Automated system helps grocery DC make smarter picks
An integrated order picking system has slashed labor costs, boosted throughput, and virtually eliminated errors at KeHe Distributors' new DC. And it used ordinary technologies to do it.
Peter Bradley is an award-winning career journalist with more than three decades of experience in both newspapers and national business magazines. His credentials include seven years as the transportation and supply chain editor at Purchasing Magazine and six years as the chief editor of Logistics Management.
Speed and efficiency are crucial to every DC operation, of course, but the pressure to achieve ever-higher productivity levels is greater in some industries than in others. The grocery business, with its razor thin margins, is a good example. As Craig Turner, vice president of operations for specialty food supplier KeHe Distributors, puts it, "The ability to reduce time and motion in a low-margin environment is extremely important in order to remain competitive in our market."
So it's no surprise that when the company began planning for a new DC in Allentown, Pa., it made productivity a priority. The project was part of a larger DC consolidation effort aimed at eliminating redundant operations. The distributor had already completed a similar project in the Dallas area, combining several operations into a single DC in Dallas. The Allentown distribution center, in turn, would absorb operations from an Albany, N.Y., facility once construction was finished.
At the same time, KeHe began looking into systems that would boost efficiency and make maximum use of capacity in the facilities it would retain. The mission took on particular urgency late last year as the company prepared to acquire competitor Tree of Life—an acquisition it completed this past February.
Tall order
Founded in 1952, Romeoville, Ill.-based KeHe has grown into a $2 billion company today. As one of the largest suppliers of natural and specialty foods, it distributes dry goods, frozen foods, and perishables—about 60,000 SKUs in total—to some 33,000 retail outlets throughout North America and the Caribbean. Its customers include some of the largest food retailers as well as independently owned natural and organic food stores.
Like its counterpart in Dallas, the consolidated Allentown operation would serve both types of customers—the big mass marketers and the independent retailers. As Gene Carter, KeHe's executive vice president of distribution and supply chain, explains: "We wanted to take two environments and combine them in a sort of hybrid DC that would support both specialty and national channel sales."
That had implications for the order picking end of the Allentown operation. To be precise, it meant the facility would have to be able to handle orders for full pallet and full case quantities as well as orders for individual items, or "eaches." And it would have to do it cost-effectively. "We had to improve our throughput sales per square foot, which are now up 50 percent," says Carter. "Reduction of fixed-cost expenses was a priority for us."
It was clear from the outset that one of the key parts of the project would be revamping the facility's each-picking system. In the past, less-than-case order picking had been a manual operation, with workers selecting items from mezzanines and flow racks with paper pick lists. But a more efficient process would be required at the new DC. It was time to automate.
To design and implement an automated each-picking system in Allentown, KeHe called on two partners: Automation Dynamics, a Wylie, Texas-based material handling systems integrator that had overseen a similar project at KeHe's Dallas DC, and Intelligrated, a Mason, Ohio-based firm that designs and produces automated material handling systems.
Turner set the bar high for the design project. The solution had to boost productivity, assure high levels of order accuracy, and do it all efficiently. The DC has a tight window for processing orders, he explains. "When orders are dropped in our environment, we have less than eight hours before the first truck goes out. We cannot waste time chasing a box through the system in order to get a 99.9-percent accuracy rate."
The solution also had to be compact. Intelligrated and Automation Dynamics would have to take an each-pick operation that had filled 50,000 square feet of space and fit it into 20,000 square feet.
On top of that, the design team would have to devise a system that could efficiently handle both fast- and slow-moving items—that is, the 2,000 SKUs that account for 80 percent of the facility's orders as well as tens of thousands of slower movers. Turner sums up the challenge this way: "How were we going to take these eaches and create a dense pick area that did not have to have levels and levels of pick modules?"
Custom package
The solution the team came up with is a customized design that combines pick-to-light technology (which is used in conjunction with flow racks) and carousels. The setup allows the company to use the optimal picking method for each kind of item, Turner says. "We took the carousel system and used that for what I call the long tail, those C and D items that have minimal movement, and integrated it with a multi-level pick-to-light system that handles [the fast-moving SKUs]," he explains.
As for equipment, the system incorporates photo-eye accumulation conveyor, spiral conveyor, and a two-level pick- and put-to-light mezzanine (to take advantage of vertical space). It also includes a warehouse management system (WMS) and eight eight-foot horizontal carousels.
Turner emphasizes that there was much more to the project than just installing the automated equipment. "We did a lot of due diligence on the application of these two technologies," he says. "We didn't just say, 'Let's take a carousel and let's take a pick to light and slap them together,' because that in itself would not be faster than what we were already dealing with. We had to get into the heart of it and actually retool each pick solution."
