Mørk et al.

BMC Public Health (2017) 17:239

DOI 10.1186/s12889-017-4149-3

RESEARCH ARTICLE Open Access

An analysis of the effects of a campaign

supporting use of a health symbol on
food sales and shopping behaviour of
consumers
Trine Mørk, Klaus G. Grunert* , Morten Fenger, Hans Jørn Juhl and George Tsalis

Abstract
Background: Since 2009, the green Keyhole symbol has been a joint Nordic initiative for signalling healthfulness of
specific food products. In 2014, the Danish Ministry of Food, Agriculture and Fisheries carried out a campaign aimed
mainly at men over 35 with a low level of education, encouraging them to use the Keyhole in their shopping process.
The objective of the study is to evaluate the campaign by measuring its effect on consumer behaviour in the store.
Methods: The impact of the Keyhole campaign was measured in selected retail stores. Sales data were analysed to
ascertain whether sales of Keyhole labelled products changed during and after the campaign. Observations and
interviews were conducted in the same stores.
Results: The campaign had a positive effect on sales of Keyhole-labelled products in two out of three retail chains
investigated. In these two retail chains, sales of Keyhole labelled products rose by about 20%. In the third chain, there
was a slight decrease of sales of Keyhole labelled products. The effect differed considerably between product
categories. Analysis of the interview data indicated that by the end of the campaign, shoppers with a short education
had a higher likelihood of mentioning health as a purchase motive, and there was a higher general tendency to look
for nutrition information.
Conclusions: Results suggest that the campaign did have effects on shopper behaviour and that it is possible to address
shoppers with a short education by a tailored campaign. However, long-term effect of the campaign was not ascertained.
Keywords: Food choice, Health symbol, Health campaign, Transaction data, Store interview

Background nature of many food choices [7] suggests that simpli-

In order to encourage healthier food choices, a variety fied and/or symbolic nutrition information may be a
of nutrition labelling schemes, both compulsory and promising way to encourage healthier food choices.
voluntary, are employed to signal healthier alternatives This includes health symbols, i.e., symbols on the
when making choices. Different types of nutrition label- package or the shelf front that designate food products
ling schemes have been investigated for their effect on as a healthier choice without invoking numerical infor-
consumers (for reviews see [1–4]) with mixed results, mation on calories or nutrients. Examples of health
suggesting that while many consumers understand such symbols are the Heart symbol in the U.S and Finland,
information and can make adequate use of it when the Keyhole in Norway, Sweden, Iceland and Denmark,
prompted to do so [5], lack of motivation to use the the Choices logo in Holland, Belgium, Poland, Czech
information and lack of attention to it in real-life store Republic, and Slovakia and The Heart Foundation Tick
environments are major bottlenecks regarding effects from New Zealand.
on actual food choice [6]. The habitual and heuristic Health symbols are often not self-explanatory, and
organizations sponsoring health symbols, both state-
* Correspondence: [email protected] sponsored and otherwise, employ a range of measures
MAPP Centre, Aarhus University, Bartholins Alle 14, 8000 Aarhus C, Denmark

© The Author(s). 2017 Open Access This article is distributed under the terms of the Creative Commons Attribution 4.0
International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and
reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to
the Creative Commons license, and indicate if changes were made. The Creative Commons Public Domain Dedication waiver
(http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated.
Mørk et al. BMC Public Health (2017) 17:239 Page 2 of 11

