The computer recommends a box. Its recorded volume is several times greater than the product’s volume. The packer places the product beside it and immediately sees the problem: the product is too long.
Nothing has gone wrong with multiplication. The plan confused enough space in total with the right shape of usable space. That small distinction is the beginning of cartonisation.
Cartonisation is the process of deciding which products should share a package and which available packaging configuration can contain them while satisfying the required geometry, protection, weight, handling and delivery constraints.
It can produce a recommendation before picking begins or support a decision at the packing station. In either case, a recommendation is not a completed parcel. The physical contents, packaging and recorded shipment must still be brought into agreement before release.
Oracle’s Warehouse Management User’s Guide describes a carton-selection implementation using grouping, weight capacity, volume, dimensions and packaging restrictions. Microsoft’s containerisation documentation describes packaging definitions and mixing rules used when warehouse waves are processed. These are primary examples of configurable software behaviour, not proof that every suggested package is physically suitable.
This is Article 124 in the extended How Logistics Works collection. Packing for Logistics owns physical preparation; Shipment Dimensioning owns measurement; Packaging Efficiency owns the wider protection-and-resource trade-off. Here, the reader’s job is narrower: choose a feasible packaging arrangement for this order, then verify it against the parcel that actually leaves.
Enough volume does not establish a feasible fit
Consider an original mathematical example: a rigid item measuring 50 × 5 × 5 centimetres and a cubic box with usable internal dimensions of 20 × 20 × 20 centimetres. The item occupies 1,250 cubic centimetres. The box contains 8,000 cubic centimetres of internal volume.
The item still cannot fit. Even the longest straight distance inside that cube, its corner-to-corner diagonal, is only about 34.6 centimetres. The rigid item’s 50-centimetre length exceeds it. Allowing arbitrary rotation cannot solve this particular mismatch.
The example is deliberately generous to the box: it ignores cushioning and closure allowances. Those requirements would reduce the usable arrangement further. Volume is one necessary screening condition, not a complete packing proof.
Multiple items add another difficulty. Each might fit individually while their combined shapes cannot be arranged together within the available space. A program that checks separate item dimensions and total volume may therefore need further arrangement logic or a validated packing recipe before its suggestion becomes executable.
Decide what is allowed to share a package before choosing its size
Two items fitting together does not establish that they belong together. They may serve different receivers, require different handling, belong to separate order commitments or need independent identification. Geometry should operate inside these rules, not override them.
Start with the permitted grouping. Which order lines may be combined? Must a kit remain together? Can a large order split across parcels? Does the customer require a particular presentation or package-level document? These questions determine the candidate groups that the packing calculation is allowed to consider.
Microsoft’s documented container-build templates include inventory-mixing rules, including the ability to prevent allocations for different customers from sharing a container. That is a software example of an important design order: first define a legitimate grouping, then search for a suitable package.
Internal capacity and external shipment size describe different boundaries
The products must fit inside the package. The transport network must accommodate the package’s outside dimensions. Those are not interchangeable measurements.
In a hypothetical carton record, usable internal dimensions are 32 × 22 × 12 centimetres, giving 8,448 cubic centimetres. Its completed external envelope measures 34 × 24 × 14 centimetres, giving 11,424 cubic centimetres. The difference represents the distinct boundaries being described; it is not free space that the packer can use.
A selection engine using external dimensions as internal capacity can recommend a package that is too small. A shipment declaration using internal dimensions can understate the parcel presented to the carrier. Store and name both attributes clearly enough that each process uses the appropriate one.
Cushioning is part of the arrangement, not an afterthought
A product that fits tightly inside an empty carton may leave no room for its required protective material. A fragile surface may need separation from another item. An insert may constrain orientation or reduce the space available elsewhere in the package.
The selection process should therefore use the authorised protective configuration where one exists. It should not assume that the packer will somehow add the necessary protection after the geometric capacity is fully consumed.
A validated product-specific packing method can be more important than a generic density target. Oracle’s documented container-load relationships illustrate this distinction: specific item-and-container arrangements can take precedence over its ordinary carton-selection logic. The relevant packaging owner must establish the real protection requirement; the cartonisation system should preserve it.
