Four tests you can run without a research budget, what each one proves, and two claims about packaging you should not pay for.
Test it in a competitive block, not on its own. Shoppers give about 13 seconds to a brand choice in store, measured by the Ehrenberg-Bass Institute in 2014, and an Ipsos eye-tracking exercise reported that they saw about 28 of 193 products on one shelf. A pack admired in isolation has not been tested at all. Four tests below need no research budget.
The one that matters most: the block test
A block test is a two-stage shelf-visibility method: packs are shown in a competitive block, and the measure is whether a design is found and recognised in context rather than admired on its own. Kantar’s PackEvaluate version shows respondents a shelf of 11 brands including the test pack. Kantar’s own support documentation describes the two stages in one line each: “Then we ask respondents to type in brands they remember seeing (spontaneous recall)” and “Then we show the test pack to assess prompted recognition.” Kantar publishes no pass threshold, so any supplier quoting a required recall percentage is inventing one.
You can run a rough version yourself, in this order:
- Print your pack at actual size, on the material closest to the final one.
- Put it on a shelf beside the products it will actually sit next to, at the height it will occupy.
- Step back to shopping distance and ask someone who has not seen the pack to find your brand.
- Time how long the search takes, and write down what they looked at first.
- Turn the pack round and ask what they remember. That is the unaided half of Kantar’s method, done for free.
What you are testing is not whether the design is good. It is whether it is findable in a place that is working against it.
Four tests, and what each one proves
Run them in this order, because each one filters the next. A print proof is a physical sample produced on the process and stock the final run will use, which is why a screen PDF cannot stand in for it.
| Test | How to run it | What it proves | What it does not prove |
|---|---|---|---|
| Block test | Actual-size pack in a competitive shelf, ask a stranger to find it | Findability in context | Whether anyone will buy it |
| Arm’s length | Stand back 2 metres and read it; ISO 24509:2019 is the published method for minimum legible size by viewing distance | Hierarchy and legibility | Colour accuracy |
| Range test | All variants side by side | The coding layer separates them | Individual pack appeal |
| Print proof | A physical proof from the actual process | Colour, finish and type at final size | Anything about shelf behaviour |
Each row proves one thing. A pack that passes all four has cleared the failures that are expensive after launch; it still has not been proven to sell, and no pre-launch test does that.
What the eye-tracking research actually supports
Two findings are worth knowing because they change what you look for.
In a mobile eye-tracking study of unplanned purchases published in PLOS ONE in March 2021 (Hummel and colleagues, 19 participants, 88 purchases), the product a shopper chose averaged 7.3 fixations against 3.1 on its label, and 92 percent of purchases carried at least one fixation on the chosen product. That says attention tracks with selection; it does not say that adding decoration buys attention.
A 2026 study in the Journal of Eye Movement Research (Guo and colleagues, 160 participants) measured that a transparent window brought the first fixation onto the product in 0.36 seconds against 0.93 seconds for a printed image of the same product, and held it longer, 5.89 seconds against 3.95. That is a structural decision, made early, and it is not something artwork can compensate for later.
Two claims you should not pay for
Some numbers circulate in packaging pitches that do not survive a look at their source.
The first is that colour increases brand recognition by a fixed percentage. The claim is widely repeated and the sourcing behind it does not hold; it was removed from our own design knowledge base for exactly that reason.
The second is that the brain processes images many thousands of times faster than text. That figure has no traceable study behind it either, and it was removed on the same grounds.
If a proposal leans on either one, ask for the study. A supplier who cannot produce it is quoting folklore, and MarkaWorks keeps a record of which claims were removed and why, precisely so they do not end up in a brief.
Five seconds, and what that test is
You will see the five-second test recommended often. It is a method, not a measured finding: you show the pack briefly and ask what the person remembers. Useful for checking whether the primary message survives a glance, and it does not tell you anything about purchase.
Treat it the same way you treat the block test. It answers one narrow question well, and it becomes misleading the moment somebody presents it as evidence that a design will perform.
When to run them
MarkaWorks runs the block test and the range test before artwork is finalised, because both can still change the architecture; we do them with printed mock-ups on a real shelf rather than on screen, and you can do the same on your own packs. Run the arm’s length test at every round, since it costs a minute. Run the print proof last, and run it on the process you will actually use, not a close approximation.
The order matters because the early tests can send you back to the structure, and the late one cannot.