to promote consumer awareness, understanding, and theoretical framework of the study before explaining the
use. This can take many forms, including mass media methodology. Following this, we will present and discuss
advertising, TV commercials, posters in stores, sweep- the results. We will conclude the article by outlining the
stakes, and web-based information. Whether such mea- significance of the results for promoting the use of
sures indeed have an effect on consumers’ use of health health symbols.
symbols is an underresearched topic. The mixed results
of studies investigating the use and effects of health Theoretical framework
symbols (e.g., [8–10]) may be, in part, related to differ- In this study, we view a consumer’s choice of a food
ences in the way and extent to which these symbols product as the result of a process that is somewhere
were supported by measures to promote awareness, un- between a deliberate, consciously formed decision and a
derstanding and use of such symbols. Research on the habitual purchase. Habitually driven decisions are repeti-
effects of such support measures could therefore be a tions of earlier decisions in the same environment [16]
valuable contribution to our knowledge on the effects and may ultimately mean that a product is purchased
(or lack of same) of health symbols aimed at promoting purely based on product recognition. The more a choice
healthier choices. resembles a deliberate decision, the more information
In the present paper, we analyse the effects of a multi- processing will be involved, in-store and/or by previous
media campaign that the Danish Ministry of Food, exposure to product-related information. However, in
Agriculture and Fisheries carried out in 2014 to promote most cases, even deliberate decisions will be simplified
the Nordic Keyhole health symbol. The Keyhole is a decisions that are characterized by the use of heuristics -
voluntary health symbol, the purpose of which is to high- simple decision rules - and thus only limited processing
light a “better choice” of food products within a product of information will be involved, especially in the shop,
category. The Keyhole is awarded if the product meets the where decisions are usually made quickly [17, 18].
criteria established by the government in a nutrient profile Heuristics involve the use of a few key criteria such as
model. Each year, The Ministry of Food, Agriculture and brand and price, but possibly also criteria related to
Fisheries in Denmark runs a campaign, usually for a health and convenience. Heuristics can be modified
period of 4 weeks, to raise awareness about the Keyhole depending on the decision situation, including the acti-
and to promote the use of the label among consumers. vation of purchase motives at the time of decision [19].
Danish consumers are aware of the Keyhole symbol, but A campaign that aims to promote the use of the Keyhole
have limited knowledge about what the Keyhole label can influence consumer choice in several ways. First, con-
stands for [11]. The 2014 campaign is especially inter- sumers, especially those with a high health consciousness,
esting because it, while addressing the general audience, may already have the Keyhole as part of their repertoire of
also had a more narrowly defined target group: Men heuristics, and a campaign has the potential to increase
older than 35 with a low level of education. It is an the frequency of its use by priming it and thus making it
often-voiced concern that the incidence of unhealthy more accessible in the purchase situation. Second, con-
dietary patterns is related to socio-economic class [12, 13]. sumers who do not usually use health-related heuristics or
For example, in Denmark – a traditionally egalitarian use other health-related heuristics can be encouraged to
country – the highest share of obese is among citizens use the Keyhole in their attempts to find healthy alterna-
with low socio-economic status [14]. In a recent national tives. Finally, a campaign could increase the presence of
study researchers found that 18.6% of Danish men have health motivation in the purchasing process, which may
an unhealthy dietary pattern (based on a diet score based affect whether or not the Keyhole is used as information
system calculated on the basis of self-reported consump- in this process.
tion, where respondents were categorized as having an un- The possibilities of influencing consumers in the store
healthy dietary pattern if they report a low intake of fruit, are limited, because the amount of information is over-
vegetables and fish and a high consumption of saturated whelming and consumers spend only little time in making
fatty acids). For women the percentage is 9.5%. When decisions. Possible promotional effects are likely to be
education is taken into account, the study finds that 4.6% based on the combined effects of multiple information
of those who completed a higher education have an un- channels. Information channels that work outside the
healthy dietary pattern compared to 24.8% of those who shops can both lead to learning new information (e.g.
did not [15]. about the meaning of the Keyhole) and the prioritization
In this paper, we will provide evidence on some effects of health motives, which in turn can increase the likeli-
of the Keyhole campaign 2014. The knowledge obtained hood that health-related information is used both inside
from the study has implications both for the design of and outside the store.
future campaigns and for the potential of health symbols Research indicates that health symbols, such as the
to affect consumer choice. We will start by outlining the Keyhole, can increase, decrease, or have no effect on the
Mørk et al. BMC Public Health (2017) 17:239 Page 3 of 11