Some constraints are non-negotiable; others are preferences
An item that cannot enter the box is infeasible. A package exceeding an applicable weight limit is infeasible for that limit. An unauthorised mixed load does not become acceptable because its cube utilisation is excellent.
Within the feasible options, the operation can compare preferences such as material use, parcel count, packing time, presentation and total delivered cost. Keeping the two layers separate prevents a favourable score from compensating for a requirement that must actually hold.
The practical selection sequence is to reject configurations that violate required constraints, then compare the remaining alternatives for the receiver’s job. The cheapest infeasible package is not a bargain. It is an exception waiting for someone downstream to discover it.
Rotation needs permission as well as geometry
A rectangular item can have several possible orientations. A packing calculation may find a fit by turning it onto another side. That mathematical option is not automatically an acceptable handling option.
Some products or packaging arrangements have a required orientation. Some shapes may only be stable in particular positions. Some inserts are designed around one placement. The data model should represent those restrictions rather than forcing the packer to remember them after the calculation has already selected a box.
Also check what the actual software supports. Microsoft’s referenced implementation distinguishes lateral dimension interchange from vertical rotation. That documented limitation is not a universal packing rule; it is a reminder to understand the model’s allowed moves before treating its result as a search of every physically possible arrangement.
Nesting and compression should never be invented to make the numbers fit
Some objects can legitimately nest. Some soft goods can be compressed within an approved packaging method. Other products lose shape, function or protection when treated that way. A generic algorithm cannot safely infer the permission from a convenient gap in the dimensions.
Use documented product relationships and validated packaging instructions where these behaviours matter. The record should distinguish a permissible compact configuration from an optimistic assumption that an operator can force the contents into place.
When no supported arrangement fits, retain that result as an exception. Selecting a larger suitable package or escalating an unusual item can be more efficient than repeating a physically impossible recommendation at several packing benches.
Weight has several meanings at the same bench
The goods have a mass. The empty carton and protective materials add mass. The completed parcel has a gross mass. The carton, handling process and chosen service may each impose a relevant limit.
Check the meaning of every field before combining them. A maximum payload allowance is not necessarily a maximum gross-package allowance. Subtracting the empty packaging twice can reject feasible options; omitting it entirely can understate the final parcel.
Oracle’s storage-container definitions distinguish container dimensions, empty-container weight and maximum weight as configured attributes. The terminology and accounting basis must be read in the relevant implementation. The wider lesson is to preserve units and meanings, not merely populate numeric fields.
One box versus several is a service decision
Fewer parcels can reduce separate handling and simplify receipt. One large package can also be awkward, expensive to move or unsuitable for the products. Splitting can improve physical fit while creating more labels, movements and opportunities for part of an order to arrive separately.
Compare only arrangements that preserve the customer promise. A set required for one appointment may need coordinated receipt even when it ships in several parcels. A customer who accepts split delivery presents a different job from one who requires the entire order together.
The decision should therefore connect cartonisation to order structure. When a split is authorised, preserve the relationship between the order and every resulting parcel. One successfully delivered carton should not make an incomplete multi-parcel order appear complete.
Compare the whole packaging-and-delivery burden
A low-cost carton can require more protective material or more packing time. A smaller parcel may qualify for a different transport price but require two packages instead of one. A dense arrangement may save space and increase damage or difficult unpacking.
For a particular decision, define a common comparison boundary. It might include cartons, protective materials, labour, freight and recurring recovery costs. The figures should come from the applicable service and local operation, not a generic claim that smaller always means cheaper.
Use observed damage and return evidence when available. Where it is absent, do not invent a reassuring damage probability to make an attractive option win. State the uncertainty and obtain the packaging assessment or trial needed for the consequence involved.
A catalogue of every possible box can create another problem
A larger carton range can provide closer matches to varied orders. It also creates more packaging stock to identify, replenish and store. Packers may spend longer finding the correct size, and infrequently used options can occupy space without much practical benefit.
A smaller range simplifies the station but can leave more empty space or force avoidable splits. The appropriate range depends on the actual order shapes and the benefits achieved by each additional size.
