Constellations, patent pending

What 500 People See · Process

Testing Shelf Perception

Packaging commits tooling, print runs and shelf space all at once, and it gets reviewed the way a business card does. The buyer sees it for a second, from four feet, next to eleven competitors.

Quick answer

Packaging is the highest-consequence design decision most brands make, because it commits tooling, print runs and shelf space all at once, and it gets reviewed the same way a business card does. A group of people in a conference room look at three comps under good light at desk distance. The buyer sees it for a second, from four feet, next to eleven competitors, with no context and no explanation. Testing element by element before the comps exist puts real response into the decision while everything is still cheap to change.

What the conference room can’t simulate

The gap between how packaging gets evaluated and how it gets encountered is wider than for any other kind of design work.

Distance and time. Comps get held at reading distance and studied. Packs get scanned from several feet away in under a second. Detail that carries the concept in a review is invisible at the moment of purchase.

Isolation against context. A comp appears alone on a table or a slide. A pack appears inside a wall of packs, most of them using the category's conventions, several of them owned by the same competitor. Whatever a design is doing, it's doing it against that wall.

Explanation. Somebody presents the comp and says what it means. Nobody stands in the aisle to do that. The pack has to carry it alone, and the audience will read intent off it in the time it takes to walk past.

Figure 1 · The review and the aisle

How a pack is evaluated in a review against how it is encountered in the aisle Two panels. On the left, the review: a presenter with a speech bubble beside one comp, held close and explained. On the right, the aisle: a shelf of twelve packs, one of them the design, the other eleven competitors. Four rows underneath name the gap: distance, reading distance against four feet; time, minutes against under a second; context, alone on a table against a wall of eleven competitors; and explanation, somebody says what it means against nobody, the pack carries it alone. THE REVIEW one comp, reading distance, somebody explains it THE AISLE a wall of packs, four feet, under a second “what it means is…” DISTANCE held at reading distance scanned from four feet TIME studied for minutes under a second, walking past CONTEXT alone on a table or a slide inside a wall of eleven competitors EXPLANATION somebody says what it means nobody. The pack carries it alone Detail that carries the concept in the review is invisible at the moment of purchase.
Every condition of the review is the opposite of the condition of the purchase. The comp that wins the room is the one that rewards attention. The aisle gives it none.

That last one connects to a finding that shows up repeatedly in perception work: audiences read images functionally, deciding fast what a thing is and what it's for. In one study, images sharing an identical visual vocabulary landed on opposite sides depending on whether they appeared to stand for something or to sell something. Packaging is the one surface where that read happens under the most pressure and the least attention.

The problem with testing finished packs

The instinct on a packaging project is to build three or four full comps and put them in front of people, or into a shelf mock-up, and ask which one wins.

It's the wrong test for two reasons.

It's too late to be cheap. By the time complete packs exist, structure, hierarchy, illustration approach and palette are all locked together into a proposal. A negative result tells you the proposal failed and gives you nothing to build the next one from.

It gets a worse read. Handing someone a complete design asks them to do the designer's job. They evaluate a whole, and their response mixes a dozen decisions into a single yes or no with no way to separate them. Which is also why “which of these three do you prefer” produces the least useful data available for the money.

Figure 2 · Comps against elements

Three finished comps collapsing into one answer, against a board of twelve elements answered one at a time On the left, in red, three finished comps labelled A, B and C bracket down to a single pill reading which one, one yes or no with a dozen decisions inside it, and a note that a no tells you the proposal failed and nothing about why. On the right, in green, a board of twelve element tiles, line, form, texture, color, type register, illustration, photo treatment, material, finish, depiction, claim and maker, each marked converged on, rejected or no strong read, leading to a pill reading the direction, built afterwards, by the creative team out of what converged. THREE FINISHED COMPS the usual test ELEMENTS ON A BOARD before the comps exist A B C “WHICH ONE?” one yes or no, a dozen decisions inside it a no tells you the proposal failed and nothing about why line form texture color type register illustration photo treatment material finish depiction claim maker converged on rejected no strong read THE DIRECTION, BUILT AFTERWARDS by the creative team, out of what converged every decision answered on its own, while it is still cheap to change A complete pack asks the audience to do the designer’s job. A board of elements asks them to respond.
Three comps produce one answer with a dozen decisions tangled inside it. Twelve elements produce twelve answers, and the direction gets built from the ones that held.

