---
name: "packaging_design"
description: "Brief and judge packaging that works on a shelf and on a phone: structure and hierarchy decisions, claims and required information checked before design, materials and print limits, and an unboxing check on the pack a customer actually receives. Use when briefing a new pack, reviewing pack artwork, preparing a print proof, or fixing a pack that fails in store, in search results, or in transit."
---

# Packaging Design

## Purpose

Get a pack briefed, checked, and proofed in the right order so the expensive mistakes never reach print. Most pack failures are not taste failures. They are order failures: artwork finished before the claims were cleared, a dieline changed after the design was built on it, a barcode that looks fine and will not scan, or a front panel that reads at full size and disappears as a phone thumbnail.

This skill runs five passes in sequence: brief, claims and required information, hierarchy for shelf and phone, materials and print, and a physical unboxing check on a filled production pack. It ships with a brief template, a claims and information checklist, a print proofing checklist, and a worked review that shows all five passes on one product.

## Workflow

### 1. Write the brief before any design work

- Fill in `templates/packaging-brief.md` completely. A blank field is a decision nobody has made yet, and it will be made later by the printer, the filler, or the retailer, usually badly.

- Lock these items first, because changing any one of them restarts the design: product and variant, fill weight or count, pack format and dimensions, dieline source and version, sales channels (shelf, marketplace, direct ship, or a mix), markets where it will be sold, and the date the pack must be in hand.

- Name the one job of the pack in that channel. On a shelf the job is to be found and understood in seconds. In a marketplace grid the job is to be legible at thumbnail size. In direct ship the job is to arrive intact and open cleanly. A pack sold in all three places needs all three tests, not an average of them.

- Record hard constraints in the brief, not in chat: retailer requirements, barcode type and number, language requirements, allergen or warning needs, sustainability targets the company can actually prove, and the unit cost ceiling.

- State what is out of scope in this round. A redesign that also changes structure, supplier, and claims in one pass cannot be judged, because nobody can tell which change caused a problem.

### 2. Clear claims and required information before design starts

- Run `references/claims-and-information-checklist.md` line by line. Do not start layout until every front panel claim has a substantiation entry and every required element has a home panel assigned.

- Identify the governing rule set for the actual product and market first. Food, supplements, cosmetics, drugs, chemicals, and toys each carry different required information, and rules differ by country. The checklist uses United States food rules as the worked example because they are specific and public. Treat other categories as a prompt to look up their rules, not as a copy of the food rules.

- For United States packaged food, reserve space for these elements before choosing type sizes: statement of identity on the principal display panel, net quantity in the bottom 30 percent of that panel parallel to the base, ingredient list in descending order by weight, disclosure of the nine major allergens (milk, eggs, fish, crustacean shellfish, tree nuts, peanuts, wheat, soybeans, sesame), Nutrition Facts, and the name and place of business of the manufacturer, packer, or distributor. Those items come from 21 CFR Part 101 and the allergen laws it sits alongside.

- Size the net quantity statement from the panel area, not from taste. Minimum letter height is 1/16 inch for a panel of 5 square inches or less, 1/8 inch over 5 to 25 square inches, 3/16 inch over 25 to 100 square inches, 1/4 inch over 100 to 400 square inches, and 1/2 inch over 400 square inches. Run `scripts/pack_check.py netqty --pdp-sqin <area>` to get the figure instead of guessing.

- Treat every claim as a promise that needs evidence on file before the pack prints. That covers nutrition and health claims, comparative claims such as 30 percent less sugar, functional claims, origin claims, and environmental claims.

- For environmental claims, follow the FTC Green Guides at 16 CFR Part 260. Hold competent and reliable scientific evidence before the claim runs, qualify any claim that covers only part of the product or only the pack, state percentages for recycled content, and avoid broad unqualified words such as green or eco-friendly that are hard to substantiate. A claim that is literally true and still misleads, for example bragging about removing a substance that is already banned, fails the same test.

- Give each claim one of three verdicts in the checklist: cleared with evidence named, reworded to what the evidence supports, or cut. No claim ships on a verbal assurance.

- Route regulated categories to a qualified reviewer before print. This skill catches missing elements and unsupported wording. It does not replace regulatory or legal review.

### 3. Set the hierarchy for shelf and phone

- Order the front panel for a shopper who will not read it. Research on shelf behaviour puts the decision window at roughly 3 to 5 seconds, often from 3 to 6 feet away, and colour registers before shape, symbols, and words. Build for that order: distinctive colour block first, then a recognisable shape or image, then brand, product name, and variant, then one differentiating benefit. Everything else belongs on a side or back panel.