Study recurring order families rather than isolated perfect fits. A box that improves hundreds of ordinary orders may deserve a permanent position. A package needed once for an unusual object may be better handled through a defined exception route.
The recommended carton must actually be available
A theoretically ideal package cannot help a packer when the station has run out of it. Packaging stock is therefore an execution dependency, not a background purchasing detail.
Define suitable fallback options in advance where possible. A replacement carton must still preserve protection, weight, grouping and handling requirements. A larger box is not automatically safe, especially when the protective method depends on a particular insert or snug arrangement.
When no validated alternative is available, escalate or hold the affected work rather than encouraging improvised packaging solely to keep a throughput number green. Record the stockout so repeated packaging shortages can be repaired at their source.
Early cartonisation and packing-station cartonisation see different evidence
An early recommendation can help plan picking containers, labels and downstream workload. It depends on the recorded item dimensions, quantities and packing rules. The benefit is preparation; the limitation is that the final physical set may not yet be present.
At the packing station, the operator can see the actual products and packaging. That can reveal a bulge, changed outer pack or missing item that the earlier plan did not represent. The trade-off is that a late decision can require repacking or changing an already prepared shipment structure.
A useful design can use early planning with final physical verification. The two stages should cooperate. The final station should not silently repair every weak recommendation without returning evidence to the data and rules that generated it.
A heuristic is a decision procedure, not proof of the global optimum
A practical system may evaluate cartons in an ordered sequence, group selected items, apply capacity tests and stop when it finds a feasible arrangement. That can be entirely useful without establishing that no better arrangement exists among every possible grouping and orientation.
Ask what the selected method optimises and what it leaves out. Minimum parcel count, smallest external cube and lowest total delivered cost are not identical objectives. A method designed for one should not be praised as universally optimal for the others.
The operating standard should be fit for purpose: reliable enough for ordinary orders, clear about its constraints and able to route exceptions. A bounded, understandable method that produces dependable packages can be preferable to an opaque score that nobody can reconcile with the packing bench.
A packer override is evidence to examine
The packer selects a different carton from the recommendation. That may reveal better physical knowledge, a packaging stockout, a changed item, an unclear instruction or an unnecessary habit. The override alone does not tell us which explanation is correct.
Capture enough reason to distinguish recurring patterns. If one product is repeatedly moved to a larger box because its recorded dimensions exclude a protective sleeve, the source data needs correction. If the recommended carton is never available at one station, the replenishment process needs attention.
The Logistics Reason Codes guide explains how to classify these observations without turning a label into blame. The objective is fewer unsupported recommendations and fewer unexplained deviations.
A changed box can require a changed shipment record
When the physical packing differs from the plan, check which downstream facts changed. The completed dimensions, weight, parcel count, contents association or carrier selection may now be different.
A two-parcel result should not leave a one-parcel declaration active without reconciliation. A replaced label should not leave an obsolete identity available for another scanner to read. Where an advance shipping notice includes the package structure, the current advice should describe the actual shipment.
This connects cartonisation to Advance Shipping Notice and Logistics Unit Labels. A good packing decision is incomplete until the network can identify and interpret what was actually packed.
A worked order: the small box that fits only before protection
The following scenario is invented. An order contains two flat boxed products, each measuring 30 × 20 × 5 centimetres. An allowed stacked arrangement produces a combined product envelope of 30 × 20 × 10 centimetres before any outer protection.
Candidate A has internal dimensions of 30 × 20 × 10 centimetres. It matches the nominal envelope exactly. Candidate B has internal dimensions of 32 × 22 × 12 centimetres. For this example only, assume an authorised protective arrangement requires one centimetre of clearance on each outer side of the combined stack. Candidate B meets that dimensional requirement; Candidate A does not.
This is a geometry demonstration, not a recommendation that one centimetre of cushioning is adequate for any real product. A qualified packaging assessment would determine the material, clearances and route suitability.
The system recommends B. At the station, the packer discovers that one product now has a raised closure that increases the actual stacked envelope. The recorded dimensions describe an earlier package version. Forcing the products into B would preserve the recommendation while breaking the physical requirement.
The packer uses the approved exception route. The affected item is measured, its identity is checked and a suitable authorised alternative arrangement is selected. The completed parcel is then measured and weighed in its actual closed state. Its label and shipment record are generated or corrected accordingly.
