Load plans: from spreadsheet sketch to executable sequence

What is a load plan?

It is the item-by-item representation of how a shipment occupies a container or trailer: the position of every case or pallet in space, its orientation, its stacking level, and a numbered loading sequence. It comes with control metrics: volume utilisation, total weight against payload, centre of gravity position, and — the part most tools hide — an explicit list of what does not fit, the leftover cargo.

A plan that doesn't display its leftovers is a dangerous plan. The most common anxiety among logistics coordinators using market tools is precisely: “were any items left out?”. If the tool doesn't answer that on its own, someone finds out on the dock.

Why spreadsheets fail

The real competitor of load planning software isn't other software — it's Excel. A spreadsheet does one thing brilliantly: it sums volumes and weights. That is exactly the trap. A sum of volumes is not a load. Thirty cubic metres of cargo “fits” arithmetically into a 33 m³ 20-foot container; in practice, long items, non-stackable cases or anything taller than the door opening can make the load impossible at 22 m³.

The spreadsheet knows nothing about geometry (two boxes can't share the same space), stacking (what crushes what), orientation (this side up), sequence (the first delivery must be loaded last) or weight distribution. The classic story: four days spent hand-building a load plan in Excel, invalidated the day before departure because one SKU changed dimensions. Everything has to be redone, cell by cell — there is no “recalculate” button on a drawing.

An optimisation engine does the same work in seconds — and more importantly, redoes it on every change, without fatigue and without oversight.

The constraints that actually matter

What makes the problem hard is not volume — it's constraints interacting with each other. The main ones any serious plan must handle explicitly:

• Stacking — three distinct levels: “nothing on top, ever”; “at most N layers”; and the only physically correct one, a maximum supported weight per item, checked transitively down the whole stack.
• Stability — no floating boxes: the standard operational criterion is at least 75% of the base area supported (100% for fragile goods).
• Orientation — an item has up to 6 possible orientations; “this side up” eliminates 4 of them.
• Sequence and multi-stop — on multi-drop routes, the last stop's goods load first (LIFO). Ignore this and the driver unloads half the trailer at stop one.
• Weight and axles — axle loads are regulated (Directive 96/53/EC in Europe); most weighbridge rejections happen on axle load, not gross weight.
• Centre of gravity — as low as possible, centred laterally, and positioned longitudinally according to the vehicle's load distribution diagram.

How an optimisation engine works

Container loading is an NP-hard problem: nobody — human or machine — enumerates every combination. A good engine works in stages. First block building: 1,000 items across 20 SKUs collapse into a few dozen homogeneous blocks — this is what makes computation fast, and it's what an experienced loader does by instinct. Then constructive heuristics place those blocks using several competing strategies (by size, by weight, by destination), a local search improves the best solution for as long as the time budget allows, and an independent validation pass re-checks every rule: overlaps, support, weight, axles.

Two properties separate a serious engine from a toy. Determinism: the same shipment must produce exactly the same plan on every run — a tool that returns 87% then 91% on identical data earns zero trust. And explainability: every rejected item must carry a readable reason (oversized, over weight limit, no space left, stacking conflict) — never a generic “computation failed”.

An executable plan, not a pretty picture

The final judge of a load plan is neither the buyer nor the planner: it's the dock supervisor. If the plan looks unworkable, the crew ignores it — and the tool is dead within three weeks. An executable plan has four traits:

• A numbered sequence: case 1, case 2, case 3… in the exact order the forklift picks them, consistent with door access.
• A step-by-step view on mobile, readable next to the truck — not an A3 colour PDF pinned to the office wall.
• Leftovers stated up front: what doesn't ship, and why, before the truck is at the dock.
• Fixable from the dock: a wrong dimension discovered while loading must be correctable — with the plan recalculated in seconds, not escalated “to the office” for tomorrow.

Where to start

You don't need an IT project. You need three columns of clean data — dimensions, weight, quantity per SKU — and a tool that computes without setup or an integration project — a free two-field account is all it takes. Start with a real shipment that already left: if the computed plan beats what was actually loaded, you know what you're leaving on the table. To size a shipment, CBM calculation is the first reflex; for palletised freight, check how many pallets fit in a truck in ten seconds; for ocean freight, the full method is in our container stuffing guide.

Frequently asked questions

What's the difference between a load plan and a pallet pattern?

A pallet pattern arranges cartons ON a pallet; a load plan arranges pallets and loose items INSIDE a container or trailer. They chain together — product → carton → pallet → truck — and a mistake at one stage is paid for at the next.

Is a load plan legally required?

The document itself is not mandated by EU regulation, but its consequences are: weight and axle limits (Directive 96/53/EC), compliant cargo securing (EN 12195), and a verified gross mass (VGM) for ocean containers. A load plan is simply the easiest way to prove compliance before departure.

What utilisation rate should I aim for?

There is no universal number: 90% on uniform stackable cartons is common, while 60% on non-stackable oversized cargo can be excellent. Rather than chasing a percentage, compare against the same load planned by hand — and make sure the leftover list is empty or deliberate.

Can I trust an automatically computed plan?

Yes, on two conditions: the engine is deterministic (same input, same plan, to the millimetre) and every violated constraint is explained in plain language. A tool that returns a different plan on every click, or renders floating boxes in 3D, should be discarded.