Orders by the batch
Because the orders handled in Allentown are relatively small, the system designers opted for a batch processing approach. For example, instead of sending a worker with a single order tote through the whole pick-to-light setup, the system is programmed to process multiple orders simultaneously. "We'll go into a zone and scan 12 boxes at a time, and lights will light up for all 12," Turner says. "We'll pick to light, then put to light. A light bar under the conveyor tells the worker which box to drop it into. So we're able to batch pick 12 orders at a time through the entire system."
He adds that the pick-to-light zones are designed with multiple entry and exit points for containers. "If a box is completed, it can exit early, which gets it quickly through the system. It also lets us introduce new boxes with different start points. The selectors can maximize their batches at all times to maintain efficiencies."
KeHe's WMS monitors the picking activity to ensure operations stay on track. For instance, the system is able to recognize immediately if a container is moved out of a zone before picks in that zone for that order have been completed. "It returns the tote back into the zone before it exits the pick-to-light environment so we don't waste time allowing a tote to travel through the whole system before we recognize something is missing," Turner says.
To assure order accuracy, each container automatically undergoes inspection as it leaves the each-pick area. A check-weight system determines if the box is within a small tolerance based on the goods ordered. If it determines a box is out of tolerance, the system diverts it to a control area for further examination. While the box is being weighed, a digital camera snaps a photo of the open container—the average box has about a dozen items—and archives the photo for future reference. "We've seen tremendous improvements in our order accuracy thanks to the various controls installed on this system," Turner says. "It has allowed us to produce nearly perfect orders."
Productivity gains have been substantial as well. "When we were in traditional flow rack with manual picking, [employees] who had been there for 10 years averaged about 110 lines an hour. Now, we're [averaging] three to four times that," Turner says. He adds that another advantage of the pick-to-light system is that minimal training is required. "New employees can now reach this level of productivity and accuracy within a few weeks," he says.
Riding along on a carousel
Orders filled from Allentown's carousel units are also handled in batches, with workers picking into as many as 16 totes at a time. When a container is completed, they move it across cantilevered racks onto the conveyor system for movement through the check-weight system and on to shipping. (In both the pick-to-light and carousel operations, workers pick directly into the shipping containers.)
To avoid having workers wait around while a carousel's bins rotate into position, the system designers programmed the WMS to automatically track how quickly goods are moving and assign the fastest movers to multiple bins. "We balance [the faster-moving items] out among a number of carousels so we're never waiting for a carousel to spin," Turner explains. The result has been a significant uptick in speed. "We are getting 400 lines per hour in an environment where in the past, the best we could do was 110," he says.
In another bid to boost efficiency, the system designers engineered a way to replenish the carousels from the rear. Restocking through the back assures that replenishment operations won't interfere with the picking process.
For all their benefits, the carousels do have one minor drawback. Replenishment requires workers to remove goods from cases before depositing them in the carousel bins, rather than just opening the cases as they did with the flow racks. But that appears to have had a minimal effect on productivity. All in all, the Allentown site has been able to reduce labor by 35 to 40 percent.
Room for expansion
While the carousel and pick-to-light systems deserve most of the credit for the productivity gains at the Allentown DC, the operation has benefited from some minor changes as well. A case in point is a relatively simple change made to the labels that are printed when the WMS "drops" an order. Each label is now color coded by customer via a colored band at the bottom. "We may print customer A in red and customer B in green, so when you go to apply a label, you can quickly match colors rather than having to look at the customer ID number," Turner says. "If you watch someone picking and see a red label on a box on a green pallet, you know there's a mistake. It has helped our quality control."
Now that the new systems are in place, Allentown is ramping up to take on more work. The facility, which ships out about 2,000 totes a day on average, will soon absorb operations from a nearby third-party DC. As business expands, the facility should have no trouble accommodating additional volume. All it has to do is install more levels of carousels and pick-to-light racks.
Asked how the new picking system is working out to date, Carter, Turner, and other company officials say they're delighted with the results. Carter says, "Being able to take an old bit of technology and retool it to a design that is very applicable and very efficient for us, we couldn't be more excited." The gains have been impressive enough, in fact, that KeHe may soon go a step further. The company is considering launching a similar project at its 1 million-square-foot DC in Romeoville, Ill.
Artificial intelligence (AI) and data science were hot business topics in 2024 and will remain on the front burner in 2025, according to recent research published in AI in Action, a series of technology-focused columns in the MIT Sloan Management Review.
In Five Trends in AI and Data Science for 2025, researchers Tom Davenport and Randy Bean outline ways in which AI and our data-driven culture will continue to shape the business landscape in the coming year. The information comes from a range of recent AI-focused research projects, including the 2025 AI & Data Leadership Executive Benchmark Survey, an annual survey of data, analytics, and AI executives conducted by Bean’s educational firm, Data & AI Leadership Exchange.
