How Do You Specify a Custom Cooler Bag That Actually Holds Temperature?

What Are the Standard Sizes and Capacities for Custom Cooler Bags?

Cooler bags are quoted in cans, not litres, because the buyer briefs a pack-out. A 12-can bag means 12 x 330 ml cans lying on their side in two rows of six, and the usable volume that produces is about 8 litres. Sizes above 24 cans stop behaving like a bag and start behaving like a box, which changes the freight calculation more than it changes the unit price. Before you quote anything, fix the external dimensions, the wall build-up and the mouth opening, in that order.

External dimensions alone are not a specification. Two bags both listed as 360 x 230 x 250 mm can differ by 1.5 litres inside, because one puts 8 mm foam in every panel and the other puts 8 mm in the base and 5 mm in the side walls. Write the build-up into the tech pack: external dimensions, foam thickness per panel, liner seam allowance, and finished internal dimensions with a tolerance. Internal dimensions are what the brand will be asked about when a customer says a 12-pack does not fit.

The load column matters more than the volume column. A 30-can bag rated for 30 cans plus four 400 g gel packs carries 9.4 kg, and the handles, the shoulder strap and the base panel all have to be specified for that number rather than for the empty weight. Plan the strap width and the handle bar-tack pattern against the filled weight and then add a margin, because a drop test will be run on the loaded bag and the carton will be stacked by the loaded weight.

How Much Insulation Do You Actually Need: 5 mm, 8 mm or 10 mm?

Foam thickness is the most over-specified and least-tested variable in this category. Buyers ask for 10 mm because they assume thicker means longer. It does, but the return is not linear, and the penalty arrives three ways at once: weight, carton cube, and a lid that no longer closes flat. On a 20-can bag, moving from 5 mm to 8 mm roughly doubles the ice-retention window. Moving from 8 mm to 10 mm buys perhaps a third again, at a higher weight cost than the first step.

Two foam families cover the market. Polyethylene (PE) closed-cell foam in 5 mm and 8 mm is the workhorse: it laminates, it welds, and it is available in every density the trade asks for. EVA closed-cell foam is denser, quieter in the hand and holds a corner better, which is why it appears in structured bags; it costs more per cubic centimetre and it recovers less well after being crushed in a container. Neither is a substitute for the other in a pattern that was cut for the other.

The other half of the thermal system is the facing. A radiant barrier — an aluminised film on the food-cavity side — reflects radiant heat back at the load and is the least expensive hour-for-hour gain in the category; it is worth roughly two millimetres of foam at a fraction of the cost. What it will not do is survive a fold. Specify it on the flat panels and on the lid, and leave it off the corners where the bag collapses for storage.

None of these numbers is a promise until the test conditions are written down. A bag holding 20 chilled cans with three frozen gel packs in a 25 °C room behaves differently from the same bag holding six cans and a great deal of air, because trapped air is what moves the heat. Put four numbers in the tech pack: ambient temperature, load, gel-pack mass and pre-chill time. Without those, an ice-retention claim cannot be falsified, and two suppliers can both claim twelve hours while delivering half of it.

PEVA, TPU or PUL: Which Liner Clears Which Food-Contact Rule?

The liner is the food-contact surface, which makes it a regulated component rather than a trimming. In the EU the framework is Regulation (EC) No 1935/2004: the material must not transfer constituents to food in quantities that could endanger human health or change the composition of the food. In the United States the equivalent route for the polymer is the indirect food additive listing in FDA 21 CFR 177, and the polymer class your liner is made from has to appear there, or the converter has to hold a letter saying it does.

Three liners cover most of the market. PEVA — polyethylene vinyl acetate — is the default: it welds, it is available with a food-contact declaration, and it is the least expensive of the three. TPU is the flexible one; it stays soft down to roughly -30 °C, which is what frozen-grocery delivery needs, and it welds cleanly. PUL is a polyester substrate with a polyurethane film bonded to one face; it is the pleasantest to handle and the most likely to delaminate at a weld if the film gauge is wrong for the dwell time.

