A system employing voice, barcode scanning and light triples the speed that an e-tailer could get costumes and party supplies out the door and to its customers.

BuySeasons, a major online retailer of costumes, accessories, seasonal decor, and party supplies, implemented an order-fulfillment system that more than tripled picking speed while accommodating a peak-season surge in shipping volumes with significant reductions in direct labor costs. 

The increases in picking speeds resulted from replacing BuySeasons’ legacy voice-picking technology with 40 batch-picking carts employing multimodal picking technologies and from the introduction of optimized picking strategies. 

The BuySeasons order-fulfillment system, from Pendleton, S.C.-based FastFetch, employs three key innovations for enhancing productivity. The first is multimodal, batch-picking technology — utilizing a combination of voice, barcode scanning, and light-directed placement — that enables pickers to fulfill multiple orders simultaneously in a hands-free, eyes-up manner without having to look at a piece of paper or a computer screen. 

The second is an overarching emphasis upon optimization of the order-fulfillment processes. A highly effective optimization algorithm is employed to assign orders to carts in a manner that minimizes the total travel required to gather items for pending orders. Additionally, at every stop in the picking area, a second optimization procedure is employed to discover orders that have SKUs in common so that the picker can distribute items into multiple customer orders with a single pick. 

The third key productivity enhancement is a bulk-picking strategy found to be optimal for a large portion of BuySeasons’ orders. The bulk-picking strategy enables a picker to make one trip through the picking area, gathering as many as 24 SKUs for numerous single-item orders. This strategy is particularly effective since pickers can quickly retrieve items for perhaps hundreds of orders on a single trip. 

The average picking rate across all pickers — including inexperienced temporary workers — increased from 38 items an hour to 132 items an hour. The current picking rate is nearly 3.5 times the picking rate achieved using the voice system prior to installing the new technologies and picking strategies. The average picker productivity now substantially exceeds that of the top five pickers using the legacy voice system. 

BuySeasons' product offerings include costumes for children and adults, makeup, jewelry, wigs, masks, party favors, and party tableware. The company operates from a 380,000-square-foot operations and distribution facility in New Berlin, Wis. In addition to its own customers, BuySeasons also fulfills e-commerce orders from such retailers as Walmart, Amazon, Target and JCPenney. 

The legacy picking strategy deployed at BuySeasons separated orders into two classes: baggable orders (e.g., costumes) and non-baggable orders (e.g., party supplies, makeup, and jewelry). Non-baggables were picked into shipping cartons and dropped off to a conveyor that transported completed orders to the shipping area. Baggable items were picked into totes and dropped off at a bagging station where they were manually matched with the order paperwork, labeled and packaged into bags before being sent to the shipping area. 

Using a manual selection process, groups of orders were assigned to individual carts. Although the manual assignment process sought to assign orders to carts so that travel through the picking area would be largely confined to a minimal number of zones, the manual assignment process was far from optimal: carts typically traversed multiple zones to pick all required items. 

A voice-picking solution guided workers to pick locations and to provide pick quantities. Approximately 100 voice-picking headsets were employed to handle peak demands. The voice-picking system was supplemented with RF guns to provide additional picking capacity. 
Especially during the peak season, the majority of BuySeasons’ orders are single-item, baggable orders. BuySeasons therefore adopted a bulk-picking strategy to improve the efficiency of retrieving SKUs for single-item baggable orders. 

Their bulk-picking strategy utilized a cart to retrieve multiple units of a single SKU and deliver those items to the bagging area where they were matched with the order paperwork, labeled, packed into a bag, and placed on the conveyor destined for the shipping area. 

Multi-item, baggable orders were picked by assigning groups of orders to carts containing totes. The batches of orders assigned to carts were not optimized to speed picking. Accordingly, pickers typically had to traverse a substantial portion of the picking area to gather required items. The totes were then dropped off at a bagging station. 

At the outset of the project, FastFetch analysis showed the following characteristics of BuySeasons’ order profile: 

• A relatively large number of SKUs with a high percentage of slow movers 
• A small number of items per order 
• A sizable percentage of single-item, baggable orders 
• Extreme surge in peak-season order volume 

BuySeasons’ voice-picking system did not provide the productivity levels needed to accommodate peak season demands. Although voice-picking enabled pickers to pick multiple orders on a single trip through the picking area, the batches assigned to the carts were not optimized. 

Carts typically had to travel through much of the picking area to fulfill each batch — and voice picking could not readily take advantage of the opportunities for cluster picks where items from a single pick location are distributed into multiple order locations on a cart. 

The BuySeasons WMS pushes orders to a FastFetch host which, in turn, assigns collections of orders to carts. Isolating the WMS from the picking processes eases the task of interfacing picking processes to the WMS and reduces risks that might potentially result from upgrades to the picking processes. 

Order assignments are sent to the carts using a Wi-Fi link. Using a distributed database architecture, as items are picked, the cart tablets update their local databases as well as the central database maintained by the WMS. 

There is no requirement for continuous connectivity between the carts and the FastFetch host database. If a network link between the FastFetch host and a cart becomes unavailable (for example, in a facility dead spot), the FastFetch host will synchronize its local database with the databases maintained on the carts whenever the network link becomes available.