The age of big data and the demands of e-commerce are generating more data than ever before — and with it, the need for "actionable" analytics to manage operations in distribution centers, says Charles Armstrong, founding partner with Orion Advisors Group.

SCB: What’s the problem as you see it in the distribution industry today?

Armstrong: Consumer demands, particularly in e-commerce, have exacerbated the challenges that we've always had in running our distribution centers. The process now is all about speed and predictability. We've added this e-commerce process, where customer orders drop continuously. That has forced organizations to respond by investing in capital, larger buildings, and robotics.

We've educated the customer. It used to take four days to deliver a package, then three days, then two days. In some cases now, it's one day. And we've got cutoff times because now we're using small-package carriers. Warehouse-management systems have responded by eliminating batching, because it was an ineffective process. Now that we have very tight windows and are leaning out our operations, we need to continuously flow orders out to the floor as they're dropping, so they can be worked according to different sets of priorities. We've taken the process of managing the D.C. floor and we’ve significantly complicated it, and increased the stakes for shipping on time.

SCB: What doesn't industry understand about this problem?

Armstrong: For the industry, it's about line balancing. It's about how you balance the mechanical resources — sortation systems, robots. Then you've got your talent pool, the associates who are picking and packing. That mix has become dynamic; it changes throughout the day, according to whether you’re doing multi-waves, single waves, cluster pick, or discrete pick. All these are variables that we didn't have to deal with 20 years ago, but now have to be managed in real time. The alternative is to spend more money on infrastructure. Optimizing the execution process is absolutely critical in order to survive in the future.

SCB: A lot of people talk about the need for analytics, but you talk about actionable analytics. What do you mean by that?

Armstrong: First let's talk about what's not actionable analytics. If you go to any distribution center across the country you're probably going to find people working with Excel spreadsheets, downloading from labor, warehouse and transportation-management systems. They're using that to create workflow or work management, but unfortunately it's not being done in real time. It's all dated material. As you lean out D.C.s that are running faster and faster, the implications of errors surface much quicker. They have a much larger impact than when you operated a building that had WIP [work in progress], where you had the ability to buffer up. We're eliminating all the buffers.

The new world says we have to operate in a much leaner fashion. So actionable analytics is the process where we start to take all of the real-time data that's now available. You've got transportation systems with API [application programming interface] connections that tell you when a vehicle's going to arrive. We need to be tracking all the activity as the orders drop. The systems have the capability to identify the work content, and the productivity required for each element of work. So actionable analytics is the process where we take the advantage of artificial intelligence, and historical information around productivity, capacity, and throughput, and create directions for the management team. They tell you to increase this workload, decrease this flow, shift personnel from this department to that one. In some cases, ultimately, the system will also tell you who to move. Because if you're in a Lean situation and need to recover and get your building smoothed back out, you want to put your “A” players into those slots. You want to get back to normal status as fast as possible.

SCB: Obviously this is a still-developing technology within the world of distribution centers. Can you give me an analogy for where this type of system is working elsewhere?

Armstrong: A great example that all of us can relate to is the airline industry. You're all aware in the past, air traffic controllers used to track planes with little placards on a radar screen. They monitored things like turbulence and distance between planes. That's how we traveled safely. In today's world, that manual effort has been replaced by new air traffic control systems, which monitor the speed of every single airplane. They understand its wind-turbulence effects on other planes. In the past, you might have wondered why you were flying past an airport before making a U-turn and coming back. Today, that doesn't happen as frequently. The reason is that these systems, with actionable analytics, can identify the plane you're on and make a turn in the pattern to get in front of another plane in a safe way. We can move more airplanes faster and in less space, reducing the time that passengers spend circling airports.

SCB: And you can achieve the same kind of results in a distribution center?

Armstrong: The same concepts apply. It's understanding throughput by product mix, job element and job function, and allowing the systems to do all the mathematical calculations on a continuous, real-time basis. That allows leadership teams to more effectively run the buildings, get product out on a timely basis, and serve the customer.

SCB: You also seem to be suggesting that it's changing the very nature of the D.C., in terms of the size of the facility, what goes on inside, and how much inventory you need to keep on hand in order to get the job done.

Armstrong: All of those things are changing. We’re trying to operate with less inventory. We have to operate with higher fill rates, which means accurate cycle counting of a SKU that seems to be out of stock when it's not supposed to be. That has to happen very quickly now, because other orders are coming right behind. Let the system determine that there's a high-priority SKU that needs an immediate cycle count in order to correct the situation, or to be able to schedule a move.

SCB: What has to happen for this to become a mature reality within D.C.s? You seem to be suggesting that we're on that journey, but we're not there yet.

Armstrong: What we need to have is more connectivity between systems. We've made progress. The good news is that the capabilities are there now. We have real-time systems and cloud-based software that can do the calculations. AI is an option, and we can do predictive analytics. So it's about connecting the various components into one tool, one process, as opposed to all of the disparate tools that we typically are operating with today.