Two documents matter at goods-in and neither is optional in a serious programme. The first is the food-contact declaration from the film converter, naming the polymer and the regulation it is cleared under, and dated within the year. The second is an overall migration test report, run on the finished liner with a simulant, at the contact time and temperature the bag will actually see. A supplier who can produce a certificate for the resin but not for the converted film is telling you they have not tested the thing that touches the food.

There is a cost reason to settle this before the sample rather than after. Migration testing is a fixed-cost laboratory exercise. Asking for it after the liner is already tooled into a programme means paying for it twice if it fails, then paying again for a new liner and a new sampling round. Our production team asks for the declaration at the quote stage, because a liner swap at that point costs nothing and a liner swap after sample approval costs the whole sampling window.

Should the Liner Be RF Welded, Hot-Air Welded or Stitched and Taped?

A cooler bag has two failure surfaces and the seam is the one that fails first. Water in the cavity from melted ice has to stay in the cavity. Three methods compete: radio-frequency welding, hot-air or hot-wedge welding, and conventional sewing with a seam tape. Each works on some materials and not others, and the choice is usually made by the liner rather than by the buyer.

RF welding melts the material itself by dielectric heating and produces the strongest and flattest seam in the category. It requires a polar polymer, which PEVA and TPU are and polypropylene is not. It also requires a die for every seam geometry, which is why it appears in volume programmes and not in 500-piece ones: the tooling has to be amortised. Hot-air welding needs no die and works on TPU and PUL, but the seam is wider and the operator variance is larger, and two operators on the same shift can produce two different seam strengths.

Sewing with a taped seam is the fallback, and it is not a defeat. On a 500-piece first run it is usually the right answer, because it avoids tooling entirely and it can be repaired. What it cannot do is make a liner watertight on its own, because the needle holes are the leak path and the tape is only what closes them. If the brief says the bag must hold melt water, specify the tape width and the tape adhesion test, and then check it: tape applied to a curved corner without enough heat is the single most common leak origin in this category.

Whichever method you choose, put a leak test in the inspection plan and not only in the sample report. A 24-hour static hold with the bag filled to a marked level, standing on a lined floor, with each unit numbered, will find the seams that hot-air welding missed and will do it before the goods are cartoned. It needs no equipment, it costs the floor space, and it catches the defect customers actually complain about.

Why Do Zippers Fail on Cooler Bags More Often Than Anywhere Else?

On most bag categories the zipper is a closure. On a cooler bag it is also a thermal component, and it is the coldest part of the bag, which is where condensation forms and where the load sits. Three failure modes account for most warranty claims in this category, and none of them is the slider breaking.

The first is coil corrosion. A cooler bag spends its working life wet on the inside, with salt from melted ice water and from the food itself sitting on the teeth. A nickel-plated coil in that environment will show corrosion inside a season. The fix is not exotic: a polyester coil with a corrosion-resistant slider and a tape that does not wick, or for a welded build, a TPU-coated tape. It costs more per metre than a generic coil and it removes the most visible failure in the category.

The second is the zipper being specified for the empty bag. A lid zipper on a 30-can bag has to close against 9.4 kg of load pushing the walls outward, and the walls will move because the foam does. A #5 coil will close the bag when it is empty and will start to separate at the corner radius when it is full. Specify the slider and chain size for the loaded geometry, and where the lid turns a corner, use two sliders meeting on a flat section rather than one slider that has to negotiate the radius.

The third is the thermal bridge. A standard coil zipper is a continuous line of uninsulated material across the lid, and on a bag with a radiant barrier it is the only uninsulated path left. The trade answer is an insulated zipper flap covering the line, or a roll-top mouth with the zipper used only for access. Neither is free, and both have to be decided at the design stage, because a flap has to be patterned in rather than added later.

