WhitePaper
Tobii® Technology
Eye tracking
as a tool
in package
and shelf testing
Foreword
This guide is written to serve as inspiration and as an introduction for those who want to know more about
eye tracking and how this tool can be used in package and shelf testing. This guide will not explain a
methodology but I hope it will inspire you to develop your own methodologies as well as your own tests and
experiments. I would also like to thank the people who have helped us with this guide, especially Poja Shams,
who wrote the first draft of this guide. I would also like to thank Michel Wedel for providing feedback on this
document.
Tommy Strandvall ([email protected])
Tobii Technology
November 2008 / version 3
© Tobii Technology AB
Stockholm, Sweden
This document can be downloaded from our website: www.tobii.com
2
Eye tracking as a tool in package and shelf testing
Tobii Technology
Table of Contents 1. Introduction 4 1.1 What is eye tracking? 4
2. Eye tracking - a tool for shelf and package testing 5 2.1 What can eye tracking tell us about packages and products? 5 2.2 Designing a study 6 2.2.1 Uncontrolled tests 6 2.2.2. Controlled tests 6 2.3 Creating the stimuli 7 2.3.1 Creating the shelf images 7 2.3.2 Positioning and randomizing the products on the shelf 8 2.4 Conducting the study 8 2.4.1 How long should the products and the shelves be shown? 8 2.4.2 How many respondents should be included in the study? 9 2.4.3 Setting up and running the test 9
3. Visualizing and analyzing eye movement data 10 3.1 Visualizations in Tobii Studio 10 3.1.1 Gaze plots 10 3.1.2 Heat maps 10 3.1.3 Clusters 10 3.1.4 Areas of Interest (AOI) 11 3.2 Eye tracking metrics in Tobii Studio 11 3.3 Visualizing eye movement data in shelf tests by using Tobii Studio 12 Where did the eyes stop first? 12 What is the average viewing pattern on the shelf? 12 How many of the products on the shelf were noticed? 13 Do shoppers see the product on the shelf? 13 How many looked at the product first? 14 Is the product able to pull attention quickly? 14 How quickly did the product manage to pull attention? 14 How much attention did the product get compared to competing products? 15 How long was the product considered? 15 How many times did the shoppers look at the product? 16 How long does it take for the shoppers to find the product on the shelf? 16 How long does it take to find and select a product on the shelf? 16 How great is the product’s ability to retain attention for a period of time? 17 Does the placement of the product on the shelf have an impact on the order in which the
consumers look at the products? 17 How large is the attraction power for different elements in the design? 18
Bibliography 19 3
Eye tracking as a tool in package and shelf testing
Tobii Technology
1. Introduction
Most of our purchase decisions are made at the point of purchase, i.e. in the stores. According to the Point of
Purchase Advertising Institute, 74% of all mass merchandise purchase decisions are made in stores 1 .
Additionally, our purchase decisions are usually very quick, as shoppers only spend a few seconds in front of
the shelf before deciding to buy a specific product. Under these circumstances it is not surprising that the
ability for a product to attract shoppers’ visual attention – i.e. to be seen - has a strong influence on the
choices they make – products that are unseen are therefore often unsold.
Eye tracking equipment has been used in market research and for product testing over the past two decades.
These studies have proved that eye tracking is a reliable way to measure attention given to objects like
packages on a shelf. By analyzing shoppers’ eye movements it is possible to gain insight into their visual
attention span, which in turn is correlated with their purchase considerations.
Consumer behavior at the point of purchase is influenced by both memory-based and visual factors; eye
tracking data is of course best suited for gaining insight in the visual factors that have an impact on purchase
decisions. The product packaging also affects the perception of the brand and the product and thus one
could say that the packaging is the product. The challenge for a product today is to break through the clutter
on the shelf where many similar products and competing brands share the same space and where every
product is trying to attract the shoppers’ attention. For the product to be noticed, and ultimately purchased,
the packaging must win when compared to competing packages and products. In this context, eye-tracking
data is especially good in revealing how well the package can break through on the shelf.
The three most common questions package research is trying to find an answer to are:
1) Package visibility and findability on the shelf
2) Image and brand communication
3) Impact on the intention to buy and on sales
Package visibility is affected by several different factors such as shape, color, logo, etc. It is also possible to
test and compare alternative packages where the goal is to determine which design performs best on the
shelf. Traditional methods such as recall are generally a poor measure of shelf visibility as it is impacted by
brand familiarity and memory rather than actual shelf visibility. This is why only asking consumers which
brands they remember seeing on a shelf will most likely provide misleading answers.
Eye-tracking data is most valuable when it is combined with other conventional methods such as interviews
where consumers are asked about their purchase intentions, brand loyalty, price sensitivity and purchase
motivation, etc.