sales of the products that bear them [8–10, 20]. How- over 35 years of age with limited education – had a high
ever, studies on the effects of health symbols have not probability of shopping there. Likelihood of the focal
usually taken into account to which extent campaigns consumer group shopping there was assessed based on
have been used to introduce and/or promote the symbol. store location and store managers’ indication that the
This may be one reason for the diversity of results re- target group was shopping there. For practical reasons,
ported in previous studies. all six stores also had to be within a reasonable geo-
graphical distance from each other. Chains A and C
Methods were typical full-assortment supermarkets, whereas
Context chain B was a discount chain with a smaller assortment.
In 2014, the Danish Ministry of Food, Agriculture and The chains differed in the share of Keyhole-labelled
Fisheries carried out a 3-week campaign promoting products in the assortment. Use of the campaign mater-
the use of the Keyhole health symbol. The campaign ial was monitored during the three campaign weeks in
was especially directed at men over 35 years of age each store on a daily basis, using a checklist. No system-
with 9–10 years of school, with or without vocational atic differences in the use of the campaign material were
training, but without further education. The Ministry detected between the different stores.
formed an alliance with three major retail chains that
received promotional material to decorate the stores, Analysis of sales data
materials for competitions with prizes for both em- The sales data analyses were limited to the following
ployees and customers, and several recipes that were product categories: Fresh fruit and vegetables, frozen
developed together with a well-known TV chef specif- fruit and vegetables, frozen ready meals, fresh fish,
ically to appeal to the special target group. These were frozen fish, fresh meat, frozen meat, flavoured fermented
available in the participating stores and on the spe- milk products, breakfast cereals, and pre-packed bread.
cially designed Keyhole website. In-store promotional The selection of these categories was based on the
activities consisted of ceiling signs, tattoos with the assumption that they were likely to contain products
keyhole on, posters featuring the TV chef and a large labelled with the Keyhole logo, while also having some
knife, flyers with reference to what the keyhole is variation in the share of products that would be eligible
based on and shelf labels also encouraging participa- for the Keyhole.
tion in a sweepstake (see Table 1). Other activities in- For each of the above categories, sales data were
store were promotional videos in the fruit and vegeta- obtained on a daily basis for all days in the following
bles section or at the counter with fresh fish, showing weeks in 2014: 14, 15, 17 (week 16 omitted due to Easter
the TV chef creating one of the special designed meals holidays), weeks 18, 19, 20 (Keyhole campaign), and
for the campaign. Outside the store environment, weeks 21, 22, 23. Data obtained were total turnover in
radio spots, TV spots, and special events ran through- terms of both volume and value per category, total
out the campaign period. turnover in terms of both volume and value for Keyhole
labelled products in category, number of transactions in
Overview of data collection the product category, number of total transactions for
The impact of the Keyhole campaign was measured in Keyhole labelled products and information about aver-
selected retail stores. The study is in two parts. The age price of each product, all on a daily basis, resulting
first part consists of the analysis of sales data in se- in 9 × 7 = 63 data points..
lected stores, analysing whether sales of Keyhole la-
belled products changed during and after the campaign. In-store observation and interview
These data were provided by the participating retail The second part consisted of observations and subse-
chains. The second part consists of observations and quent interviews with customers in the selected stores.
interviews in selected stores. While the analysis of sales This part of the study was carried out in four out of the
data does not allow differentiating the effects by differ- six stores used in part 1 of the study, those from retail
ent target groups, the observation/interview study al- chains A and B. It was not possible to carry out inter-
lows us to distinguish effects on different groups of views in retail chain C due to resource constraints and
shoppers, including the target group of the campaign. delays in getting the necessary approval.
Shoppers were observed and subsequently approached
Selection of stores for a short interview at the shelf with breakfast cereals and
Six retail shops belonging to three different retail chains at the counter with ready meals. Our decision to concen-
were selected based on nominations from the participat- trate on breakfast cereals and ready meals was based on
ing retail chains. A criterion for the selection of stores the assumption that the breakfast cereals category covers
was that the campaign's focal consumer group - men a wide variety of products, both more and less healthful
Mørk et al. BMC Public Health (2017) 17:239 Page 4 of 11

Table 1 Examples of campaign material

Campaign material Design
Ceiling signs

Tattoos
Mørk et al. BMC Public Health (2017) 17:239 Page 5 of 11

Table 1 Examples of campaign material (Continued)

Posters

Commercial flyers
Mørk et al. BMC Public Health (2017) 17:239 Page 6 of 11

Table 1 Examples of campaign material (Continued)