A fourth is worth naming even though it is not a zipper: handle attachment. On a 30-can bag carrying 9.4 kg, the handle webbing pulls on a panel that is mostly foam, and foam has poor tear resistance. The pattern answer is a webbing loop that continues under the panel and is bar-tacked at two points, not a handle sewn to the surface. Stitch type and stitch density on that load path are specified under ASTM D6193, which is the reference to cite when the inspector asks what the tech pack means.

Which Parameters Drift Between the Approved Sample and the Second Run?

The approved sample is a photograph of one day of work. The second run is a different day, a different foam roll, a different operator on the welding head and a different dye lot. The things that move are predictable, and if they are named in the tech pack they can be measured instead of argued about.

• Foam density, not foam thickness. Thickness is visible and gets checked; density is not, and it is what decides whether a panel recovers after compression. A 25 kg/m³ roll and a 33 kg/m³ roll are both 8 mm.
• Liner gauge. A 0.15 mm PEVA and a 0.12 mm PEVA weld differently and tear differently, and the difference is invisible until a seam fails.
• Weld temperature and dwell. These are set by an operator, and they drift across a shift and across a season as the ambient in the room changes.
• Colour, in two places: the shell fabric against a Pantone reference, and the trim — webbing, binding, zipper tape — which usually comes from a different dye house entirely.
• The print. A screen print that was crisp on the sample will be heavier or lighter depending on how long the ink has been sitting in the mesh, and a heat transfer will change gloss with the press dwell.

The practical control is a golden sample that is kept and a tolerance sheet that is attached to it. The full argument is set out in keeping reorders consistent; the short version is that tolerance has to be written as a number with a measurement method before it has any force. This is the least expensive control available in the category and the one most often skipped.

The second control is a pre-production sample built from the actual bulk materials rather than from the materials that were handy. There is a gap between the two and that gap is where the drift lives. On a 35-50 day bulk window there is time to make one. Asking for it after the goods are finished is asking for the 35-50 day window a second time.

What Does an ISO 2859-1 AQL 2.5 Inspection Actually Do on Your Lot?

AQL is not a quality level despite the name. It is the defect rate the sampling plan is prepared to accept most of the time. Under ISO 2859-1 at General Inspection Level II and AQL 2.5, a lot of 3,201 to 10,000 pieces is inspected by pulling 200 pieces; the lot passes if 10 or fewer defective units are found and fails at 11 or more. That is the entire mechanism, and it belongs on the purchase order, because a supplier who quotes "AQL 2.5" without the lot size, the level and the plan has not actually said anything.

The share-of-lot column is the one buyers never read and should. On a 500-piece order you inspect between 6.7% and 16.0% of the units; on a 35,000-piece order you inspect under 1%. That is not a fault in the standard, it is what the standard is for: it is a lot-acceptance tool, not a guarantee that every unit is sound. If the consequence of one faulty liner seam is a food-safety incident or a retailer chargeback, sampling at General Level II is not the right control, and the conversation has to be about 100% leak testing on the line instead.

The other half is defect classification, and it is where most inspection instructions are useless. A critical defect is one that makes the article unsafe or non-compliant — a liner with no food-contact traceability, or a migration result that fails. A major defect is one that would cause a return: a leaking seam, a zipper that will not close, a handle pulling away from the panel. A minor defect is a cosmetic one a customer would notice and not return. AQL 2.5 normally governs major defects; critical defects are zero-tolerance at any AQL, and a separate, tighter acceptance number for minors is common. Write all three into the inspection instruction with examples from your product, or the inspector will classify for you.

One number to be aware of, because a competitor may quote it at you: AQL 1.5. It is a real AQL value and it is stricter, but 1.5 is not among the AQL grades normally paired with ISO 2859-1 and ANSI Z1.4 in commercial practice, and moving to it buys a smaller accept number at the price of a much higher chance of rejecting an acceptable lot. The reasoning is set out in AQL 1.5 vs 2.5; the short version is that it changes who bears the risk, not how many defects leave the building.