1.1 What is eye tracking?
An eye-tracking system measures how the eyes typically move over a digital image that is shown to a
respondent from the target group. The human eyes are constantly moving until they stop and focus on a
point. There are over ten different types of eye movements, of which the most important ones are saccades,
fixations, and smooth pursuit. When the eyes stop to focus it is called a fixation and the movements between
these fixations are called saccades. The length of the stops, when the eye fixates, varies from about 100 to
600 milliseconds and during this stop the brain starts to process the visual information received from the
eyes 2 . Saccades are extremely fast jumps from one fixation to another. The average length of a saccade is
about 20-40 ms. During this period the eyes do not send visual information to the brain. The human eye has
a visual field of about 200º but the highest number of light sensitive cells on the retina are located in the part
called fovea, which is the only point in our eyes where we are able to see a sharp and colorful image of the
world around us. This area is fairly small and covers only about 1 - 2 degrees of our vision (which is about the
size of a thumbnail at an arm length’s distance). It is only from these cells our brain can receive detailed visual
information. Our perception is slightly larger than the area of foveal vision. When, for example, reading a text,
1
2
4
POPAI (1997), Consumer Buying Habits Study
Rayner, K. (1998) Eye Movements in Reading and Information Processing: 20 Years of Research
Eye tracking as a tool in package and shelf testing
Tobii Technology
we can read about 12-15 letters to the right and 3-4 letters to the left, which would indicate there is a
perceptual span of around 18 characters centered asymmetrically around the fixation point 3 . Using our
peripheral vision outside the area of the foveal vision we can see an unclear image, but our peripheral vision
is generally very poor and is only good at picking up movements and contrasts.
It is possible to move our covert attention (the attention of our mind) around the entire visual field when our
eyes are at rest. Thus it is possible to move our attention around without eye movements. However, our foveal
vision is usually a valid measure for determining the target of our covert attention as our brain can process
very little information from complex stimuli from the area outside the fovea (such as packages on a shelf). 4 In
such cases it is more efficient for our brain to focus attention on the fovea rather than on our peripheral
vision, as the brain needs to process blurry visual information requiring more effort to interpret than the visual
information from the fovea. This is why it is possible to tell something about human behavior by just following
eye movements, especially the fixations, as we know that we can only see something clearly when we fixate
on an object or are very close to it. If a person does not fixate on, or is not very close to the package on a
shelf, he or she would not be able to see the product and read the text on it. The eye tracker records these
movements and the location of the foveal vision when the eyes fixate.
By analyzing our eye movements it is possible to tell something about our behavior. The length of a fixation is
usually an indication of information processing or cognitive activities as this is when the brain interprets the
visual information from the eyes. For example, in reading studies it has been proven that for common words
the fixation lengths are shorter than for less common words where the fixation lengths are longer. The
amount of fixations can indicate how easy it is to find, for example, a specific package on a shelf. The fewer
fixations used the more efficient the search process.
2. Eye tracking - a tool for shelf and package testing
2.1 What can eye tracking tell us about packages and products?
Pre-testing and post-testing methods have traditionally been used to test the selling properties of a package.
In pre-tests, focus groups, interviews, and questionnaires have been used to reveal the attitudes of shoppers.
In post-tests, the quantity sold after changing the design is measured and compared to data collected
previously. In general, eye tracking is used as a tool to answer questions about the saliency of a package in a
store environment. Saliency is an indicator of how well the package stands out compared to other packages
on the shelf. An assumption in eye tracking studies is that packages that are viewed more have a higher
saliency than products that are viewed less.
When analyzing eye-tracking data from a shelf test the intention is usually to diagnose problems rather than
finding specific designs that work well and could be used in other situations. In comparative studies new
designs are tested against the current package. The data from an eye-tracking test are often combined with
focus groups, questionnaires and in-depth interviews to get a better understanding of how the package and
the product are perceived by shoppers. Other data that could be included are, for example, unaided shelf
recall, which can reveal which products the shoppers remember having seen on the shelf, or brand
consideration, which can reveal which brands on the shelf the shoppers would consider buying.
Consideration or purchase intention usually increases with the number of fixations on the product on the
shelf. Noting a product is, however, not necessarily a direct indication of purchase intention as consumers
usually note many more products on a shelf than they actually consider buying. In a shelf study conducted by
The Wharton School of Business they found a positive and significant correlation between how much
5
products were looked at (and reexamined) and purchase intention . If the shopper looks at a product, looks
away and then looks back, the purchase intention increases. Evidence has also been put forward that purely
memory-based consideration is possible since unseen is not always unsold. There is, however, a strong
positive relationship between the number of fixations on a product and consideration, as products fixated
3
Rayner, K. (1998) Eye Movements in Reading and Information Processing: 20 Years of Research
Pieters,. R & Wedel, M. (2007) Informativeness of eye movements for visual marketing: six cornerstones.