- Limit the front panel to five jobs: brand, what it is, which variant, net quantity, and one reason to pick it up. If a sixth item wants front panel space, something already there has to leave or move.

- Run the distance test. Print or mock the front panel at real size, view it from about 6 feet, then from 3 feet, and write down what can actually be read at each distance. Product name and variant must survive at 6 feet. The benefit line must survive at 3 feet. Fine print is pick-up copy and should be designed as such.

- Run the shelf set test. Place the pack, physically or in a rendered shelf image, between its two or three closest competitors at real spacing. Judge contrast and colour ownership in that set. A pack that looks strong alone and vanishes next to a competitor has failed its main job.

- Run the thumbnail test. Shrink the front panel image to 150 pixels wide, the size it gets in many search grids, and check it again at 200 pixels. Brand, product name, and variant should still read, and a pale pack should not blend into a white grid. If it fails, simplify: fewer elements, larger product name, stronger contrast, and a coloured field or border where the channel allows it. See the worked review in `references/worked-pack-review.md` for a thumbnail failure and its fix.

- Check variant discipline across the range. Variants should differ by one clear system, usually a colour band plus the variant name in the same position and size on every pack, so a returning buyer can find theirs without reading the whole panel.

- Check accessibility basics on pack copy: body text with a contrast ratio of at least 4.5 to 1 against its background, and large display text at least 3 to 1. Run `scripts/pack_check.py contrast --fg <hex> --bg <hex>` for a ratio instead of judging by eye. Avoid setting type over busy photography where the ratio changes word by word.

### 4. Choose structure and materials for the route the pack travels

- Match the format to the job. A folding carton is single-ply paperboard that ships flat and prints well. It is a shelf pack and needs an outer shipper for transit. Corrugated board, a fluted layer between liners, is the transit format: it stacks, cushions, and survives conveyors. A rigid box is thick wrapped board that feels premium, does not ship flat, and costs the most per unit and in freight. Plenty of products need two of these, a folding or rigid pack inside a corrugated shipper, and the brief should say so plainly.

- Pick board by product weight, then confirm with the converter and a filled test. As a starting point used by converters, light items under about half a pound sit in roughly 250 to 300 gsm board, medium retail items up to about 2 pounds in roughly 320 to 400 gsm, and heavier items move to heavier board or corrugated. Grain direction, scoring, and moisture exposure can override the weight guide, which is why the filled test in step 5 decides.

- For corrugated, specify the system, not an adjective. Write the flute and strength, for example single-wall C flute with an edge crush value, instead of asking for a strong box. C flute is the common shipping default, B flute resists crush well, and E flute gives a smoother surface for printed mailers.

- Decide finishes last and price them honestly. Foil, embossing, spot gloss, soft-touch film, and windows each add a plate, a pass, or a hand step, and each one can hurt another goal: film can block a recyclability claim, a window can weaken a panel, and metallic inks can shift under different lighting. Record in the brief which finish earns its cost and what claim or function it supports.

- Check closures and seals as part of structure, not as an afterthought: how the pack opens, whether it recloses, whether tamper evidence is required for the category, and whether the opening method works for a person with weak grip. A pack that needs scissors and teeth to open will be remembered for that.

### 5. Build and proof for print

- Work through `references/print-proofing.md` before releasing any file. The short version follows.

- Build artwork on the current supplier dieline and name its version in the file. Keep the dieline on a separate non-printing layer, and put each finish, foil, emboss, spot colour, and white ink on its own labeled layer.

- Set colour for print, not screen: CMYK or named Pantone spot colours, with no RGB left in the file. Raster images at 300 dpi at final size. Fonts outlined or embedded.

- Apply the bleed and safe zone the printer specifies. A common carton specification is 3 mm of bleed on every cut edge with text and logos kept at least 3 mm inside the cut line. Some label printers specify less, so the printer spec in the brief wins over any default in this skill.

- Keep critical copy off folds, creases, scores, glue flaps, and seal areas, and keep it clear of corners that will bend. Words that look fine flat can split across a crease once the pack is folded.

- Place the barcode where it will stay flat and scannable. For EAN-13 and UPC-A, 100 percent magnification means an X-dimension of 0.330 mm and a total width of 37.29 mm including quiet zones. Stay inside the 80 to 200 percent magnification range, keep the quiet zones empty (EAN-13 needs 11 modules on the left and 7 on the right, UPC-A needs 9 on each side), and print it in one dark colour on a light background that is not red, orange, or pink. Run `scripts/pack_check.py barcode --magnification <percent>` for the dimensions, then scan a printed proof with a real checkout-style scanner or a verification app, not only a phone camera.