What Is the Realistic MOQ, Sampling Window and Bulk Lead Time?

MOQ is 500 pcs per style and per colourway, and it can be discussed down to 300 on a repeat. The reason is not arbitrary. A colourway is a separate setup on the print line, a separate foam and fabric lay, and on a welded build, a separate set of welding parameters. Two colourways of the same bag at 250 pieces each is not the same order as 500 pieces of one, and it does not price the same.

Sampling takes 6-10 working days from an approved tech pack, and 12-15 working days for a complex build — a welded liner, a new die, a moulded base. That clock starts when the pack is final, not when the first email arrives. The two things that stop it are an incomplete tech pack and a change of direction after the sample is cut. Neither is a production delay, but both get recorded as one.

MOQ

500 pcs per style / colourway (300 on repeat)

Sampling

6-10 working days (12-15 for complex builds)

Bulk production

35-50 days from sample approval

Inspection

AQL 2.5, ISO 2859-1, General Inspection Level II

Payment

T/T 30/70

Incoterm

FOB Xiamen (CIF / DDP available)

Capacity reference

7 production lines, 149 machines, 200,000 pcs per month

Operating since

2014

Bulk is 35-50 days from sample approval, and the honest reading of that range is that it is a distribution rather than a promise. The spread is not slack. It covers material lead time — a food-contact liner in a specific gauge is not a stock item — the welding or sewing queue behind the orders already booked, and the inspection and packing window at the end. We do not promise the fastest 20 days. We promise a delivery date you can calculate, and the way to calculate it is to work backwards from the milestones rather than to negotiate the headline number.

The way to use the window is to book it. A 35-50 day bulk schedule plus transit is a planning input, not a surprise. The buyers who miss a season are the ones who treat the 50 as a worst case that will not happen rather than as a case that happens often enough to plan for. If the goods are for a fixed date, work back: landed date, minus customs entry and inland delivery, minus transit, minus the 35-50 days, minus the 6-10 working days of sampling, minus your own approval time. The date you arrive at is the date the tech pack has to be final.

What Actually Drives the Unit Cost of a Custom Cooler Bag?

The unit price of a cooler bag is not driven by the shell fabric. The fabric is usually a modest share of the bill of materials. What drives it, in order: the thermal build, the liner, the closure hardware, the labour in the seam, and only then the shell.

• Foam, by volume. Foam is bought by the cubic metre and converted by the square metre, so a 10 mm build costs roughly double a 5 mm build on material — and then costs again in freight, because the bag occupies more of the carton.
• The liner, by material and by method. TPU film costs multiples of PEVA per square metre, and a welded liner costs less per seam than a sewn-and-taped one at volume but more at 500 pieces, because the die has to be paid for either way.
• Closure hardware. A corrosion-resistant #8 coil with a covered flap costs multiples of a generic #5 coil, and it is the difference between a bag that survives a season and one that does not.
• Seam labour. Every metre of weld and every metre of tape is paid for twice: once in labour and once in the time it occupies the line.
• Carton cube. A 30 L bag that does not fold ships mostly air, and air is what the freight forwarder charges for. A folding design with a crease panel can halve it.

The two costs that are invisible on a quote and real on a P&L are tooling and testing. A welding die, a moulded base and an embossing plate are one-offs that appear on the first invoice and nowhere else; a migration test and a drop test behave the same way. Ask for both to be quoted separately from the unit price, because a supplier who folds them into the unit price is quoting a number that will not hold when you reorder at a different volume.

There is one lever that is usually available and usually missed. Freight is charged on the greater of weight and volume, and cooler bags are volumetric. Nesting the empty cartons, folding the bags flat for the outbound leg, and specifying the carton to the folded dimensions rather than the erected ones will move the landed number further than a 3% negotiation on the unit price. That is what the landed cost calculator is for: the unit price is one line of eight.