5
Chandon, Hutchinson, Bradlow and Young (2007) Measuring the Value of Point-of-Purchase Marketing with Commercial EyeTracking Data
4
5
Eye tracking as a tool in package and shelf testing
Tobii Technology
more are more likely to be considered. Each fixation provides a new chance to consider the product and thus
there is a relationship between purchase intention and the number of fixations.
Products are of course considered even though they are not fixated on. This is because some packages and
products are so well known that shoppers can even use their peripheral vision to identify them on the shelf, or
they simply expect that a well known product should be on the shelf. Additionally, attitudes towards the
brands play an important role when considering buying a product. Simply raising the visibility for brands that
people dislike, or for brands that are liked by everybody, does not necessarily increase the intention to buy the
product.
2.2 Designing a study
The method used for a shelf test with eye tracking should be chosen based on which questions you want to
answer. To be able to answer some questions it is necessary to have an experimental design of the study to
be able to measure impact of, for example, design changes and placement. Eye tracking tests can generally
be designed in two ways, either as a controlled or an uncontrolled test. In an uncontrolled test, the purpose of
the test is unknown to the respondents. The respondent is only instructed to look at the shelf, or the image of
the shelf, without any specific instructions. The main purpose of an uncontrolled test is to measure the
general saliency of the products on the shelf. In a controlled test the respondent is given a set of instructions,
for example to choose a product or to look for certain products, product properties, or product attributes. The
main purpose of a controlled test is to observe the performance of the tasks, for example measuring how
long it took to find a certain package on the shelf.
Regardless of the design you choose, you should always include questionnaires or interviews to gain more
information on the behavior of the respondents. These questions are not covered by this guide.
2.2.1 Uncontrolled tests
The performance indicator in an uncontrolled test is the time the respondents spend looking at different
products. The viewing time, during which time the shelf is shown, can either be constant or variable
depending on how the test is constructed. If you have the same viewing time for every respondent, the
observation times for different products are directly comparable between the respondents. However, if you
give the respondent the possibility to control the viewing time (this is called a relative measurement), the
observation times are not directly comparable. However, it is possible to see how big the attraction power is
for the different products and the shelf. A tachistoscope method can also be employed in uncontrolled tests;
in this case the shelf is shown for a very short time, for example 0.5 seconds, after which the respondents are
asked to mention which products they remember seeing on the shelf. The test can then be repeated with a
longer viewing time of, for example, 3 seconds. An uncontrolled test should always have a predetermined
randomization of the tested product on the shelf. Respondents tend to focus on the center of the shelf during
an uncontrolled test and therefore it is important to randomize the tested products in different placements on
the shelf to minimize the effect of the position.
Example questions in uncontrolled tests
•
•
•
•
•
Do shoppers see the product on the shelf?
How long does it take for the shopper to see the product?
How much attention did the product get compared to competing products?
Do shoppers see the product before they see the competing products?
Does the placement of the product on the shelf have an impact on the duration or order in which the
shopper looks at the product?
2.2.2. Controlled tests
In controlled tests the indicators of performance are generated from instructions and tasks given to the
respondents. Examples of such tasks are; to look for a certain product (brand A), or a product category
(cookie), or product attributes (blue color and cylindrical shape), or a product property (everyday cookie).
Instructing respondents to imagine that they need to buy a product and then asking them which product they
6
Eye tracking as a tool in package and shelf testing
Tobii Technology
would consider buying will provide you with both eye tracking data and brand consideration data. The
performance results are based on how the respondents conducted these tasks, such as time spent
searching, choices they made, etc. Process-based results are the eye movements that occurred while
conducting the tasks, for example what they were looking at and for how long, which elements on the
package they looked at, etc.
An important aspect of process-based data is that the quantity of time spent looking at a certain product on
the shelf does not automatically mean that the shopper will buy the product. However, it will give you insight
into the decision-making process of the respondent and reveal if the product is considered as an option for
purchase.
The tasks in controlled tests can be anything from perceptual features such as; finding the most exclusive
product on the shelf or looking for light products on the soda shelf to conceptual features like finding cookies
on the shelf or finding Brand A cookies on the shelf. The perceptual features are connected to the
perceptions of the products and the brand. As an example, luxury, economical, and inexpensive are
perceptual features. Perceptual features are used to see if the packaging is communicating the message it is
supposed to communicate. Conceptual features are connected to the long-term memory and can be seen as
the mental picture of a product, brand or category. Brand A is a conceptual feature, which is connected to the
mental picture of Brand A in the mind of the respondent.
Example questions in controlled tests
•
•
•
•
•
How long does it take for the shopper to find the product on the shelf?
Can we change the time needed to find the product on the shelf by changing different properties
(color, shape, glossiness) of the packaging?
How much attention does a package get when the shopper is looking for a product with certain
properties, for example an exclusive product?
