Supply chain transformation in the AI era

Transcript: Supply chain transformation in the AI era

I’m just going to start by saying that I feel a bit of a fraud being stood in a suit today. Those of you who I work with closely know I’m normally more comfy in a hoodie, normally on that side of the room coding away with my team, but actually the thing I really like to be doing is getting dressed like this.

I’m a bit of a supply chain nerd. I spend the majority of my time, luckily, at Peak being able to explore people’s supply chains, getting into the  nitty gritty of what’s going on and then exploring.

The slight downside of that, I get to wear the hard hats and the fantastic gear on the left hand side for the food safety. I then get the chance to talk to many businesses like yourself where they doll me up a bit like this and I look very, very uncomfortable right there. A little bit like now. So thank you.

We’ve only got 20 minutes and those of you who work in supply chain know we could probably talk for about two hours on supply chains and supply chain problems. So I’m not going to talk about a lot of stuff. I’m not going to mention all of the things that have been happening over the last few years. 

We had COVID. We had gas prices. We had, somewhere in there, there was a big ship stuck in the Suez Canal that caused chaos. 

This is the reality for those of us who work in supply chains. And there was a brief period where we were talking in the COVID pandemic when we’re all talking about this sort of nebulous idea of a “new normal”. That sort of horrible phrase. I think we all got completely sick of it, after a while. And for those of us who work in supply chain, this is the continuous normal. We are continuously challenged by things happening that we can’t control. We have to embrace that; that’s the thing I want to start with first.

There’s always going to be new challenges. There’s always going to be things happening. There’s always going to be things that have happened that we have to react to. And how can we use technology and AI to help us react faster? And get the outcomes that our businesses want and our customers need. There’s also some structural changes that we’re currently facing. For those who work in supply chain, many of us are now completely addicted to getting things delivered within less than 24 hours. And that’s now starting to move into those B2B relationships where customers are increasingly demanding.

Supply chains have reacted. The factories I go around now, they’re no longer messy places full of guys in hard hats and dust and all those things. There’s robots. There’s automated production lines. You know, these things are producing huge amounts of increased productivity, but it’s completely changing how supply chains are structured such that we can respond quickly.

With the way the global world is, we’re seeing a big amount of reshoring. So new factories being opened in the US, things are changing. Structurally, things are changing, and we have to change with that. So that’s a lot of words to just say things are always going to change. So how do we respond quicker, faster and better.

At the same time, this is not something I’m particularly familiar with, but I get the chance to work with lots of businesses. And there are other changes that are coming along.

Some of our previous panelists talked about the rise of Facebook, which wasn’t a thing. Demand planners are having to react to completely brand new types of demand.  I know there’s some people here from the concrete industry who probably don’t have people tweeting and TikTok-ing their products, causing big spikes of demand. Have sympathy for your fellows who are dealing with that in the CPG space right now.

How has business responded? The quickest answer is to throw technology at a problem, if we don’t have people, and supply chains have suffered from an acute lack of talent over the last several years, throwing technology at the problem is a good starting point. 

And what we’ve ended up with is a large number of point solutions generating huge amounts of data that’s ended up being siloed. So different parts of our business, not just in the supply chain, have ended up owning that data, which could be really useful for predictive purposes from our CRMs, to the production data coming off these machines we have. Joining that together has been a challenge that many businesses really struggled with and will continue to be while we keep throwing more and more different point solutions at the problem.

And what we have to remember is that what really matters is value. I very rarely talk about technology. I really appreciate one of the previous speakers talking about how it’s very easy for us to get really down in the weeds of AI, ML, what’s the difference between those two? What models are we talking about? Those of you, again, some of you know me well. I can talk forever on different machine learning models for forecasting. But at the end of the day, as a business, what we need is to generate value and we have to keep a few key things in mind.

At the end of the day, most supply chains are about moving products from one area to another. As quickly and efficiently and as cost effectively as possible. So how can we ensure we’re getting the right product in the right place, in the right quantity?

And that solved the problem. So we can go home now. We know what the problem is. We can solve it. OK? It’s never quite as simple as that, unfortunately.