Which Five Design Mistakes Cause Most Cooler Bag Complaints?

Every one of these is a decision that costs nothing at the design stage and costs a sampling window after it. They are also all things a third-party inspector will not catch, because an inspector checks against the approved sample, and these mistakes will be in the approved sample too — the bag was made the way the tech pack said. The control is the tech pack, not the inspection.

A fifth, which is not a design mistake but produces the same result, is a missing test condition. An ice-retention claim with no ambient temperature, no load and no gel-pack mass is not a specification, and two suppliers can both meet it. If the bag is sold on a hold time, put the hold time into the acceptance criteria along with its conditions, and make the pre-production sample prove it.

How Do You Classify a Cooler Bag for Customs and Land the Cost?

Insulated bags generally enter under HS heading 4202, and the subheading follows the outer surface. A cooler bag with a textile or plastic outer surface normally falls in 4202.92; leather outer surfaces fall elsewhere in the heading. Duty rates are set by the destination tariff and they change, so treat any rate you are quoted as an estimate and confirm it against the official tariff of the importing country or with a licensed broker before you price the programme.

The classification is worth getting right for a reason that has nothing to do with duty. Preferential origin claims and duty-relief arrangements are tied to specific tariff lines, and a misclassification discovered at entry is a post-entry amendment rather than a conversation. If the bag ships with a branded food product inside it, or as part of a meal kit, the treatment of the set can differ from the treatment of the empty bag — and that is a question for the broker, not for the supplier.

Use the HS code finder to pin the line, then the Incoterm decision tool to test whether FOB Xiamen is the right allocation of risk on your lane. On some lanes the buyer is better served by CIF or DDP; on others the buyer is paying the seller a freight margin to avoid a phone call. FOB Xiamen is the default and it is quoted that way, and it is not always the right answer.

Then the arithmetic, in eight lines: unit price, inland haulage to the port, origin charges, freight on the greater of weight and volume, insurance, destination handling, duty at the confirmed rate, inland delivery. There is a calculator for it, and the reason it exists is that the two lines buyers forget — destination handling and duty — are the two that sit outside the supplier quote.

What Temperature-Hold Test Belongs in the Tech Pack?

The test is inexpensive and almost nobody writes it down, which is why ice-retention claims in this category are close to unenforceable. A usable specification has six fields: the load, the gel-pack mass and its pre-freeze temperature, the ambient conditions, the start time, the pass criterion, and the number of units tested.

• Load: not "12 cans" but "12 x 330 ml at 4 °C, plus two 400 g gel packs frozen at -18 °C for 24 hours".
• Ambient: "25 °C ± 2 °C, in shade, standing on a wooden pallet and not on concrete". The surface matters; concrete is a heat sink and changes the result.
• Pass criterion: "temperature at the geometric centre of the load stays at or below 8 °C for 8 hours". Centre-of-load probe, not a probe near the wall.
• Units: three, from three different cartons, tested in the same week. One unit is an anecdote.
• Report: a time and temperature log, not a statement. If it was not logged, it did not happen.

A second test belongs beside it, and it is the one that prevents the complaint rather than the return: a 24-hour static leak test, filled to a marked level with the lid closed, one unit in ten of the pre-production sample, each unit numbered and the floor checked. The two together — hold and leak — cover the two things a cooler bag is bought for. Both can be run without a laboratory, which means both can be run on every repeat and not only on the first one.

If the bag is sold into the EU as a food-contact article, the documentation set travels with it. The record to keep is the food-contact declaration for the converter of the liner, the overall migration report on the finished liner, the declaration under Regulation (EC) No 1935/2004 for the finished article, and the traceability link between the three. It is a paperwork exercise, and it is what makes an incident survivable rather than terminal. The references are listed at the foot of this page.

Frequently Asked Questions

People Also Ask