Which products in the same product category are seen as having a certain property, for example
being exclusive?
What attributes do other products with the same property have and how can these be used in a new
design?
Example
Suppose a respondent is asked to find Brand A cookies on a shelf and to click on the product with the
mouse when it is found. The performance is the time it takes for the respondent to find and click on the
Brand A product. The process consists of the eye movements during this time. Some interesting questions to
answer are: How long does it take for the average respondent to find Brand A? What products were
confused with Brand A and contributed to the delay of finding Brand A?
How do we use this information? If Brand A needs to differentiate the package design but still keep the
feeling of Brand A while raising the findability level, then the results can reveal design details that
communicate Brand A to the shopper. When the essence of the Brand A design is found we can manipulate
the package to enhance these communicative properties. The details that communicate Brand A may be any
attributes on the package such as the color, shape, glossiness, brand name and font. Eye tracking can be
used as a tool to help you find these details. In this case, the question we want to answer might be; what was
the first point on the package the respondents looked at when identifying it as Brand A?
2.3 Creating the stimuli
2.3.1 Creating the shelf images
It is important to use good quality images in a shelf test. Details that can grab attention and are not related to
the products should not be present. If it is important to also include price as a factor in the test, then price
tags can be included on the shelf. There are three main ways to create the picture of the shelf; using a virtual
planogram, photographing the products on a real shelf or photographing the products one by one in a light
box and then composing a shelf out of the images. If you are photographing a real shelf then randomizing
7
Eye tracking as a tool in package and shelf testing
Tobii Technology
products on the shelf can be time consuming and there might also be problems with the quality of the image.
Photographing the products one by one and using photo-editing software to create the virtual shelf might
result in better quality. Additionally, it is easy to make changes to the shelf using this method, for example
when changing the positions of the products. Another aspect of this method is that all the products will have
the same face angle. If the face angle is the same for all the products on the shelf, the products on the lower
shelf level will be seen from the same angle as the products in the middle and the top. The advantage is that
no bias will be present on any part of the shelf. The disadvantage is that the picture will be unrealistic for
respondents who notice this visual glitch. 3D modeling software can also be used to create more realistic
pictures of the shelf or the products.
2.3.2 Positioning and randomizing the products on the shelf
If the test is uncontrolled, it is important to randomize the products on the shelf since the attention of the
respondents tends to focus around the center of the picture. Randomizing the positions on the shelf will
remove this bias. The positions used on the shelf should follow the logic for how real products are placed on
shelves in stores. This logic is based on brands, products, and sometimes the product quality. Randomization
will give a better understanding of how the position of the products affects the results. By randomizing the
products you can observe if the product gets the same amount of attention independent of position on the
shelf. To be able to compare whether or not the position has an impact one would typically show the pictures
to two different groups of similar people; one group would see the product in position A and one group would
see the product in position B (see image below).
If the test is controlled, it is not recommended that you randomize the positions because the purpose is to
measure the attention based on instructions given to the respondents. In more complex studies where, for
example, you want to measure the impact of design changes, it is necessary to use an experimental design of
your study and to carefully randomize the positions of the packages on the shelf.
Group A
Group B
Compare
Compare
2.4 Conducting the study
2.4.1 How long should the products and the shelves be shown?
Shoppers typically make their purchasing decisions in front of the shelf within only a few seconds. A shopper
spends on average of 10 seconds on a product category before approaching a certain part of the shelf. 6 In a
recent eye tracking study 7 involving 309 shoppers where the respondents were asked to look at a picture of
a shelf of detergents and juices for as long as they wanted in order to make a purchase decision, the average
viewing time for the detergents turned out to be 17.99 seconds, and for juices 25.06 seconds. Other in-store
observation studies have revealed that shoppers spend between 12.2 and 13.2 seconds in front of a shelf
with detergents.
Deciding how long the shelf should be shown to the respondents depends on the type of test you are
conducting and the number of products on the shelf. If the test is uncontrolled you can show every shelf for a
fixed time. In controlled tests the viewing time should stop when the respondent has fulfilled the task. In
uncontrolled tests, the suggested viewing time for small shelves could be about 3 to 7 seconds, and for larger
shelves with more products, up to 15 or 20 seconds. It is also possible to let the respondents decide
6
Rowan, Claire (2000), Packaging by design, Food engineering International, February
Chandon, Hutchinson, Bradlow and Young (2007)Measuring the Value of Point-of-Purchase Marketing with Commercial Eye-Tracking
Data
7
8
Eye tracking as a tool in package and shelf testing
Tobii Technology
themselves how long they want to view the image of the shelf. By doing this it is possible measure how well
the packages manage to engage the respondents and hold their attention. In tachistoscope tests (T-scope),
the image of the package or the shelf is shown very quickly to the respondents. The viewing time can be
parts of a second up to a few seconds. Usually the viewing time is gradually increased (for example 0.5, 3
and 7 seconds) to measure how the recognition improves. In these T-scope studies the respondents are
asked directly after the viewing to recall what they have seen on the shelf.