You have to be aware of where the actual problem is coming from. If you boil most supply chains down it’s one very simple loop. You have planning and then you have action. So we make plans. We do forecasts. We build all those fancy models, we create them and then we take some action. We make some products. We buy some products. We put some products through a secondary process. We do whatever we need to do.

Unfortunately, as often in life, best made plans, reality starts to happen.

By the time that we get a chance to action our plans, they’re already out of date. The data is not correct. The suggestions are wrong. And then we really struggled. S&OP has existed in our businesses for decades, but still S&OP ends up being a monthly meeting where we all turn up just to talk to each other nicely or not so nicely. 

And actually, this is too simple because the real reality is all this stuff happens. And we go absolutely crazy because, so many things are happening so much that supply chain leaders are really struggling to cohesively respond to the challenge that they’re facing.

And this is all without even talking about the big picture stuff that I mentioned at the start. This is the day-to-day, and then imagine a COVID pandemic comes along and completely destroys your ability to respond. We believe that AI and machine learning can help close that gap.

By shortening that response time by bringing the planning and the actions much closer together, you can get a much more cohesive response. We can never fix the fact that, you know, a truck doesn’t turn up or a big ship gets stuck in the Suez Canal. But how can we more effectively respond when that does happen? How can we be more convinced in ourselves that we are making the right choice under uncertainty?

We’re never going to get rid of that uncertainty, but we can make more effective choices more quickly. I’ve had the pleasure of working with a few, and I’ve said hello to a few people and I got a smile from Andy from Marshalls there. At Peak, we’ve spent a long time learning from our customers as well. So we’ve now been able to embed a lot of that learning and understanding of how the best-in-class supply chains work and how we can bring that together into a product that helps all supply chains to be the best that they possibly can be.

So I’ve just got to check the time because I know I’m going to keep talking forever. How am I doing? Not too bad. That’s good. Oops. Sorry, Chris.

I should have stated at the start. Now that I live in America, I learned quite a few different things about how to communicate. I normally have to start all my talks in America with “if I say anything weird, it’s not because I’m a data scientist, it’s because I’m English, and so I say things a bit strange.”

I have to warn them this way around because I started saying inventory rather than inventory, which is what we say here as proud northern as most of us.

And to riff off another famous American, we need to be asking not what our supply chains can do for us. But what as business leaders can we do for our supply chains?

So, unfortunately, one of the other slightly negative things that came out of the pandemic was there was a brief shining moment when me and the other people in the room who were supply chain nerds were held up as the masters of industry because we got some toilet paper to you all, and that was a brief beautiful moment. But unfortunately now that moment has passed, the supply chain is seen as a cost center. You have to do more with less. You can’t get the people. You can’t do what you need to do.

So hopefully we can talk about some of the few things that I’ve seen as I’ve spoken to businesses, where there are easy, sort of actionable, items that can either help today or can prepare you for a future where you are taking more of those actions.

The first thing that seems to be a big thing that’s enabled by technology is a level of granular detail that’s never been previously possible. Every SKU is different. You know, at Peak, we really believe that every business is different. And then within your business, you know every SKU is different as well. It either has different customers, different production processes, different constraints, so many things. And many of these meetings, these S&OP meetings, get really bogged down in that detail because there’s always a reason why something is like it is, but that stops you seeing the big picture.

By allowing us to feed in those business guardrails and that understanding of a SKU by SKU basis, it allows us to take those individual actions down at that level that drives the value across the supply chain. So that’s the first one. Start thinking about granular detail.

Deal with a long tail. Many businesses over time naturally agglomerate more and more and more SKUs. Customers are more and more demanding. They want it in every color all of the time. And I know there’s a few sales people in the room, so apologies.

We also believe that we cannot possibly function as a business unless we have everything in stock of every color of every type, even though the data shows that we’ve never sold it.

That purple T-shirt that was going to be cool three years ago, is probably still sat in the warehouse or whatever. You have to embrace this though. Many businesses are set up rightly when you have a limited amount of time and attention to focus on your top selling SKUs.