- Proof in this order and sign off each stage separately: flat digital proof for spelling, claims, and panel assignment; folded paper mockup at real size for orientation, hierarchy, and copy crossing folds; printed proof on the actual substrate for colour and finish; and a scan test for the barcode on that proof. Approving the flat proof alone is the most common cause of orientation and fold errors reaching production.

- Read the proof like a stranger. Check the product name, variant, net quantity, allergen line, dates and codes area, barcode number digit by digit against the source, and every claim against the cleared checklist. Spelling checks run by the person who wrote the copy do not count as a proof read.

### 6. Run the unboxing check on the pack a customer receives

- Test the finished, filled, sealed production pack, not a hand-made sample. Hand samples are glued better, filled more carefully, and handled less than production packs.

- Drop it the way transit will. As an in-house proxy before any formal testing, drop the filled shipper on one corner, then on edges, then on faces, from about waist height, then open it and record product damage, pack scuffing, seal splits, and insert movement. For formal assurance, match an ISTA procedure to the route, for example ISTA 1A, 2A, or 3A for parcel distribution, and agree pass and fail criteria before the test runs.

- Stack it. Put the weight of a realistic stack on the shipper for a day and check for crush, bulge, and leaning. A pack that survives a drop and collapses in a warehouse stack has still failed.

- Open it as a customer, on camera, in one take. Time it. Note the first thing seen, whether the opening order makes sense, how much packaging has to be removed before the product is reachable, whether instructions appear before or after the step they explain, and what is left to throw away or recycle.

- Score the unboxing on five plain questions: Did everything arrive undamaged? Did it open without tools or frustration? Was the product the first clear thing seen, or was it buried in filler? Did the inside of the pack match the promise on the outside, including counts, colours, and claims? Would the leftover material fit in a normal household recycling or bin routine in the market where it is sold?

- Feed every failure back to a specific earlier pass: damage goes to structure and materials, confusion goes to hierarchy, a wrong count or broken claim goes to the claims checklist, scuffed or shifted print goes to proofing. Fix at that pass and retest the physical pack. Do not patch a transit failure with a sticker that says handle with care.

### 7. Judge a finished design

- Score each pack against the same six questions and require evidence, not impressions: Can a shopper name the brand, product, and variant in 5 seconds at 6 feet? Can they do the same from a 150 pixel thumbnail? Is every claim on the pack in the cleared checklist with evidence named? Is every required element present, on the right panel, and at a legal size? Did the printed proof scan, fold, and read correctly on the real substrate? Did a filled production pack survive the drop, stack, and open test?

- Give the pack one verdict: ship it, fix and re-proof the listed items, or go back to the brief because the format or channel assumption was wrong. List fixes in the order of the passes above so the team fixes causes before cosmetics.

- Keep the signed proof, the cleared checklist, the brief, and the unboxing notes together as the pack record. The next redesign starts from that record, not from memory.

## Output Contract

Deliver, for each pack or redesign:

- A completed packaging brief using `templates/packaging-brief.md`, with dieline version, channels, markets, and cost ceiling filled in.
- A completed claims and information checklist using `references/claims-and-information-checklist.md`, with a verdict and evidence source for every claim and a panel assignment for every required element.
- A hierarchy test record: what was readable at 6 feet and 3 feet, the shelf set result against named competitors, and the 150 pixel and 200 pixel thumbnail result.
- A print proofing sign-off using `references/print-proofing.md`, covering dieline version, colour mode, bleed and safe zone, barcode dimensions and scan result, and proof stages approved.
- An unboxing check record: drop, stack, and open results on a filled production pack, with photos or video noted, and each failure mapped to the pass that owns the fix.
- A final verdict in one line: ship it, fix and re-proof, or back to brief, with the fix list in pass order.

## Operating Rules

- Claims and required information come before design, every time. Layout that starts before the checklist is cleared will be redone.
- The printer spec and the governing regulation for the actual product and market override every default number in this skill. Defaults here are starting points for common carton work.
- Judge the pack in context: in a shelf set, in a search grid, and in a customer's hands. A pack judged only as a flat image on a screen has not been judged.
- Test the production pack, not the sample. If only a sample exists, say so in the verdict and treat transit and unboxing results as provisional.
- Never invent a regulation, a test result, a percentage, or a certification. If a fact cannot be verified from a primary source or a test you ran, cut it from the pack and from the review.
- No claim ships on a verbal assurance, and no proof is approved by the person who wrote the copy reading it alone.
- Keep the style plain in all pack copy and reviews. Short lines a tired shopper can parse beat clever lines they have to decode.