2.4.2 How many respondents should be included in the study?
A general rule when deciding how many people to include in the study is that the test should always be
conducted among people who are in the target group for the brand or the product category. The sample size
to be used in a shelf-test with eye tracking depends on three main factors: 1) the sub-groups that will be
analyzed independently; 2) the degree of risk involved in the decisions being made based on the results; and
3) the amount of available resources, mainly time and money. In the Tobii Studio software you can use
demographic variables such as age, gender, and income level to filter the results. If you want to be able to
compare the results based on different background variables you need to include enough people from each
sub-group or cell in the study to get reliable results that can be generalized to the entire population. If you are
planning to use statistical tests on the results, for example, to find correlations, you need to adapt the number
of respondents to the type of statistical tests you are planning to conduct. In commercial shelf tests with eye
tracking, a sample size of about 100 per cell is usually used to get robust and reliable data. For example, if it
is important to compare men and women independently then it is advisable to have about 100 from each
group in the study. Tests can of course also be conducted with fewer participants, depending on the goals
and available resources.
2.4.3 Setting up and running the test
Using a Tobii x120 eye tracker
The Tobii x120 eye tracker is most suitable for shelf testing as it
can be used together with life-size shelf images projected on a
screen or shown on a large monitor. When used together with a
projector and a screen the respondent should stand or sit a couple
of meters away from the screen, depending on the size of the
screen. The eye tracker itself should be placed at a distance of
about 70 cm (27.5”) from the respondent’s eyes. The eye-tracker
should be attached to a robust floor stand and the height and tilt
should be adjusted so the eye tracker is able to track the
respondents’ eyes regardless of where on the screen the
respondent is looking. We recommend using the Tobii floor stand
where the height is easily adjusted. A good high definition projector
and a high quality screen should also be used for maximum quality.
When deciding the size of the screen and the distance from the
screen to the respondent, please calculate the correct distance and
size based on the maximum viewing angle of the eye tracker, which
is 35 degrees (see image below). For smaller shelves (not lifesize) a large monitor or television screen can be used to present
the images to the respondents. For more information about how to
use the Tobii x120 together with a projector and a screen please
refer to the x120 User Manual.
If the Tobii x120 eye tracker is used together with a scene camera,
a real shelf with real products could be used in the test. In this case
the setup would be similar to the setup above, but instead, a real
shelf with real products would be presented to the respondent. In
this setup a video camera needs to be placed close to the
respondent so it can record the entire shelf during the test.
Viewing time can be controlled by covering the shelf using a roll-up
screen. Additionally, the same roll-up screen can be used when
conducting the calibration. Using a setup with a real shelf is
9
Eye tracking as a tool in package and shelf testing
Tobii Technology
70 cm
27”
complicated and the quality of the results might also be affected, especially since the resolution of the
images will be much lower as they are taken directly from the video camera. For more information on how to
use the x120 in a scene camera setup please refer to the Tobii x120 User manual.
Using a Tobii T60 or T120 eye tracker
For testing small shelves or individual packages, a Tobii T60 or T120 eye tracker can be used. In this case
the stimulus is shown on the integrated 17” TFT screen with a resolution of maximum 1280 x 1024. When
using a Tobii T60/T120 the respondent can either sit or stand in front of the eye tracker. The distance from
the respondent’s eyes to the eye tracker should be around 70 cm (27.5”).
3. Visualizing and analyzing eye movement data
3.1 Visualizations in Tobii Studio
After collecting the eye movement data it is time to analyze the results. This is done in the Visualization view
in the Tobii Studio software package.
3.1.1 Gaze plots
The Gaze Plot tool in Tobii Studio is a useful tool when visualizing
scan paths and search behavior. Each fixation is illustrated with a
dot where the radius represents the length of the fixation. The
longer the fixation - the larger the dot. It is recommended that you
use the gaze plot for only one or a few participants since the plots
will be hard to interpret if more people are included. The scan path
image helpful when drawing conclusions on the search behavior of
the individual participants. By decreasing the time being shown in a
gaze plot it is possible to create illustrative images representing a
short time interval, such as the eye movements during the first 0.5
seconds when the shelf was shown.
3.1.2 Heat maps
The Heat Map plot is a powerful way to visualize the gaze behavior
of an entire group of recordings. The Heat Map plot consists of the
shelf picture as background image and a Heat Map mask
superimposed on top of this. It shows the aggregated results of
many respondents as hot and cold spots on the shelf picture or as
a gaze opacity picture where the fixations create holes in an
otherwise black surface. A heat map can be created based either
on the fixation count or fixation duration. It is possible to set the
scale for the heat map when creating a heat map based on fixation
count or absolute duration. You can easily see whether the
average respondent has seen a part of the shelf or not by looking at a heat map. By decreasing the time
being shown in a heat map it is possible to create illustrative images representing a short time interval, such
as the eye movements during the first 0.5 seconds when the shelf was shown.
