Replenishment: The next big target

Once seen as merely a cost of doing business, warehousing and distribution are rich with potential competitive advantages. For a time, replenishment from reserve to forward pick faces was the new necessary evil as attention to picking productivity intensified.

Now, companies are adjusting to e-commerce and a broader need to fill more precise orders on a more frequent basis. Each product movement and touch now faces tremendous scrutiny, and their impacts can inform processes within the facility as well as the wider supply chain.

According to Richard Rodgers, director of solution development for Dematic, “Historical best practices were economically driven, with each level in the supply chain asking, ‘What’s easiest for me? What can I get away with, and what will my customer put up with?’ Everyone managed to evade as much cost, labor and effort as possible.”

Suppliers used to control what they manufactured and when they shipped it, with an eye toward economies of scale. “Whether retailers or consumers, downstream customers are now driving the bus, not the manufacturer, and it has completely inverted the supply chain of command,” Rodgers says. “Before everyone got smart about carrying all the costs related to inventory, suppliers would load up a buffer and send single-SKU truckloads of goods. The warehouse just needed to figure out how to store it.”

There’s a lot more to figure out these days, but the experts Modern spoke with for this article agree the replenishment function is a good place to start. After all, the small transactions between reserve storage and forward pick areas often mirror the larger relationship between nodes throughout the supply chain. The traditional, linear flow of product from manufacturer down to store shelf has been disrupted by consumer demand to “buy anywhere, ship anywhere.” Inside the four walls, a case might move from receiving directly to a pick face, to another truck, to a value-added service or packing station, or any combination of those. It might even make sense for the manufacturer to skip the DC altogether and ship direct to the consumer.

“Replenishment gets less attention than order picking,” Rodgers says. “Everyone is laser-focused on pick productivity, and replenishment tends to be a less high-profile or less automated area. But, it is the next target.”

New key performance indicators
Those who have spent some time tightening their picking processes won’t have to go far to find that target. Although it occurs right behind the pick face, replenishment is often painfully divorced from picking processes. Luther Webb, director of operations and solutions consulting at Intelligrated, offers an example.

Say a carton flow rack location can hold 15 units, but there is now demand for 16. The system might quickly re-profile the location to show 16 units available for picking, but the stocker will put the extra item on the floor behind the location, or maybe in a secondary reserve area. Either way, Webb says, at some point the picker will wonder where 16 is, and someone has to manually find it. Even with timely and efficient replenishment, a single unit has now wasted lots of time and effort. The effect is almost the same as if the 16th unit had never been brought forward. Webb suggests such stock-outs or “hot replenishments” should never be higher than 3% to 5%. More than 30% can mean trouble.

“Traditional techniques were manageable until the SKUs got out of hand,” says Eric Lamphier, senior director at Manhattan Associates. “There were all these established KPIs like the ratio of reserve to pick, and the presumed synchronization of processes has been the default way it was done. For years it seemed like some kind of art form to issue a replenishment closely with a pick task, but with e-commerce our customers now want batch algorithms that support single-item slotting in batches of one in real time.”

Without even basic software to manage replenishment functions, DC operators of the past would try to keep as much stock as possible up front and only reserve if they had to. “Today people still make pick locations big so they don’t have to replenish often­—because they can’t do it well,” says Michael Howes, vice president of software engineering and service for Forte. “Previously, when you created a task, it sat in the queue for hours. The quantity never changed but picking continued. If the need changes from 100 units to 130 at the last second, you’ve created and delayed the next replenishment task.”

With next-generation replenishment software, the ideal quantity is retrieved at the last possible second, goes to the best location in the best quantity, just in time. “You literally couldn’t have done any better,” Howes says. “Ballpark, with that kind of system, you will see half the travel time for replenishment pickers as compared to what we would expect to be normal travel with modern systems.”
Customers are also looking to evaluate replenishment labor as a percent of total labor and identify its cost per case moved.

Managers might like to stay below a certain number of SKUs or stock-outs as a percent of units shipped. “That’s fine, but say you have a large percentage of stock-outs, but it’s for inexpensive items with small margins,” Webb says. “This is the idea of scoring tasks, and it can go beyond the value of merchandise. If you run out of a promotional product, the cost of that item is one thing, but the value of your promise to customers is another.”

Replenishment might be a non-value-added step, but optimizing it can prevent value loss. “You can look at how to get ‘free’ replenishment through inventory analysis and task interleaving, since it doesn’t cost anything to complete a task while I’m there,” says Jim Barnes, CEO of enVista. “Or you could design and slot a facility with the goal of eliminating replenishment altogether.”

Barnes offers the example of a customer that used to move net 26 full cases per man-hour. “They could have replicated their previous facility, a $7.5 million project with lots of conveyor, three pick modules, a shipping sorter, a routing sorter,” Barnes says. “In the new facility, they now touch each case only once, and with the no-replenishment approach they’re now well north of 135 cases per man-hour.”

New paths to productivity
To accomplish that sort of velocity, many companies rely on an array of picking technologies—from automation to voice to light-directed to wearable scanners—to establish a baseline of accuracy. This also means associates can pick and replenish simultaneously, work with dynamic SKU locations, and generally touch and move items with minimal fear of introducing errors.

In addition to cross-training, you need a strong software foundation to efficiently coordinate all these tasks, whether that be a warehouse management system, warehouse execution system or other slotting module. “Some tend to think of replenishment as an inbound operation. Take in and put there,” explains Dibyendu Ghosh, vice president of product development for Softeon. “There’s much more to it. Software can bring demand-driven replenishment, dynamic replenishment, crossdocking or direct-to-pick during putaway. The dynamic nature keeps increasing, and it no longer makes sense to hold a pile of inventory, but many keep doing just that.”