The discrepancy returns to the master-data owner. The old package dimensions are not silently overwritten in a way that misdescribes earlier shipments; the relevant current product configuration is corrected. Open orders using the same weak assumption are reviewed as appropriate.
The immediate order now has a valid package. The next order benefits only when the correction reaches the selection process. Without that return, a capable packer can keep protecting customers while the software repeats the same mistake indefinitely.
Test ordinary orders, difficult orders and changed orders
A cartonisation trial should include more than a few neat rectangular products. Select representative order families and cases known to create difficulty: long items, mixed shapes, protected products, authorised splits and unavailable carton sizes.
Check physical feasibility, packing time, protection and agreement between planned and actual shipment structure. Follow the result far enough to see receiving or damage evidence appropriate to the trial. A package that closes at the bench has passed only one boundary.
Keep the trial’s limits visible. Successful results for one product family do not establish suitability for every route or handling condition. Introduce new rules through controlled change and retain an exception path for situations the validated design does not cover.
Do not confuse less empty space with a complete sustainability result
Reducing external package size can reduce the space a shipment demands. That is a useful operational mechanism. It does not by itself quantify a reduction in vehicle trips, total material use or emissions.
A smaller carton could require a different insert, create more parcels or affect damage and returns. Whether freed cube changes transport activity also depends on the wider load and network. State the improvement actually observed rather than converting a volume percentage into an unsupported environmental claim.
The defensible sequence is to measure the package change, examine protection and complete-service effects, then assess wider benefits with the required evidence. Sustainability should follow the real delivery system rather than stop at an attractive picture of a tightly packed box.
Measure the decisions the system gets wrong
Track recommendations that cannot be packed as proposed, authorised overrides, carton stockouts, unexpected splits and differences between planned and measured parcel dimensions. These measures identify where the selection process is losing contact with the physical operation.
Pair them with complete-order outcomes: packing time, correct contents, damage, receiving issues and total relevant cost. A smaller box is not success when it increases breakage. A low override rate is not success when packers are discouraged from reporting an unsuitable recommendation.
Compare like with like. A period with smaller orders can produce better cube utilisation or lower parcel cost without any rule improvement. The collection’s Logistics Benchmarking guide explains why the workload and measurement boundaries must remain visible.
Questions that keep the selection honest
Is cartonisation the same as intermodal containerisation? No. This article concerns selecting packaging arrangements for products and orders. Some warehouse software uses the word containerisation for that process. The separate logistics topic of intermodal containers concerns standard transport units moving across modes.
Should the smallest carton always win? Only among configurations that meet the actual requirements, and only when that criterion fits the objective. Protection, handling, parcel count and the receiver’s needs can justify another feasible option.
Can the packer change the recommendation? The operation should provide an authorised exception route. A physical mismatch needs a response, but the changed package and its downstream records must remain controlled.
What is the best first improvement? Investigate one recurring recommendation failure. Establish whether its cause is geometry, product data, protection rules, grouping or packaging availability. Repair the demonstrated mechanism before expanding the carton catalogue or replacing the entire system.
Evidence, calculations and scope
The primary software references are Oracle’s Cartonization, Consolidation, and Packing, Microsoft’s Containerization and Oracle’s Defining Storage Containers. They show documented configurations and data requirements. They are not independent performance trials or universal packaging specifications.
The volume, diagonal and stacked-product examples are original mathematical constructions. The clearance assumption belongs only to the stated hypothetical example. No carrier price, generic protective thickness or guaranteed emissions reduction is asserted. Real packaging and specialist handling decisions require the relevant competent owners. Editorial evidence review: 5 September 2026.
Choose for the journey, then verify the parcel
Return to How Logistics Works for the complete route from available goods to usable receipt. Cartonisation sits at an important boundary within it: several products become one or more physical packages that the next organisation must handle.
A good cartonisation decision produces a feasible, properly protected and correctly identified shipment—not merely a high fill percentage. The selection earns trust when its assumptions match the objects, its constraints match the service and the completed parcel gives the downstream network the same account of reality that the packing bench can verify.