3.1.3 Clusters
Clustering of fixation data can be seen as an automatic generation
of Areas of Interest. The cluster function reveals areas with a high
concentration of fixation points. The maximum distance between
fixations to include in a cluster can be set as a distance threshold in
the settings.
10
Eye tracking as a tool in package and shelf testing
Tobii Technology
3.1.4 Areas of Interest (AOI)
A very useful tool for quantifying gaze data on a higher level is to
use areas of interest (AOI). You can draw the areas of interest on
your shelf image, and after this, the gaze data from these areas are
available for further statistical analysis in Tobii Studio or other
software. The data for the different AOIs can easily be exported for
further analysis by using the built-in SPSS export function in Tobii
Studio. The AOI should be slightly larger than the exact graphical
object you want to analyze; this is because you also want to pick up
fixations that are not completely on target due to the inaccuracy of
human vision.
3.2 Eye tracking metrics in Tobii Studio
All the eye tracking metrics included in Tobii Studio are affected by the fixation
filters, which are used for defining fixations and cleaning up the gaze data. By
changing the settings for the fixation filters the results will vary since the idea
behind the fixation filters is to reduce the amount of data by combining individual
fixations into groups of fixations and thus reduce the amount of data. In turn, the
duration of the fixations will be longer.
There are two different ways of presenting the gaze data for the different Areas of
Interest (AOI) defined in Tobii Studio; either as results for the different individual participants
(recordings/participants versus AOIs) or as averages over all participants (Images versus AOIs). As a default,
the values presented for the individual participants are sums of the gaze data within the respective AOIs. The
values presented for all the participants (Images versus AOIs) are averages over all participants for the
different AOIs. By using the Cell values button in Studio, it is possible to get other values such as averages
for individuals.
The following metrics are currently included in Tobii Studio:
Time to first fixation
The time in seconds from when the stimulus was shown until the start of the first fixation within an AOI.
Fixation length
The length of the fixations in seconds within an AOI.
Fixation count
The number of fixations within an AOI.
Observation length
The total time in seconds for every time a person has looked within an AOI, starting with a fixation inside the
AOI and ending with a fixation outside the AOI.
Observation count
The number of visits and re-visits to an AOI.
Fixations before
The number of fixations before the participant fixated within an AOI for the first time.
Participant %
The percentage of participants that have fixated at least once within an AOI.
11
Eye tracking as a tool in package and shelf testing
Tobii Technology
Exporting data for further analysis
Using Tobii Studio you can compare groups based on background variables by filtering the results to only
show the results for a specific sub-group. For example, if you want to search for significant differences
between groups you need to export the gaze data from Tobii Studio and analyze it using other software.
There is an integrated SPSS/Mathlab/Excel export function in Tobii Studio that will export the data for
different AOIs to a text file. It is also possible to export raw gaze data for every participant separately by using
the Text Export function found in the Replay view.
3.3 Visualizing eye movement data in shelf tests by using Tobii
Studio
Where did the eyes stop first?
By looking at the fixations in a gaze plot you will see where the respondents fixated during the test. On the
timeline you can define the time interval to be displayed, for example the first few fixations on the shelf. If you
notice that the first fixation is interfered with by the previous stimuli shown, we recommend starting the gaze
plot analysis at the second fixation. It is possible to save the selected time on the timeline and then generate
a heat map or a gaze opacity heat map for the same time interval to illustrate what caught the respondents’
attention when they first saw the shelf.
A Gaze opacity heat map and a gaze plot showing the attention during the first second.
What is the average viewing pattern on the shelf?
If you want to reveal the respondents’ average viewing pattern you can use the “Time to First fixation” metric
for your AOIs. The simplest way to get a basic illustration of the viewing pattern is to analyze which AOI has
the shortest average Time to First fixation, second shortest, etc. In the Table view in Tobii Studio you can
organize the values in the Time to First Fixation column to get a list of the average viewing order of the AOIs.
For a more robust analysis of the viewing order you need to export the AOI data for the individual
respondents and do the analysis as illustrated below. In this case you will be able to calculate the percentage
of people that have seen the AOI in a certain order.
Time To First Fixation
Person A
Person B
Person C
Person D
1st
2nd
12
Product X
25%
75%
Product X
0.45
0.80
1.35
0.75
Product Y
1.10
0.45
0.65
0.55
Product Y
75%
25%
Eye tracking as a tool in package and shelf testing
Tobii Technology
Product X order
1
2
2
2
Product Y order
2
1
1
1
The average Time to First Fixation values for the Areas of Interest defined on the shelf displayed as a
graph and as a table in Tobii Studio.
