Snap-on Tools was working to reduce inefficient packaging and absurd box sizes for its e-commerce customers. Like Amazon, Target and other online retailers, the company was responding to rising shipping costs, environmental concern and general nuisance of all that cardboard. An artificial intelligence solution by FastFetch Corp. turned out to be the perfect fit.

The humble cardboard box is having a moment. Once the realm of stockrooms and cargo trucks, boxes are now branding tools, environmental influencers and, in the era of e-commerce, the new storefront.

For companies that ship dozens or thousands of items in any given day, packaging presents big opportunities for customer experience and cost savings — but matching items to appropriate boxes can turn into a complicated science. At tools manufacturer Snap-on Tools, artificial intelligence is changing that.

Snap-on is a $3.7-billion supplier of tools, equipment and service solutions based in Kenosha, Wisconsin, that serves customers in more than 130 countries. Its product line of more than 65,000 SKUs includes hand and power tools, diagnostic software and shop equipment for industries like automotive, agriculture, construction and more. 

A distribution center in Crystal Lake, Illinois, plays the crucial role of connecting the manufacturer to its franchisees by retrieving items from inventory and packing them into containers for shipment. While automated systems are increasingly taking over these roles, most still require people. And the job isn’t always easy in practice. 

Products at Snap-on range in size and shape from small keychains to bulky tool storage units, and each shipment contains a different collection. Packing these items into boxes is often a game of Tetris — where bad moves (read: wasted space) become costly. 

Of course, the industry has a number of solutions. One approach is to maintain a large inventory of assorted box sizes, but there are shortcomings with this strategy: 

• A large inventory requires adequate storage space;
• maintaining a large box inventory complicates any changes in product inventory (since on-hand boxes must first be consumed or wasted); 
• and finally, getting the right-sized boxes from storage to packing stations slows the whole operation. 

Alternatively, some DCs employ machines that build perfectly sized boxes on the spot. Even the fastest among them, however, struggle to keep pace with today’s packing production lines, and not all companies can justify the cost. 

A Collaborative Solution 

In 2018, Snap-on partnered with FastFetch Corp, an order-fulfillment systems provider, for a custom solution. 

Together they designed a carton-management system that determines box sizes, dispenses them at packing stations and replenishes as they’re used. Three carton racks that stand amongst six packing stations hold an assortment of boxes in various sizes that were predetermined by AI, based on Snap-on’s historical order data.

Here’s how it works. 

The Perfect Fit

Before applying AI, FastFetch analyzed nearly 85,000 order records to identify 40 to 50 box sizes that would accommodate all of Snap-on’s past shipments with minimal wasted space. Then, the AI software used a two-step calculation process to cut those choices by about 30% — and spit out the best assortment of sizes. 

“Generally speaking, having a larger number of carton sizes increases the likelihood of selecting a good fit for the items to be packed,” says FastFetch’s Jack Peck, but this “requires more storage space and perhaps includes a number of infrequently used carton sizes.” 

“At some point, increasing the number of carton sizes provides diminishing returns and increases costs,” he says.

Once the boxes were on site and ready for packing, another AI formula came into play. Much like gaming algorithms that can determine a good chess move, FastFetch’s software explores possible item placements, orientations and other properties such as nesting (for example, tapered trash cans stacked together) and containability (items with a hollow center that can contain other items) to determine the best box for a given order. 

Upon scanning an order’s barcode, the system computes optimal dimensions and chooses a box in less than a second. (Packing is left to the worker’s discretion, as item-placement instructions tend to slow the process, FastFetch says. Consequently, the system always chooses boxes that provide a small bit of wiggle room.) 

Packers retrieve boxes from one side of a carton rack, and replenishers fill gaps on the other side. LED strips on both sides light up to indicate which boxes to choose for packing and restocking. 

A single carton rack can serve multiple packing stations by employing different colored LED lights for each station. At Snap-on, each carton rack supports two packing stations. 

‘Fast and Accurate’ 

Two replenishment carts employ barcode and voice technology to ensure fast and accurate retrieval of boxes from storage as follows:

1. A worker logs in via the cart’s tablet screen and requests a set of box orders. 
2. Using voice technology, the cart directs the worker to the first storage location.
3. The worker scans a local barcode to verify arrival. Then, the cart announces how many boxes to pick and highlights where to place them. Numerical displays on the cart also indicate what’s needed. 
4. Once boxes are loaded, the worker confirms proper placement by touching a light-sensitive proximity switch. 

This process continues until the job is done, and then the tablet sends the worker to restock the carton racks. 

For Snap-on, achieving right-sized packaging has boosted fulfillment efficiency — and generated “significant cost savings.” FastFetch’s system was easy to install and use, and the return-on-investment (ROI) period was just three months, according to management. 

Since installation, the distributor cut shipping costs by more than 11% and packing labor by 30%. Corrugated material usage also fell, improving Snap-on’s green footprint by reducing landfill waste and fuel consumption required for transporting packages.