Shelf labels

Reproduced with permission from the Danish Food and Veterinary Administration

[21, 22]. Furthermore, the special target group was question), if they had looked for nutritional information
assumed to be frequent consumers of ready meals. This (yes/no/can’t remember) and if yes, if they had looked
assumption is corroborated by studies linking men to be for any labels on the product (open question, but to be
more frequent consumers of ready meals, in part because confirmed by respondent by showing a picture of the
of less developed cooking skills and a more positive atti- label). If the customer indicated that s/he had looked for
tude towards the product category [23, 24]. In both cat- the Keyhole, s/he asked to show where they found the
egories, some products carried the Keyhole while others symbol, if they knew what the Keyhole means (open
did not. Chain A had fewer Keyhole labelled products on question), how often they look for the Keyhole when
its shelves, 459, than chain B, which had 961. shopping (never/rarely/often/always), if they have no-
During the observations, we noted whether the cus- ticed any special Keyhole activities in the store or on the
tomers looked at the product and, if so, whether they flyers (open question) and how much time they had to
looked at the front or other areas of the product. We shop for groceries that day (little/appropriate/plenty).
also noted whether they observed the campaign-related Answers to the open questions were coded afterwards.
material in the store and how long it took them to select In total, 1411 respondents were recruited in the four re-
the product. If a selection took place, we asked the shop- tail stores. All respondents had to be 18 years of age or
per to participate in an interview. Observational data older. Half of the respondents were men over 35 years of
from non-consenting shoppers were deleted. The inter- age with between 9 and 10 years of school, with or with-
view employed a questionnaire gathering data about age, out vocational training, but without further education.
employment, educational background, which item they Half of the respondents were recruited at the cereal shelf,
had just selected, if they had purchased that item before and the other half at the counter with ready meals. Half of
(yes/no/can’t remember), why they chose this item (open the respondents were recruited and interviewed before
Mørk et al. BMC Public Health (2017) 17:239 Page 7 of 11

the Keyhole campaign in week 17 and the other half to- chains, the price differential for Keyhole labelled products
wards the end of the Keyhole campaign in week 20. affects the odds for buying Keyhole labelled products. For
retail chains A and C, the odds for buying Keyhole la-
Results belled products rises by about 20% during the campaign
Analysis of sales data period, and this effect is sustained in the 3 weeks after the
Figure 1 shows the mean share of transactions with campaign. For the discount chain B, where the share of
Keyhole labelled products across all product categories transactions with Keyhole labelled products was highest
in the dataset for the three retail chains before, during before the campaign, the odds for buying Keyhole prod-
and after the campaign. The data show that the share of ucts decreased by about 10% during the campaign, and
transactions with Keyhole labelled products differs again a negative effect remained in the 3 weeks after the
considerably between the three retail chains, with campaign. All effects are statistically significant. For
discount chain B having the highest share. Share of additional illustration, Table 3 shows total number of
transactions with Keyhole labelled products decreased transactions and total value for Keyhole-labelled and not
slightly during the campaign for chain B, increased dur- Keyhole-labelled products for the three retail chains pre-,
ing the campaign for chains A and C, and levelled off during and post-campaign. Again, the results indicate con-
after the campaign for chain A, while it continued to siderable differences between the three retail chains. In
raise for chain C. These figures, however, cover strong the discount chain B, most of the transactions and the big-
variations between product categories. Some product ger share of the total value relates to products with the
categories, most notably fresh and frozen fish, but also keyhole. This may be partly due to the smaller assortment
cereals, show a strong increase, whereas other product of the chain, which includes a larger share of staple and a
categories remain almost constant or even decrease. smaller share of specialty products.
Because of the different context in the three retail In order to provide additional support to the conclu-
chains, statistical analysis was done separately for three sion that the differences between the periods are related
chains. For each, chain, a multi-level logistic regression to the campaign and not just to seasonal fluctuation, we
was estimated in SAS with the logit of the share of keyhole obtained, for retail chain A, the same data for the same
labelled products per day and per product category as the periods the year before the campaign and estimated the
dependent variable. Time period (during/post campaign same model. The results indicated no significant effect
vs. pre campaign) and the relative price differential of Key- for the same periods.
hole labelled products compared to the remaining prod-
ucts in the category were estimated as fixed effect Analysis of store observations and interviews
parameters. Product category was specified as a random Data from a total of 1411 respondents were gathered:
factor. The results are shown in Table 2. For all three retail 602 respondents within the focal consumer segment of