Test elements. Line, form, texture, color, type, illustration register, photographic treatment, material and finish cues. This is audience perception mapping applied to a pack: an audience responds to images instead of describing them, and the pattern shows where the group agrees, where it splits, and why. A common shape is two boards of twelve images, sometimes three or four, crossed against each other so the sampling covers the range rather than one person's preferred corner of it. The direction gets built afterwards, by the creative team, out of what the audience converged on.

What to put on the boards

Some categories worth separating out on a packaging project specifically.

Category cues against differentiation cues. Every category has visual conventions that signal what the product is, and packaging that abandons them risks not reading as the product at all. Test which conventions are load-bearing and which are just habit. The robot study is a useful parallel here: the unsettling responses were about ambiguity of category rather than about form. A pack that doesn't read as its category has the same problem.

Material and finish. Matte against gloss, uncoated against coated, foil, emboss, transparency. These carry an enormous amount of the premium or value read and they're usually decided last, on cost, without ever being tested.

Product depiction. Whether and how the contents appear. The Coca-Cola study is directly relevant: with 467 participants, the imagery that produced craving was cold, wet, in-motion liquid, condensation on glass, a bottle at the lips. The brand's own icons and its sentimental material didn't do that job. On a food or beverage pack, how the product is shown decides more of the response than anything else on the surface, and it tends to get treated as a photography brief rather than as a decision.

Typographic register. Not the typeface. What the type is doing: shouting a claim, naming a maker, listing an ingredient, carrying a tone.

Color, tested against the category rather than in isolation. A palette that reads premium on a white board can read cheap next to the shelf it will live on.

Segments matter more on packaging than most work

Packaging usually addresses several audiences at once and they don't want the same thing.

The existing buyer, who finds the pack by recognition and will be annoyed by anything that slows that down. The new buyer, who has no recognition to draw on and is reading the pack cold. The retail buyer, who is evaluating it as an item in a category plan. Internally, the sales team and the leadership team, who each have a picture in their head.

Run those as segments and look at them both separately and together. Where everyone converges, that's core and it stays fixed. Where the segments diverge, that's a real decision about who the pack is for, and it's better made in front of the data than assumed away in a meeting; what to do when the segments disagree is its own question. If there's any chance a group sees it differently, collect the segment. Segments can be blended back together afterwards and cannot be recovered later.

What it protects

The reason to do this on packaging rather than on some other surface is the asymmetry of the mistake.

A website that reads wrong gets changed on a Tuesday. A pack that reads wrong is sitting in a warehouse in a quantity somebody negotiated a price on, on stock that was ordered against a forecast, with a listing date attached to it. The revision isn't a revision. It's a write-off plus a reprint plus a delay, and in a lot of categories it's also a conversation with a retailer about why the item on shelf isn't the one in the plan.

Figure 3 · The asymmetry of the mistake

The cost of a website that reads wrong, a pack that reads wrong, and the test that would have caught it Three horizontal bars. A website that reads wrong is a short blue bar, changed on a Tuesday. A pack that reads wrong is a long red bar in four segments, the write-off, the reprint, the delay and the retailer conversation, with a note that it sits in a warehouse in a quantity somebody negotiated a price on, with a listing date attached. A perception test at the element stage is a short green bar, a fraction of a single print run, back inside 48 hours. A website that reads wrong changed on a Tuesday A pack that reads wrong the write-off the reprint the delay the retailer conversation in a warehouse, in a quantity somebody negotiated a price on, with a listing date attached A perception test at the element stage a fraction of a single print run, back inside 48 hours The revision isn’t a revision. The check that would have caught it is the smallest bar on the chart.
The website mistake and the packaging mistake are the same error at very different prices. The test that separates them is the smallest bar on the chart.

Against that, a perception test at the element stage costs a fraction of a single print run and returns inside 48 hours. It also gives whoever has to defend the design in front of the leadership team something better than their own conviction, which is usually the thing that decides whether the right pack survives the last review.

Frequently asked questions

How do you test packaging design before production?

Test design elements rather than finished packs. Line, form, texture, color, type, material and finish cues, and how the product itself is depicted. Put a range of those to real buyers, then build the design on what they converged on.

Why not just show shoppers three finished pack designs?

Because it's too late for the result to be cheap and the read is worse. A complete pack collapses a dozen separate decisions into one yes or no, and it asks the audience to do the designer's job.

What matters most on a food or beverage pack?

How the product is shown decides more of the response than anything else on the surface. In the Coca-Cola study, cold, wet, in-motion liquid drove craving while the brand's own icons and sentimental material did not.

Which audiences should a packaging test include?

Existing buyers, new buyers and, where relevant, the retail buyer, run as separate segments. They read the shelf differently, and their disagreement is a decision worth seeing rather than averaging out.