Shelf test of packaging: methodology of conduct
What is specifically measured during the shelf test
If eye-tracking is available, the set of metrics is standard, and it is precisely this that provides an objective picture, not the designer's guess about whether the packaging 'catches the eye'.
- 1st Viewed — the number of seconds after the shelf appears in the field of vision when the respondent first fixes his gaze on the packaging.
- Viewed Time — total viewing time of the area of interest.
- Viewed Time % — the proportion of this time from the entire experiment, showing the relative weight of the packaging against the background of competitors.
- Revisitors — percentage of respondents who returned their gaze to the packaging more than once.
- Revisits — average number of such returns per person.
Separately from eye-tracking, the 'three layers of efficiency' technique is also useful: the visual layer (noticeability, attention-grabbing), the contextual layer (position and perception among competitors), and the conversion layer (influence on the intention to buy). In its full version, the evaluation can be conducted based on an extended set of parameters — this may be excessive for a small brand, but the logic of the three layers should be kept in mind at any budget level: noticeability without the intention to buy is a beautiful, yet useless packaging.
The real shelf, a mock-up, or a virtual model — what to choose
The difference between the three options lies not so much in price as in control over variables.
| Format | Pros | Cons | When to use |
|---|---|---|---|
| The real store shelf | Natural behavior, real purchase context | It is difficult to control lighting, neighboring products on the shelf, time of day | Final validation before launch |
| Mock-up shelf (hall-test) | Full control over variables, cheaper than shooting in retail | An artificial setting may give an inflated sense of noticeability | Comparison of 2-4 design concepts at the development stage |
| Virtual shelf | Cheap, fast, easy to scale the sample | Not suitable for products where packaging tactileness is important | Products sold primarily on marketplaces |
Mistake — choosing the format based on budget without thinking about the task. If the product mainly lives on Wildberries, testing it on a paper shelf mock-up in an agency meeting room — money spent in the wrong direction.
How many respondents are needed and where to look for them
The sample size depends on what exactly you are testing. For an express test, where you need to quickly filter out unsuccessful design variants, 30 respondents are sufficient — that's enough to see an obvious concept failure. For a full comparison of several packaging variants in a hall-test with shelf modeling, a sample of 150–200 people is needed — with a smaller group, the results will be noisy, and the decision to launch the design will be risky. If the test is built into an earlier stage — concept validation through focus groups and online surveys — the sample is increased to 300–500 people, because here the final design is not evaluated, but the idea itself.
Recruiting respondents must be strictly by the profile of the target audience: demographics, income level, actual purchasing behavior in the category. Testing dairy products on people who have never bought them will give beautiful figures and a useless result — they react to packaging differently than the real buyer at the shelf.
Can you do without eye-tracking and a projector
Yes, and this is not a compromised option 'for the poor' — with proper setup, such tests provide quite reliable data, provided the minimum sample size for each method is not violated.
- Find test — respondent is shown 4 slide with varying placement of the test product among competitors, and the time until detection is recorded. The metric is simple: the faster they find it, the higher the visibility. It works on a sample of 50 people for stable results — less gives a high variance in reaction time.
- Blur Meter test — the packaging is shown in a blurred form and the sharpness is gradually increased, recording the moment of recognition. It nicely demonstrates the strength of the color code and silhouette without any special equipment. 40–60 respondents are sufficient, because the metric (moment of recognition) is itself stable and does not require a large group to smooth out the noise.
- Tachystoscopy (T-scope) — a sample is displayed for 0,007–0,5 seconds, which simulates immediate peripheral perception on the shelf. Only a slide timer is needed, which is available in any online survey tool. For a stable result, 60–80 people are required: too short an exposure gives more individual variation than in Blur Meter.
All three methods are implemented in a regular online survey or even in a presentation with a stopwatch timer for in-person interviews. Eye-tracking enhances accuracy but is not a prerequisite for the methodology.
