Carton Sample Approval: What a Seven-Day Sample Proves — and Doesn't

The sample arrives on day nine. It is square, the product drops in with a satisfying fit, the print looks right under the office lights, and you sign the approval scan the same afternoon. Five weeks later the production cartons come off the container and a third of them will not square up on the packing line: the corners bow, one crease has cracked along the outer liner, and the joint sits two millimetres out on every box in the batch. Nothing about your sample predicted any of it.

Carton sample approval fails in this specific way because a sample carton and a production carton are made by two different manufacturing processes that happen to produce the same shape. The sample is usually cut one at a time on a sample table with an oscillating knife and a creasing wheel, from whatever board is at hand, and closed by hand with tape or a bead of glue. The production carton is struck by a rotary die or slotted and scored between matched wheels at speed, on board made or bought to your grade, and joined by a stitcher or an automatic folder-gluer. Almost everything you actually care about lives in the gap between those two descriptions.

That does not make the sample useless. It makes it a test of one thing — geometry — that people routinely treat as a test of everything.

What the seven-day sample genuinely proves

A well-made sample settles a short and valuable list:

• Fit. Your product goes in, with the clearance you specified, and comes out again. Load it with the real product, not a mock-up.
• Internal dimensions and panel layout. Measure the flat blank before anyone erects it. Crease-to-crease distances are the numbers that matter, and they are easier to check flat.
• Structure and closure logic. The flaps meet or they gap; the tuck engages or it does not; the insert seats or it fouls a crease.
• Artwork placement. Which panel carries which element, whether the barcode clears the crease, whether the legal copy fits.
• A shared physical reference. From this point on, both sides can point at an object instead of describing one.

That is a real return on seven days. It is roughly a quarter of what buyers assume they are getting.

The seven things the sample cannot tell you

The pattern is consistent: the sample proves the design, and the production tooling proves the manufacture. Approving one and assuming the other is the standard way this goes wrong.

The question that changes the whole conversation

Ask the factory one question before the sample is even made: is this sample cut on the production die and run on the production line, or is it cut on a sample table?

Both answers are legitimate, and both are common. Sample-table cutting is faster, cheaper and does not commit anyone to tooling, which is why a seven-day sample turnaround usually implies it. Cutting a real die first is slower and costs money up front, but the sample it produces is the actual product. What matters is that you know which one you are holding, because the answer determines what your approval means and what you still have to gate later.

If the sample is table-cut, your approval signs off geometry only, and you need a second gate at first-off. If it is die-cut on production tooling, your approval can carry more weight, but you should still see a run-start check because board, register and score settings all vary between a tooling trial and a live run.

For a high-volume or structurally awkward carton it is worth asking to pay for a die-cut sample outright. You are buying the tooling early and taking a tooling trial as your sample. Ask what that does to price and lead time, and ask who retains the die afterwards — the second question is far easier to negotiate before the die exists than after.

A related question with the same shape: is the print on this sample the production print process? A colour box sampled on a digital printer tells you the layout. It tells you nothing about how the ink will lie on your liner. Ask what the colour-control method will be on the run — an approved wet proof, a drawdown on the liner, a spectrophotometer reading to a stated tolerance, or an eyeball match to a retained master — and get the answer in writing before you approve anything on colour.

A four-gate approval structure that actually holds

Replace the single "sample approved" event with four gates. This is the part you can write into a purchase order.

Gate 1 — Structural sample, plain board. Purpose: fit, internal dimensions, closure, insert clearance. Request it shipped flat and unassembled so you can measure crease positions and erect it yourself; a factory-erected sample hides crease bow and joint squareness behind a finished shape. Approve fit and structure explicitly, in those words, and state that colour and manufacture remain unapproved.

Gate 2 — Colour proof on the real substrate. Purpose: ink on your liner. A drawdown or press proof on the specified liner, viewed under a stated light source, with your Pantone references written on the back. Sign two copies; keep one.

Gate 3 — First-off from production tooling. Purpose: the manufacture. The first 20 to 50 pieces off the real die, real board and real joining method, checked before the run continues. Ask for photographs of a first-off carton with a tape measure laid across it, the board caliper measured, the joint photographed, and the run's board combination confirmed. This is the gate that would have caught the bowed corners in the opening paragraph, and it is the one most buyers do not have.