How many of the products on the shelf were noticed?
By defining an AOI around all the products on the shelf it is possible to compare the attraction power for all
of them. Use the function “Participant %” in Tobii Studio to see how many of the respondents noted the
different products on the shelf i.e. fixated at least once within the defined AOI.
The graph shows the percentage of respondents that have fixated at least once within the defined
AOIs.
Do shoppers see the product on the shelf?
The fastest way to analyze whether or not the respondents have seen the product on the shelf is to make a
heat map including data for all respondents. If there is attention on the product you can compare it to other
products on the shelf by creating an AOI around the products. After this, you are able to use the metric
“Participant %” in Tobii Studio, which will reveal what percentage of the respondents looked at the product on
the shelf at least once.
A heat map reveals if there is attention on the products. By using Participant % you can compare how
much the different products on the shelf were seen.
13
Eye tracking as a tool in package and shelf testing
Tobii Technology
How many looked at the product first?
If you want to find out how quickly the products managed to attract attention you can calculate what
percentage of the respondents looked first at a product. This is done in Tobii Studio by selecting “Time to
First fixation” and then going to the Table view and comparing the individual results for all participants. Count
how many people have the shortest time for Product 1 and the shortest time for Product 2 and then generate
the percentage. For example, in this case 21 out of 25 respondents first looked at Product 1 resulting in
84%.
A Gaze plot showing only the first second can reveal what the respondents looked at first. By using Time
to First fixation and the table view you can reveal which products received more first looks.
Is the product able to pull attention quickly?
By looking at the fixations in a gaze plot you will see how the respondents are fixating and moving their eyes
over the shelf. For example, by changing the time interval being shown you can manipulate the gaze plot
image to show results for the first second or only the first few fixations. We recommend starting the gaze plot
analysis at the second fixation if the first fixation is affected by the previous image shown.
By using the metric “Time to First Fixation” in Tobii Studio you will be able to see how quickly the different
AOIs were seen in seconds. The lower the average Time to First Fixation for an AOI, the quicker it has been
detected.
A Gaze plot showing only the first second or the first few seconds can reveal what the respondents looked at first.
By looking at the graph for “Time to First fixation” you can reveal which product had the quickest average Time to
First Fixation.
How quickly did the product manage to pull attention?
By using the timeline in Tobii Studio you can choose to only view results for a certain time interval. Use the
“Participant %” in combination with the timeline to see how many participants looked at the product within
different time frames; for example, within 1 second, within 2 seconds and within 5 seconds, etc. This is an
indicator of the time it took for an element or a product to pull attention.
14
Eye tracking as a tool in package and shelf testing
Tobii Technology
1 second, Participant %
2 seconds, Participant %
5 seconds, Participant %
How much attention did the product get compared to competing
products?
By defining an AOI around the products you can compare the differences between the products. You can
compare the amount of attention they received by looking at metrics like “Observation count”, “Observation
length”, “Fixation length” or “Fixation count” in Tobii Studio.
For example, compare the different products on the shelf using AOIs and the metric Observation Length.
How long was the product considered?
The length of time the respondents looked at the product or a design element indicates the level of
involvement with the product. This is measured in Tobii Studio by using the metric “Observation length” or
“Fixation length”. Observation length will show you in seconds how much time the respondents on average
spent looking within an AOI.
By comparing Observation Length you will see, on average, how many seconds the respondents spent looking at
the different products.
15
Eye tracking as a tool in package and shelf testing
Tobii Technology
How many times did the shoppers look at the product?
By looking at “Observation count” in Tobii Studio you are able to see, on average, how many times the
respondents looked at the products within the AOIs. For example, by analyzing the values in the Table view in
Tobii Studio for every individual separately you can also manually generate a percentage describing how
many times the respondents re-examined a product at least once.
The metric Observation Count will reveal how
many times, on average, an AOI was visited and
re-visited. In the table view you are able to see the
values for the individual respondents.
How long does it take for the shoppers to find the product on the shelf?
The product’s findability is revealed by looking at the time it takes to find the product on the shelf. By
measuring the time it takes for the respondents to find or fixate on the product you get a findability score. In
Tobii Studio this could be done by measuring “Time to first fixation”. If there are different designs available for
the same product, you can use different groups of respondents to compare whether there are any
differences in findability between the different designs.
Time to First fixation will reveal the average of how long it took for the respondents to find the product on the shelf.
How long does it take to find and select a product on the shelf?
By having the participants click on the product they would like to purchase, or that they are asked to find on
the shelf, you can gain insight into how long it takes to find a specific product. The clicks on products are
visualized in both heat maps and gaze plots. When analyzing individual respondents in the Replay view in
Studio you will be able to see how long it took until a person clicked on the product in the Events window.
