
17:38What's that?
17:39Of course, you can share.
17:40I haven't shared anything publicly yet.
17:42So typically, we'll have a number of engagements with our clients,
17:45really track CI and how we're driving change,
17:47particularly on contract by contract basis.
17:49But certainly, as a company, we're making large bets on CI for
17:53the future in terms of driving bottom line for us.
17:56Maybe, even though the manufacturing sector has been relatively flat
18:00the last few months, there's 500,000 open manufacturing jobs in the US right now.
18:05And if you believe all the investment that's coming from the Chips Act,
18:09IRA, infrastructure bill, that's gonna grow to somewhere between a million and
18:14a half and two and a half million open manufacturing jobs over the next five
18:17years.
18:18Over the last couple of years,
18:19we've been running between zero and 1% productivity on average.
18:22We need to do 5% per year productivity improvement
18:26to kind of close that gap with productivity alone.
18:29That's a startling number in the gap.
18:31And so my personal view is that in the pre-pandemic world,
18:35automation and AI were four letter words, right?
18:38And I think now, oftentimes,
18:40that's what's actually required to kind of fill the labor force that you have.
18:44And I think what we've seen through our research is actually a lot of the things
18:47that you can automate, whether it's a physical job or a back office job,
18:50are the parts of the job that the employee actually likes the least.
18:53So if you can do it in a way where you're selling the employee on the opportunity
18:58to upscale, and they actually become more marketable when they have these skills,
19:02it actually can be a win-win.
19:05It just, can you get that framing right?
19:07Absolutely.
19:08All right, rapid fire, key takeaway, key takeaway.
19:12For me, it's about getting specific, right?
19:14What are the two or three problems that you're trying to solve that AI and
19:17automation can really help move the needle on?
19:20And my plea to everybody is continue to make supply chain a board level topic.
19:25I think it's really important.
19:27I would say key takeaways are really just focus on the themes.
19:30So skate towards the puck.
19:31So think about automation as it relates to labor.
19:34I would say how do you use technology as a pragmatic tool?
19:37So outcome-based, outcome-based, outcome-based is something I'm a big believer in.
19:41And then certainly, the other areas I would focus on is really just how do you
19:44de-risk your supply chain?
19:45So I would focus on tabletop exercises.
19:48I know that sounds like an old school process, but really at the board level and
19:52throughout the organization, where are your risk points?
19:54Highlight them and actually plan contingencies around them.
19:57Well, there's certainly no dull day in your workflow.
19:59Thank you, gentlemen.
20:00This was great.