How to incorporate shelf testing into the packaging development process
A test conducted once at the end of the project 'just for the sake of it' has almost no impact on the design — the decision is already made, it's too late and expensive to change. The correct logic is four sequential stages.
- Researching the target audience and competitor shelf placement — to understand what the packaging will actually be next to.
- Focus groups and online testing on 300–500 people — validating several concepts at an early stage, before final rendering.
- Hall-test with shelf simulation for 150–200 people, eye-tracking is optional — comparing final variants against each other.
- Monitoring sales and recognition after launch — checking whether behavior in laboratory conditions matches the real market.
Such a sequence allows for design adjustments when it's still cheap, not after the packaging print run has been printed.
Misinterpretations that cause the test to lose meaning
- A sample of 20-30 people is perceived as the final verdict. This is a normal sample for rapid screening, but not a basis for launching the design without additional validation on a larger group.
- Only visibility is evaluated, without contextual and conversion layers. Packaging can win on Viewed Time and lose on intent to buy — if you only look at the first metric, it's easy to choose a 'loud' but not sales-driven design.
- There is no control group or current packaging for comparison. Without a benchmark, numbers mean nothing: 'packaging was noticed in 2 seconds' — is that good or bad if you don't know how long the previous version took to be noticed.
- Trust in declared preferences instead of behavioral data. Respondents in the survey often praise the "more stylish" option, but in reality, the first thing that catches the eye on the shelf is a completely different package — people poorly predict their own perception, and the gap between "liking" and "noticing first" can be enormous.
What to consider a good result in the end
The shelf test does not give a single correct number after which you can print the run with a peaceful conscience. It gives a comparison — the new packaging with the old, one concept with another, your brand with the nearest competitors on the same shelf. If the new design is noticed faster than the old one, stays longer to hold the gaze, and the declared intention to buy does not decline — this is a working result that can be defended before the client with numbers, not the designer's opinion. If at least one of the three layers — visual, contextual, or conversion — fails, the design needs to be refined before printing, not after it.
What is specifically measured during the shelf test
If eye-tracking is available, the set of metrics is standard, and it is precisely this that provides an objective picture, not the designer's guess about whether the packaging 'catches the eye'.
- 1st Viewed — the number of seconds after the shelf appears in the field of vision when the respondent first fixes his gaze on the packaging.
- Viewed Time — total viewing time of the area of interest.
- Viewed Time % — the proportion of this time from the entire experiment, showing the relative weight of the packaging against the background of competitors.
- Revisitors — percentage of respondents who returned their gaze to the packaging more than once.
- Revisits — average number of such returns per person.
Separately from eye-tracking, the 'three layers of efficiency' technique is also useful: the visual layer (noticeability, attention-grabbing), the contextual layer (position and perception among competitors), and the conversion layer (influence on the intention to buy). In its full version, the evaluation can be conducted based on an extended set of parameters — this may be excessive for a small brand, but the logic of the three layers should be kept in mind at any budget level: noticeability without the intention to buy is a beautiful, yet useless packaging.
The real shelf, a mock-up, or a virtual model — what to choose
The difference between the three options lies not so much in price as in control over variables.
| Format | Pros | Cons | When to use |
|---|---|---|---|
| The real store shelf | Natural behavior, real purchase context | It is difficult to control lighting, neighboring products on the shelf, time of day | Final validation before launch |
| Mock-up shelf (hall-test) | Full control over variables, cheaper than shooting in retail | An artificial setting may give an inflated sense of noticeability | Comparison of 2-4 design concepts at the development stage |
| Virtual shelf | Cheap, fast, easy to scale the sample | Not suitable for products where packaging tactileness is important | Products sold primarily on marketplaces |
Mistake — choosing the format based on budget without thinking about the task. If the product mainly lives on Wildberries, testing it on a paper shelf mock-up in an agency meeting room — money spent in the wrong direction.