The key to replenishment is constant monitoring, Ghosh says. “Am I doing the right thing right now? Has product velocity changed? Have product dimensions changed? Is it slotted in the right place? Is it a seasonal product? The logic behind these systems should be flexible enough to accommodate seemingly simple changes that can clog up a process.”

“It’s very simple to bring a case forward from reserve, but say you need to present it to the picker with the box top cut off, or empty the product into a tote,” says Brian Lindenmeyer, principal solutions consultant for IBS (International Business Systems). “Some systems can skip zones, but what if it makes sense to bring a full pallet directly to pack-out?”

For example, if 100 orders of the same teddy bear are bound for 100 different customers, whether the pick face is carton flow or a pallet location, it is not ideal to send 100 pickers to that location. Instead, Lindenmeyer says, a full pallet can be processed at the packing location.

“The concept of direct-to-active has been supported in various flavors for some time,” Lamphier says, “but e-commerce makes operations get more aggressive about finding those opportunities.”

There is another side to that coin. Although it often makes sense to move as much product as you can while you are there, more is not always better. If you are expecting 20 orders for car seats in a given shift, it is not ideal to move all 20 forward in the morning and force the picker to work around them all day. Bringing only five at a time might sound counterintuitive since the number of replenishment tasks has now quadrupled, but because tasks can be blended across orders and associates, the overall effort can be optimized.

Reacting to need
Companies don’t necessarily need robust software to manage all this, Howes says. Even simple spreadsheets and basic tools to slot intelligently can make a big difference. “I might carry three to seven days of stock in each location. More than that will expand the footprint a lot,” he says. “That sort of discipline is not about how you replenish, but the timing of when you need to.”

According to Bill Ostermeyer, vice president of sales for viastore systems, those simple software improvements are the first step toward locating where more sophisticated or automated replenishment might make sense.

“In the traditional setup, you receive to reserve and there’s an inventory buffer in the picking media,” Ostermeyer explains. “The replenishment function occurs continuously throughout the shift. The nice thing about the old way: It’s very flexible, lending itself to discrete or batch picking and the ability to ‘throw bodies at it’ to keep up.”

A new model of automated case picking means everything is in reserve and is retrieved on a per-order basis. Well, not exactly everything. “Some companies look for an automated solution for the entire SKU range, but that’s not realistic,” Ostermeyer says. “Some fast movers might be better handled manually.”

Customers are surprised when Ostermeyer suggests the top 5% of SKUs should not enter an automated storage and retrieval system (AS/RS) and that a more conventional pallet flow approach is better—even for piece-picking applications.

Similar to the concept of scoring tasks, locations in a given pick module can be given preferred status. “Storage systems and their usage have changed, and are morphing into case-ready storage accessed by persons not equipment,” says Art Eldred, client executive for systems sales for Vargo. “I think of it more as ‘preferred versus not preferred’ instead of ‘reserve and forward.’”

Preferred status can­—and should—change from moment to moment, Eldred says. For example, baseballs and gloves might be slotted close together because they’re being sold as a bundle. But, it may turn out that gloves and hats are selling together more often, and that the hat has to be retrieved or conveyed long distances to get the order’s SKUs together.

“Retail has always been built on plans,” Eldred says. “When you mix with e-commerce it’s water and oil. The industry is having a tough time releasing control of these systems, and that has hurt us over the years. The name of the game now is versatility and flexibility.”

Even if fast movers could be accurately predicted, condensing them in a single area can lead to congestion. “For some e-commerce customers with a long tail of slow-movers, that approach is no longer your friend,” Eldred says. “The idea is to keep SKU locations as random and flexible as possible to spread the wealth.”

Matt Aprea, director of product management for Harvest Automation, agrees. He describes a system of mobile autonomous robots working alongside pickers in a traditional warehouse pick module, with an 8-foot pick area and bulk storage above. Instead of conveyors, carts or lift trucks, a picker uses totes on the shelf as pick sources, order containers or replenishment destinations.

“In the traditional mode, picking and replenishment are seen as discrete operations with separately trained associates,” Aprea says. “They might inhabit the same place, but they’re not doing the same things. That will change, and it will decouple the picker from a certain limited set of orders. Each person will be responsible for as many orders as the building can handle.”

In Aprea’s scenario, consider a tote with 10 items. The same associate can take five, put them into an empty tote, and complete an order that a robot will later bring to pack-out. The other five items can be placed in a tote for an order that will need one more item from the other side of the facility. The robot will later ferry that tote, but the empty donor tote has immediately switched from a storage location to an order container.

The whole idea behind touches is that some amount of time elapses, some task is completed, or some person or piece of equipment was in motion consuming time and energy. If replenishment, picking and order consolidation can all happen on the same shelf, in the same locations and by the same person, the time and effort between each task is virtually eliminated.

Webb expects to see more automated storage, goods-to-person, mini-loads and shuttles, and robotic solutions play a powerful role in replenishment. “When we talk about replenishment historically, the first thing that came to mind was that it would need to be conveyorized,” he says. “Now with e-commerce, SKU proliferation, flash sales, short SKU lives, labor costs and service levels, you can’t afford to have something stocked out, much less replenish something you need to ship same day. We’re now way beyond conveyorized replenishment.”

Companies mentioned in this article
• Dematic
• enVista Corporation
• Forte Industries
• Harvest Automation
• IBS (International Business Systems)
• Intelligrated
• Manhattan Associates
• Softeon
• Vargo
• viastore systems