By Luis Lopez, AI transportation consultant, CEO of Go Hub.io Holdings Corp and subsidiaries, and host of the Freight Guru Podcast
Every palletized case shipment starts with a decision most people never see: how many cases go on each layer, and how many layers go on the pallet. In warehouse shorthand that is the Ti-Hi. Ti is the number of cases per layer. Hi is the number of layers. Multiply them and you get cases per pallet.
A good Ti-Hi fills the pallet footprint, stays stable through handling, protects the cartons from crushing and fits the trailer. A bad one wastes space, leans, or arrives with collapsed bottom layers. Pattern software has helped with this for a long time. AI adds the ability to learn from what actually happens to pallets after they leave the dock.
The basics that do not change
Before any software gets involved, a few physical rules apply to every pallet pattern:
- Footprint. The most common pallet footprint in the United States is 48 by 40 inches, but customers and retailers may specify others. The pattern has to be built for the actual pallet used.
- Column stacking. When cases sit directly on top of each other with corners aligned, the carton walls carry the load straight down. This preserves compression strength, which matters for heavy or tall stacks.
- Interlocked (brick) patterns. Rotating cases from layer to layer ties the load together and improves stability, but the corners no longer line up, so compression strength drops.
- Overhang. Cases that hang past the pallet edge lose support at their weakest point and are more likely to crush or tear.
The core trade-off is between stability and compression strength, and between using every inch of cube and keeping the load safe. Different products land in different places on that spectrum.
How pattern software and AI pick a Ti-Hi
A pattern engine starts with inputs from the item master:
- Case length, width and height
- Case gross weight
- Carton strength or a maximum stacking load, if known
- Orientation rules, such as “this side up” or cases that cannot be laid on their side
- Pallet type and footprint
- Maximum pallet height and weight, set by the trailer, the racking or the customer
From there it tests possible layer arrangements, checks how many layers fit under the height limit, and checks that the bottom case can carry the weight above it. Traditional tools stop at that geometric answer. AI-assisted tools go further by scoring candidate patterns against history: which patterns for similar cases led to damage claims, leaning loads or rework at the receiving dock.
Trailer height matters too. A pallet that is a few inches too tall to double-stack inside the trailer can waste most of the space above it. Good planning looks at the pallet and the trailer together; the trailer side is covered in AI load planning and trailer cube optimization, and pallet counts per trailer in pallet sizes and how many fit in a trailer.
Mixed-SKU pallets
Single-SKU pallets are the easy case. Mixed pallets, built for store orders or e-commerce replenishment, are harder because every case is different. Software building mixed pallets usually follows rules such as:
- Heavy and strong cases on the bottom, light and fragile on top
- Similar case heights grouped into flat layers where possible
- Liquids and crushable items kept away from heavy corners
- Cases for the same store aisle or destination grouped together when the customer asks for it
AI can help by predicting which combinations tend to shift or crush and by adjusting the build sequence to reduce that risk. It cannot fix a case whose recorded weight is wrong, which leads to the biggest limit below.
Customer and retailer pallet specs
Many retailers and distributors publish pallet requirements: maximum height, maximum weight, pallet type, label placement, whether overhang is allowed, and how the load must be wrapped. A pattern that is optimal for cube may still be rejected at the receiving dock if it breaks those specs. A practical system stores each customer’s rules and checks every proposed pattern against them before it is approved. Wrapping and securing the finished pallet is its own topic, covered in stretch wrap for pallets explained.
Feeding damage data back into patterns
This is where AI earns its place. Most companies already have the signals they need, scattered across systems:
- Damage claims and their photos
- Receiving exceptions from customers
- Rework records when pallets are rebuilt in the warehouse
- Carrier inspection notes
When those records are linked to the SKU and the pattern used, a model can spot that a certain carton crushes when stacked beyond a certain number of layers, or that a brick pattern for one item leads to leaning loads on long routes. The recommended fix might be fewer layers, a column pattern, a tier sheet between layers or a stronger carton. As an AI transportation consultant, I see this feedback loop as the most useful part of the whole process, because it turns damage history into better patterns instead of just claim paperwork.
Limits: bad item master data
Pattern optimization is only as good as the case data. If case dimensions in the system are rounded, out of date, or entered for a different pack size, the software will produce patterns that do not fit the real cartons. Common problems include:
- Dimensions entered for the inner pack instead of the shipping case
- Weights copied from a similar item rather than measured
- Supplier packaging changes that never reached the item master
- Missing carton strength data, which forces the system to guess
Measuring cases on receipt and correcting the item master is unglamorous work, but it decides whether pattern software helps or hurts. For more on keeping that data clean, see AI and item master data in freight.
Where to start
Begin with your highest-volume SKUs and your most frequent damage complaints. Check that their item master data is accurate, compare current patterns with what the software recommends, and run trial pallets before changing anything at scale. Keep people who build pallets every day involved; they often know why a pattern fails before the data shows it.
For more conversations on freight and AI, subscribe to the Freight Guru Podcast.
About the author: Luis Lopez is a Miami-based AI transportation consultant and logistics entrepreneur, the CEO of Go Hub.io Holdings Corp and subsidiaries, and host of the Freight Guru Podcast.