The five trends range from the promise of agentic AI to the struggle over which C-suite role should oversee data and AI responsibilities. At a glance, they reveal that:
Leaders will grapple with both the promise and hype around agentic AI. Agentic AI—which handles tasks independently—is on the rise, in the form of generative AI bots that can perform some content-creation tasks. But the authors say it will be a while before such tools can handle major tasks—like make a travel reservation or conduct a banking transaction.
The time has come to measure results from generative AI experiments. The authors say very few companies are carefully measuring productivity gains from AI projects—particularly when it comes to figuring out what their knowledge-based workers are doing with the freed-up time those projects provide. Doing so is vital to profiting from AI investments.
The reality about data-driven culture sets in. The authors found that 92% of survey respondents feel that cultural and change management challenges are the primary barriers to becoming data- and AI-driven—indicating that the shift to AI is about much more than just the technology.
Unstructured data is important again. The ability to apply Generative AI tools to manage unstructured data—such as text, images, and video—is putting a renewed focus on getting all that data into shape, which takes a whole lot of human effort. As the authors explain “organizations need to pick the best examples of each document type, tag or graph the content, and get it loaded into the system.” And many companies simply aren’t there yet.
Who should run data and AI? Expect continued struggle. Should these roles be concentrated on the business or tech side of the organization? Opinions differ, and as the roles themselves continue to evolve, the authors say companies should expect to continue to wrestle with responsibilities and reporting structures.
Shippers today are praising an 11th-hour contract agreement that has averted the threat of a strike by dockworkers at East and Gulf coast ports that could have frozen container imports and exports as soon as January 16.
The agreement came late last night between the International Longshoremen’s Association (ILA) representing some 45,000 workers and the United States Maritime Alliance (USMX) that includes the operators of port facilities up and down the coast.
Details of the new agreement on those issues have not yet been made public, but in the meantime, retailers and manufacturers are heaving sighs of relief that trade flows will continue.
“Providing certainty with a new contract and avoiding further disruptions is paramount to ensure retail goods arrive in a timely manner for consumers. The agreement will also pave the way for much-needed modernization efforts, which are essential for future growth at these ports and the overall resiliency of our nation’s supply chain,” Gold said.
The next step in the process is for both sides to ratify the tentative agreement, so negotiators have agreed to keep those details private in the meantime, according to identical statements released by the ILA and the USMX. In their joint statement, the groups called the six-year deal a “win-win,” saying: “This agreement protects current ILA jobs and establishes a framework for implementing technologies that will create more jobs while modernizing East and Gulf coasts ports – making them safer and more efficient, and creating the capacity they need to keep our supply chains strong. This is a win-win agreement that creates ILA jobs, supports American consumers and businesses, and keeps the American economy the key hub of the global marketplace.”
The breakthrough hints at broader supply chain trends, which will focus on the tension between operational efficiency and workforce job protection, not just at ports but across other sectors as well, according to a statement from Judah Levine, head of research at Freightos, a freight booking and payment platform. Port automation was the major sticking point leading up to this agreement, as the USMX pushed for technologies to make ports more efficient, while the ILA opposed automation or semi-automation that could threaten jobs.
"This is a six-year détente in the tech-versus-labor tug-of-war at U.S. ports," Levine said. “Automation remains a lightning rod—and likely one we’ll see in other industries—but this deal suggests a cautious path forward."
Editor's note: This story was revised on January 9 to include additional input from the ILA, USMX, and Freightos.
Logistics industry growth slowed in December due to a seasonal wind-down of inventory and following one of the busiest holiday shopping seasons on record, according to the latest Logistics Managers’ Index (LMI) report, released this week.
The monthly LMI was 57.3 in December, down more than a percentage point from November’s reading of 58.4. Despite the slowdown, economic activity across the industry continued to expand, as an LMI reading above 50 indicates growth and a reading below 50 indicates contraction.
The LMI researchers said the monthly conditions were largely due to seasonal drawdowns in inventory levels—and the associated costs of holding them—at the retail level. The LMI’s Inventory Levels index registered 50, falling from 56.1 in November. That reduction also affected warehousing capacity, which slowed but remained in expansion mode: The LMI’s warehousing capacity index fell 7 points to a reading of 61.6.
December’s results reflect a continued trend toward more typical industry growth patterns following recent years of volatility—and they point to a successful peak holiday season as well.
“Retailers were clearly correct in their bet to stock [up] on goods ahead of the holiday season,” the LMI researchers wrote in their monthly report. “Holiday sales from November until Christmas Eve were up 3.8% year-over-year according to Mastercard. This was largely driven by a 6.7% increase in e-commerce sales, although in-person spending was up 2.9% as well.”
And those results came during a compressed peak shopping cycle.