Gate 4 — Pre-shipment check against the retained master. Purpose: the run, not the first box of it. Random cartons drawn across the production, checked for internal dimensions, squareness, score cracking, joint integrity, print register and colour against the signed master. The general mechanics are covered in pre-shipment inspection; for cartons, the specific measurements to demand are ID, diagonal squareness, board caliper and joint overlap.

Signing a master sample so it cannot drift

The retained master — the "gold sample" — is only useful if it is tamper-evident and if both sides hold one.

• Sign and date the sample flat, across the manufacturer's joint and across at least one crease, so panels cannot be swapped.
• Photograph the signature and the whole blank, and email the photographs to the factory the same day, so the file has a timestamp neither side controls alone.
• Keep one counter-sample in your own office, sealed. A master that lives only in the supplier's showroom settles nothing.
• Write the reference into the purchase order: production to be manufactured against approved sample No. ___ dated ___; first-off approval required in writing before the balance of the run proceeds.

That last clause is worth more than any amount of email agreement, because it turns a courtesy into a contractual condition on the run.

The clock: why sample rounds cost more than they look like

Bulk production lead time in this industry is generally quoted as running after artwork and sample approval, not from the purchase order. That means every revision round pushes the delivery date by its own duration plus the shipping time each way — a seven-day sample that needs one revision is not seven days, it is closer to four weeks door to door once courier legs and an internal review are counted.

Two practical consequences. First, budget two sample rounds into the critical path rather than one, and be honest about it in your own planning, because the alternative is a rushed approval on a sample you have not properly measured. Second, front-load the questions: pre-press format, colour control, joint type and whether the sample is table-cut or die-cut all cost nothing to ask at RFQ stage and each one can save a round later. The general timeline logic is set out in sample to mass production, and the mechanics of running a first sample order in the sample order process guide.

Common questions

The factory sent a beautiful sample. Should I be suspicious?

Not suspicious — but do not read it as evidence about the run. A sample table with a skilled operator produces cleaner creases than a production line does, because it is slow, hand-fed and unhurried. A sample that looks better than a production carton is the normal outcome, not a warning sign. The warning sign is a supplier who will not tell you how the sample was made.

What should I measure on a carton sample, and with what?

A steel rule or tape for crease-to-crease distances on the flat blank; a caliper or micrometer for board thickness; a diagonal measurement corner to corner on the erected box for squareness. Measure the internal dimensions with the box erected and loaded with your real product, not empty, and re-measure after 24 hours. Write the readings down; a sample you did not measure is a sample you did not check.

How do I stop the second production run drifting from the first?

Keep the signed master, and reference it by number on every subsequent purchase order together with the board combination, the flute, the grammages and the strength target. Repeat orders drift when the paperwork gets shorter, not when the factory changes its mind. Restating the full board spec on every order is the cheapest control available, and it is the same discipline that makes ECT and burst grades comparable between runs.

What to ask before you sign a carton sample

1. How was this sample made — table-cut or production die; hand-joined, stitched or machine-glued?
2. Where did the sample board come from, and is it the same combination quoted for the run?
3. What is the production print process and the colour-control method, and can I have a drawdown on the actual liner?
4. What are your slot, score and squareness tolerances, in millimetres, in the quotation?
5. Will you send a first-off from the run — photographs with a tape measure, board caliper, joint close-up — and hold the run until I approve it?
6. Which inspection records travel with the shipment — incoming board check, first-off form with a signature and a time on it, in-process check sheet — and can I see the blank forms and one completed example from a recent job now? Any certificate a supplier names is a separate exercise: ask for it in full, with number, scope, issuing body and validity dates, and verify it with the issuer rather than with the supplier.
7. Who signs and retains the master sample, and will you accept a counter-sample held by me?

For a concrete case to run those questions against, the CMH listing for Jindong Packaging describes a Shenzhen paper-packaging manufacturer founded in 2007 in Xinqiao, Bao'an, working ODM on a full-material basis and building to your drawing or sample, quoting sampling at about seven days and bulk production at roughly fifteen working days after artwork and sample approval, from 1,000 pieces with tiers at 3,000 and 5,000 on T/T terms, with printer-slotters, slotting and creasing machines and gluing and stitching machines listed as in-house equipment. What that listing does not say is how the seven-day sample is produced — table-cut or on production tooling — or where the sample board is sourced. That is question one, and it is question one for every carton supplier you approach. The answer decides whether your signature on the sample is an approval of the design or an approval of the run, and those are not the same document.