How many respondents are needed and where to look for them
The sample size depends on what exactly you are testing. For an express test, where you need to quickly filter out unsuccessful design variants, 30 respondents are sufficient — that's enough to see an obvious concept failure. For a full comparison of several packaging variants in a hall-test with shelf modeling, a sample of 150–200 people is needed — with a smaller group, the results will be noisy, and the decision to launch the design will be risky. If the test is built into an earlier stage — concept validation through focus groups and online surveys — the sample is increased to 300–500 people, because here the final design is not evaluated, but the idea itself.
Recruiting respondents must be strictly by the profile of the target audience: demographics, income level, actual purchasing behavior in the category. Testing dairy products on people who have never bought them will give beautiful figures and a useless result — they react to packaging differently than the real buyer at the shelf.
Can you do without eye-tracking and a projector
Yes, and this is not a compromised option 'for the poor' — with proper setup, such tests provide quite reliable data, provided the minimum sample size for each method is not violated.
- Find test — respondent is shown 4 slide with varying placement of the test product among competitors, and the time until detection is recorded. The metric is simple: the faster they find it, the higher the visibility. It works on a sample of 50 people for stable results — less gives a high variance in reaction time.
- Blur Meter test — the packaging is shown in a blurred form and the sharpness is gradually increased, recording the moment of recognition. It nicely demonstrates the strength of the color code and silhouette without any special equipment. 40–60 respondents are sufficient, because the metric (moment of recognition) is itself stable and does not require a large group to smooth out the noise.
- Tachystoscopy (T-scope) — a sample is displayed for 0,007–0,5 seconds, which simulates immediate peripheral perception on the shelf. Only a slide timer is needed, which is available in any online survey tool. For a stable result, 60–80 people are required: too short an exposure gives more individual variation than in Blur Meter.
All three methods are implemented in a regular online survey or even in a presentation with a stopwatch timer for in-person interviews. Eye-tracking enhances accuracy but is not a prerequisite for the methodology.
How to incorporate shelf testing into the packaging development process
A test conducted once at the end of the project 'just for the sake of it' has almost no impact on the design — the decision is already made, it's too late and expensive to change. The correct logic is four sequential stages.
- Researching the target audience and competitor shelf placement — to understand what the packaging will actually be next to.
- Focus groups and online testing on 300–500 people — validating several concepts at an early stage, before final rendering.
- Hall-test with shelf simulation for 150–200 people, eye-tracking is optional — comparing final variants against each other.
- Monitoring sales and recognition after launch — checking whether behavior in laboratory conditions matches the real market.
Such a sequence allows for design adjustments when it's still cheap, not after the packaging print run has been printed.
Misinterpretations that cause the test to lose meaning
- A sample of 20-30 people is perceived as the final verdict. This is a normal sample for rapid screening, but not a basis for launching the design without additional validation on a larger group.
- Only visibility is evaluated, without contextual and conversion layers. Packaging can win on Viewed Time and lose on intent to buy — if you only look at the first metric, it's easy to choose a 'loud' but not sales-driven design.
- There is no control group or current packaging for comparison. Without a benchmark, numbers mean nothing: 'packaging was noticed in 2 seconds' — is that good or bad if you don't know how long the previous version took to be noticed.
- Trust in declared preferences instead of behavioral data. Respondents in the survey often praise the "more stylish" option, but in reality, the first thing that catches the eye on the shelf is a completely different package — people poorly predict their own perception, and the gap between "liking" and "noticing first" can be enormous.
What to consider a good result in the end
The shelf test does not give a single correct number after which you can print the run with a peaceful conscience. It gives a comparison — the new packaging with the old, one concept with another, your brand with the nearest competitors on the same shelf. If the new design is noticed faster than the old one, stays longer to hold the gaze, and the declared intention to buy does not decline — this is a working result that can be defended before the client with numbers, not the designer's opinion. If at least one of the three layers — visual, contextual, or conversion — fails, the design needs to be refined before printing, not after it.
Recent articles
MegaFon, BMW, VTB and more 240+ companies. See how it looks in the showcase reel:
Our branding agency is approached to order:
Branding agency services