“The increase in spending came despite the shorter holiday season due to the late Thanksgiving,” the researchers also wrote, citing National Retail Federation (NRF) estimates that U.S. shoppers spent just short of a trillion dollars in November and December, making it the busiest holiday season of all time.
The LMI is a monthly survey of logistics managers from across the country. It tracks industry growth overall and across eight areas: inventory levels and costs; warehousing capacity, utilization, and prices; and transportation capacity, utilization, and prices. The report is released monthly by researchers from Arizona State University, Colorado State University, Rochester Institute of Technology, Rutgers University, and the University of Nevada, Reno, in conjunction with the Council of Supply Chain Management Professionals (CSCMP).
The overall national industrial real estate vacancy rate edged higher in the fourth quarter, although it still remains well below pre-pandemic levels, according to an analysis by Cushman & Wakefield.
Vacancy rates shrunk during the pandemic to historically low levels as e-commerce sales—and demand for warehouse space—boomed in response to massive numbers of people working and living from home. That frantic pace is now cooling off but real estate demand remains elevated from a long-term perspective.
“We've witnessed an uptick among firms looking to lease larger buildings to support their omnichannel fulfillment strategies and maintain inventory for their e-commerce, wholesale, and retail stock. This trend is not just about space, but about efficiency and customer satisfaction,” Jason Tolliver, President, Logistics & Industrial Services, said in a release. “Meanwhile, we're also seeing a flurry of activity to support forward-deployed stock models, a strategy that keeps products closer to the market they serve and where customers order them, promising quicker deliveries and happier customers.“
The latest figures show that industrial vacancy is likely nearing its peak for this cooling cycle in the coming quarters, Cushman & Wakefield analysts said.
Compared to the third quarter, the vacancy rate climbed 20 basis points to 6.7%, but that level was still 30 basis points below the 10-year, pre-pandemic average. Likewise, overall net absorption in the fourth quarter—a term for the amount of newly developed property leased by clients—measured 36.8 million square feet, up from the 33.3 million square feet recorded in the third quarter, but down 20% on a year-over-year basis.
In step with those statistics, real estate developers slowed their plans to erect more buildings. New construction deliveries continued to decelerate for the second straight quarter. Just 85.3 million square feet of new industrial product was completed in the fourth quarter, down 8% quarter-over-quarter and 48% versus one year ago.
Likewise, only four geographic markets saw more than 20 million square feet of completions year-to-date, compared to 10 markets in 2023. Meanwhile, as construction starts remained tempered overall, the under-development pipeline has continued to thin out, dropping by 36% annually to its lowest level (290.5 million square feet) since the third quarter of 2018.
Despite the dip in demand last quarter, the market for industrial space remains relatively healthy, Cushman & Wakefield said.
“After a year of hesitancy, logistics is entering a new, sustained growth phase,” Tolliver said. “Corporate capital is being deployed to optimize supply chains, diversify networks, and minimize potential risks. What's particularly encouraging is the proactive approach of retailers, wholesalers, and 3PLs, who are not just reacting to the market, but shaping it. 2025 will be a year characterized by this bias for action.”
Under terms of the deal, Sick and Endress+Hauser will each hold 50% of a joint venture called "Endress+Hauser SICK GmbH+Co. KG," which will strengthen the development and production of analyzer and gas flow meter technologies. According to Sick, its gas flow meters make it possible to switch to low-emission and non-fossil energy sources, for example, and the process analyzers allow reliable monitoring of emissions.
As part of the partnership, the product solutions manufactured together will now be marketed by Endress+Hauser, allowing customers to use a broader product portfolio distributed from a single source via that company’s global sales centers.
Under terms of the contract between the two companies—which was signed in the summer of 2024— around 800 Sick employees located in 42 countries will transfer to Endress+Hauser, including workers in the global sales and service units of Sick’s “Cleaner Industries” division.
“This partnership is a perfect match,” Peter Selders, CEO of the Endress+Hauser Group, said in a release. “It creates new opportunities for growth and development, particularly in the sustainable transformation of the process industry. By joining forces, we offer added value to our customers. Our combined efforts will make us faster and ultimately more successful than if we acted alone. In this case, one and one equals more than two.”
According to Sick, the move means that its current customers will continue to find familiar Sick contacts available at Endress+Hauser for consulting, sales, and service of process automation solutions. The company says this approach allows it to focus on its core business of factory and logistics automation to meet global demand for automation and digitalization.
Sick says its core business has always been in factory and logistics automation, which accounts for more than 80% of sales, and this area remains unaffected by the new joint venture. In Sick’s view, automation is crucial for industrial companies to secure their productivity despite limited resources. And Sick’s sensor solutions are a critical part of industrial automation, which increases productivity through artificial intelligence and the digital networking of production and supply chains.