Because that is if you’ve only got so many hours in a day, you need to make sure those SKUs are looked after because otherwise your business would have failed. However, that long tail is sat in that warehouse, not moving, not being thought about, probably being tripped over by the guy who has to look after the warehouse every day. And not being dealt with.

This is a hard conversation to have. It’s often thinking about what things do we need to have in stock? What is my strategic goal? Am I trying to be the business that’s known for having every color of everything, or are we specialists in a certain set of items?

This one’s always controversial, out with rules of thumb. Humans are great. I’m an AI guy. Humans are great. Not trying to replace all humans, not trying to destroy the world.

But we have tricks. When we’re dealt with complexity, we face that complexity by trying to simplify the world. And that means we quickly fall back on rules of thumb. Oh, we need for every SKU to have a certain level of stock weeks of cover regardless of what that SKU is. Or we try and categorize very simply. We go for ABC analysis for three categories for 10,000 different SKUs.

But at the core, you always know that there’s something sat in your core that’s saying that’s not quite right. Every SKU is not quite an A or a B, it should be somewhere in between.

But this takes a lot of change because customers are so used to, you know, everyone is so used to trying to simplify and put into these buckets. It won’t surprise you, I’m a bit of a weirdo. I have a book on ERP design from the 1980s, which is when ABC analysis was from the 1950s, and this was designed for a time before computers. This was designed for a time when you had to go into a filing cabinet to know what your sales records were. We can move so much faster now, but that takes a big mindset change.

And the final bit, just to talk a little bit about what business leaders can be doing who are not in the supply chain. Almost everything I’ve just said, if you’re a supply chain leader, you were nodding. I saw a few. There were a few vigorous nods on trying to get rid of a few things.

But supply chain lives within a wider ecosystem of the business. So what can we, as business leaders, more generally do to help our supply chains move forward into that new, bright future.

The first one is always quite difficult; understand the real limitations in your business. And again, this requires a little bit of unlearning a few things that maybe you’ve done the same way for 10, 20, 30, 40, 50, 60, 140 years for some of the businesses in the room. And really get down to the nitty gritty of what is my actual limitation?

Because we need to unlearn that to be able implement an adoption becomes a real critical thing when it comes to machine learning, you’ll hear us talking about it a lot. And it’s because we care about value, and as Kelly said earlier, if you don’t adopt it, it can never generate any value because it’s just going to sit on your laptop or sit in the cloud, not doing anything.

So take that step to understand what is going on in that ginormous Excel spreadsheet that one person seems to understand what’s going on in and why do they do it that way? And how can we get that out of that one person who is also a critical factor within our business without us realizing it. If that one person’s ill, they’re the only person who knows it. How are we going to function as a business?

I always like these things. I know there’s some IT leaders here as well, but for business leaders, one of the interesting things is… most business leaders can name their CFO, but they can’t name the person who’s their technology counterpart. And that’s particularly true in supply chains in my experience.

So you know who controls the money, but you don’t know who controls access to data that you need to make your decisions. You don’t know how they work, what their constraints are, what are their strategic priorities. But they do, often within businesses, particularly legacy businesses, they control access to that data. They control the resource that’s going to get you access to it. So find them, give them a hug. You know, say hello, welcome them, you know, bring them into your supply chain. Like, I love going around, factories and stuff. You’ve all got interesting businesses. Take those IT leaders along with you, find out how you both work together and how you can create a more effective solution for your business. And the idea I’ll just leave on is just focus on decisions.

It’s so easy to focus on technology, algorithms, you know, systems, all this kind of thing. And really it’s what is the decision that we’re having to take every single day? How many decisions are we having to take? Why are we taking those decisions?

There’s always layers, we can automate some of these. We can add to them. We can take the human in the loop to make an active decision. But understand what are the critical decisions in your business, and then how can we make them better using machine learning?

That’s it from me. I don’t know if I’ve been super fast or super slow or bang on. Thank you, Chris. And yeah, really pleased to meet you all, and anyone who wants to talk about forecasting algorithms, you’ll find me at lunch wandering around, enjoy the rest of the day, and thank you for your attention.