The missing data problem in supply chain design

Why does the supply chain that served you so well yesterday seem designed to end your company today? Don’t be surprised; that’s how nature works. Yesterday’s design is optimized for yesterday. Tomorrow it’s a whole new ball game.

That’s why, as a supply chain executive, you must update your user’s manual about the term “optimization.” It’s a moving target and must be because the future is always changing.

Adapt or die, baby.

You know this already, but in business, it means you can never get too comfortable. Optimize, improve, observe, adjust, and repeat. Survive, thrive, and then break it again. Anything else and you’re just a dinosaur waiting for the next asteroid. How do you secure a future for your dino babies and dino tribe?

What is supply chain design?

Supply chain design is the emerging discipline of constantly remaking your current supply chain (the one you have) into a better one (the one you need—all praise to Herbert Simon). This is done by pulling all the data together about your supply chain’s configuration and operations and creating a digital “model” of the current system. You can then test changes, small, medium, and big, and measure the key metrics: service, financial, and risk—and just about anything else you care to observe.

Design the future supply chain you’ll need and go make it happen.

This supply chain design paradigm is always about the future, which hasn’t happened yet. That means doing things you haven’t done before, operating in ways you haven’t operated, and using strategies and tactics you haven’t attempted before. And yes, sometimes you move the freakin’ warehouse location.

But if you’re doing something new…

• Where do you find the data?
• Where are the lane rates when you have no shipment history?
• What’s the capacity utilization on lines making mixes you’ve never put together before?

Since you are modeling the future, if the changes are truly new, you won’t have data or history to rely on. This is called the “missing data problem.”

I wouldn’t say I’ve been missing it, Bob.

Engineering and design always rely on assumptions. We do the best we can with what we’ve got to work with. You’d like to have great data for everything, but if you’re talking about the future—and you are, in design—you must accept the reality that you are ALWAYS going to be missing data.

Not having data about costs, demand, line rates, inventory levels, lead times—that’s not a temporary situation. It’s the heart of the craft. Missing data doesn’t mean you don’t know anything at all about the future. It means you must make a good estimate and you know it won’t be precise. But that doesn’t mean it’s inaccurate.

Supply chain design is a people business. I’d take the good estimate of a smart, informed supply chain practitioner over the precise and complete spreadsheet excretions of a miserable team of low-wage robotic data grinders any day of the week and twice on Sunday.

There is real skill and art in making good estimates to drop into the missing data gaps. The most skilled and experienced pros don’t just make estimates; they develop the sense of their own fallibility: “How accurate do I think this guesstimate is? Is it plus or minus 5% or 50%?” You do the best you can, but a sense of humility and self-knowledge is what separates the best from the rest.

As a side note, that’s why you should mistrust the know-it-all jargon-filled self-important morons that derive their confidence from their roles as “experts.” Real experts don’t tell you to trust them because they are experts. If a super-smart expert can’t explain something clearly to you, in a way you understand, then they’re not an expert, nor are they smart or super.

Technology doesn’t solve problems. People do.

Technology, optimization and simulation models and supply chain design software provide unprecedented power to model the future supply chain you have in mind at the SKU, order, and shipment level of detail.

Missing data is always going to be with us. That’s true for you, and it’s true for your competition. When you realize that, you also realize that getting good at making reasonable assumptions is a competitive differentiator. The wider the confidence interval about the data, the more important it is to test your assumptions. Don’t run the model once. Run it hundreds of times. Thousands of times. This is what the best teams do now. Dealing with uncertainty and missing data is just another thing to get better at.

In a sense, supply chain design is converging with what savvy operators have been handling for centuries (millennia, in fact). Leaders who manage logistical operations build redundancy and uncertainty into their plans. They know what they know and what they don’t know.

As supply chain design future state analyses can now handle digital models at roughly the same level of detail as a planning system, we must become savvy operators ourselves. Supply chain designers are the planners of tomorrow’s supply chain. When those changes get made, the hand-off happens, and the supply chain planning organization takes over. Albeit, gratefully, because you’ve dealt them a better hand.

It looks like you’re trying to write a letter! Errr, design a supply chain. Write a poem? Would you like to know more?

Although I’ve used the letters A and I in this article numerous times, I haven’t actually put them together yet. And you can’t write a relevant supply chain article without talking about AI. So how about it, does AI have a role in supply chain design?

Yes, and like the ego of many pontificating buffoons and ethically compromised, self-promoting bloggers on LinkedIn, it’s growing all the time.

Large learning models (LLMs), which have seized the meme-worthy momentum these days, spend their time trying to please and impress you by sounding smart. So do many supply chain “evangelists.” They string together words and phrases that sound smart and send them to you, hoping you’ll think they sound smart too. But for AI, like the “experts,” it’s just language soup or word salad. Trendy and substance-free. There is no danger that AI is going to start designing your future supply chain for you.

That’s not the whole story, though. AI, as a data-driven means of extracting patterns that human analysts can’t detect, is an incredibly powerful and transformative technology. It’s here to stay. Like any other disruptive tool, it is going to be used by people and organizations to solve specific problems in data, yes—missing data, multi-run sensitivity analysis and scenarios, and many other troublesome tasks.

There's no shortcut or silver bullet for designing a better supply chain. If someone offers you a silver bullet, do everyone a favor and fire it at their laptop on your way out the door. AI is in the process of finding its place among all the other modeling and analysis techniques that are out there. What it’s doing now is just work in process.

The silver bullet for missing data

Supply chain design is an increasingly powerful ability to use digital, data-driven models to create the future supply chain your organization is going to need.

Since the future hasn’t happened yet, missing data is simply a part of every project.

Supply chain design capability is a mission-critical team-based approach to using tools, technologies, and domain knowledge to design future supply chains. Your people, armed with domain knowledge and humility, will make estimates, and then build in their own confidence factor. Your team, armed by that and breakthroughs in scalable design platforms, will test those assumptions, building credibility and robustness in their recommendations and decisions.

Making predictions is difficult, especially when they’re about the future. But that’s what design does, and that’s why supply chain design is first and foremost about the people. Get good people and build a great culture. Partner with consultants, peers, and vendors who share your cultural values. You will always deal with missing data, but you won’t actually have missing data. You’ll just have the data that hasn’t happened yet.

SC
MR