Every click or key press has a time stamp. You may also see the click as an event on the timeline in the
Replay view. Another way of getting the data is by using the Bee Swarm function in Tobii Studio. In this case
the viewing time for an image must have been set to end when clicking on the image. Select the image you
want to analyze, then right-click on the timeline and choose “Copy Activity Data to Clipboard”. The data will
show how long every participant has seen the image, i.e. how long it took until they clicked on the product.
16
Eye tracking as a tool in package and shelf testing
Tobii Technology
When you have all the time-to-click values for every respondent, calculate averages, maximum, minimum and
standard deviation for the findability (time-to-click). You may also combine the time-to-click values with “Time
to First fixation” to see whether there is a difference between these two values.
Clicks are visualized in heat maps in Tobii Studio. To calculate the time-to-click rates you can calculate the time
difference between the ImageStart event and the LeftMouseClick event.
How great is the product’s ability to retain attention for a period of
time?
By comparing the total viewing time of the picture and the observation length for the AOIs it is possible to
measure how much time, out of the total time, the respondents spent looking at the different products. If the
picture was shown for 5 seconds and the “Observation Length” for Product A was 1.08 and for Product B
1.21 seconds then Product A received 22 % of the total attention and Product B received 24 % of the total
attention. The attention percentage can also be calculated based on the total “Observation Length” instead of
the viewing time. This is done by including an AOI covering the entire image and using this as an indication of
the total attention on the picture.
By comparing Observation Length with the total viewing time or total Observation Length for the picture it is possible
to calculate how much attention the different products received during the test.
Does the placement of the product on the shelf have an impact on the
order in which the consumers look at the products?
By putting the products in different locations on the shelf you will be able to measure the impact of the
placement. Evaluate the different placements by using the AOI tool in Tobii Studio and comparing, for
example, “Observation Length”, “Fixation Count”, “Observation Count” and “Time to First Fixation”. When
comparing the impact of placement it is important to use different groups of respondents for the different
placements, otherwise the results might be affected by users seeing the same product twice, i.e. to have an
experimental design of the test.
17
Eye tracking as a tool in package and shelf testing
Tobii Technology
For example, compare the impact of different locations on the shelf by looking at Time to First Fixation.
How large is the attraction power for different elements in the design?
By defining the design element you are interested in as an AOI you are able to analyze its attraction power.
You can compare it to other products and elements with the metric “Participant %” which reveals how many
of the participants noticed the element.
By comparing the Participant % values for different AOIs you can find out which design elements were seen.
By using the Cluster tool in Tobii Studio it is possible to find
clusters of fixations on the shelf. For example, by using this
function you can reveal design elements that attract plenty of
fixations. Changing the Distance threshold under Settings will
have an impact on the size of the clusters. The clusters can be
exported as AOIs for further analysis.
18
Eye tracking as a tool in package and shelf testing
Tobii Technology
Bibliography
Chandon, P., Hutchinson, W., Bradlow, E., & Young, S. (2008). Measuring the Value of Point-of-Purchase
Marketing with Commercial Eye-Tracking Data. In Wedel, M. & Pieters, R. (Eds.), Visual Marketing: From
Attention to Action. New York: Lawrence Earlbaum
Oshima, V. (2004). Are you being seen? Being visible in the cluttered marketing space of Asia. Paper
presented at ESOMAR’s Asia Pacific Conference March 2004, Shanghai, China.
Pieters,. R & Wedel, M. (2008). Informativeness of eye movements for visual marketing: six cornerstones. In
Wedel, M. & Pieters, R. (Eds.), Visual Marketing: From Attention to Action. New York: Lawrence Earlbaum
Rayner, K. (1998). Eye Movements in Reading and Information Processing: 20 Years of Research.
Psychological Bulletin, 124, 372-422.
Rowan, Claire (2000). Packaging by design, Food engineering International, February.
Russo, E., & Leclerc, F. (1994). An Eye-Fixation Analysis of Choice Processes for Consumer Nondurables.
The Journal of Consumer Research, Vol. 21, No. 2. pp. 274-290
Stingo, R. & Gallucci, F. (2007). Biofeedback and eye-tracking – the emotional and cognitive experience in
store.
Van der Lans, R., Peters, R. & Wedel, M. (2008). Competitive Brand Salience. Marketing Science, 27: 922 931.
Wedel, M. and Pieters, R. (2008) Eye Tracking for Visual Marketing. Foundations and Trends in Marketing, Vol.
1: No 4, pp 231-320.
Wedel, M. & Pieters, R. (Eds.) (2008) Visual Marketing: From Attention to Action. New York: Lawrence
Earlbaum
19
Eye tracking as a tool in package and shelf testing
Tobii Technology