Fig. 1 Share of transactions with Keyhole labelled products by retail chain

Mørk et al. BMC Public Health (2017) 17:239 Page 8 of 11

Table 2 Share of Keyhole labelled products as explained by campaign period and price
Predictor Coefficient p Odds ratio Coefficient p Odds ratio Coefficient p Odds ratio
Retail chain A Retail chain B Retail chain C
Fixed effects
Intercept −1.091 .03 1.230 .04 .594 .02
Relative price −.955 .00 −1.129 .00 −1.216 .00
During campaign .186 .00 1.204 −.114 .00 0.892 .175 .00 1.191
Post campaign .162 .00 1.176 −.219 .00 0.803 .283 .00 1.327
Random effects
Bread Intercept −2.457 .00 .513 .29 −.012 .78
Cereals Intercept 1.700 .00 .004 .99 .544 .20
Fresh fish Intercept 1.900 .00 1.428 .00 1.831 .00
Frozen fish Intercept 1.302 .00 .513 .30 .029 .94
Fresh meat Intercept −2.351 .00 −.863 .08 −1.964 .00
Frozen meat Intercept −.718 .10
Fresh greens Intercept .903 .04 1.670 .00 −.439 .30
Frozen greens Intercept .582 .18 1.183 .01 2.011 .00
Frozen ready meals Intercept −1.268 .00 −2.271 .00 −1.879 .00
Chilled ready meals Intercept −1.216 .00
Sour milk products Intercept −.205 .64 −2.177 .00 .701 .10

men over 35 years with low education, 528 respondents category and retail chain, and consumer segment. In order
with an advanced educational background, and 281 to emphasize possible differences between the three con-
other respondents (e.g., not male or with age <35) with sumer segments (men > 35 with a short education, the
short/without education. Demographics characteristics special target group of the campaign; other shoppers with
of the sample can be seen in Table 4. a short education; shoppers with a long education), three-
ANOVAs and logistic regressions were estimated to as- class models were estimated in LatentGold. When the
certain the effect of the campaign as opposed to the pre- Wald statistic indicated that there were no significant dif-
campaign period for the following dependent variables: 1) ferences between the three segments (p > .1), coefficients
the time it took consumers to decide on their purchases, for the three segments were constrained to be equal.
2) the probability that the consumers looked at the front There was no campaign effect regarding the time used to
of the product’s package (the location of the Keyhole), 3) make a purchase decision (mean time was 21.8 s, SD =
whether health was a motive mentioned by consumers as 23.1). Likewise, there was no significant campaign effect
a reason for their choice in the corresponding open-ended on the likelihood to look at the front of the package, al-
question in the interview 4) the probability that respon- though the likelihood differed by retail chain.
dents looked for nutritional information, and 5) the prob- The likelihood of mentioning ‘health’ as one of the pur-
ability that respondents looked for the Keyhole logo. chase motives for the purchase just made was affected by
Descriptive statistics for these measures can be seen in the campaign (see Table 6). Likelihood of mentioning
Table 5. In all cases, the analysis controlled for product health as a purchase motive was almost double in the

Table 3 Average daily number of transactions and volume of transactions w/o Keyhole before, during and after campaign (value in
in Danish kroner, numbers aggregated over selected product categories)
Mørk et al. BMC Public Health (2017) 17:239 Page 9 of 11

Table 4 Demographic characteristics by shopper segment The likelihood that one looked for nutrition infor-
Segment Segment 2: Short Segment 3: Men > 35 mation on the package before making the purchase
1: Longer education, but not years with short eduction was likewise different pre/during campaign. At the
education men > 35 years (special target group)
end of the campaign, this likelihood was 60% higher
Age than before the campaign. This effect did not differ
18–35 29.0% 63.7% 0% between segments. The likelihood was also higher in
36–44 19.9% 5.7% 47.7% retail chain B.
45–54 22.0% 6.8% 18.3% For those shoppers who answered that they did look for
55–64 18.2% 9.3% 16.3%
nutrition information, it was additionally asked whether
they had looked for the Keyhole. This probability was not
65 + 11.0% 14.6% 17.8%
affected pre/during campaign.
Gender
Male 44.3% 32.4% 100.0% Discussion
Female 55.7% 67.6% 0% The choices made in grocery stores every day are import-
Highest level of education ant for the health of the general population. The introduc-
Primary school 0% 11.4% 14.3%
tion of health symbols has the aim to further healthy
choices. However, research on the effects of such symbols
0% 46.6% 15.3%
Secondary school on food choice has had mixed results. One of the reasons
for this may be that health symbols need to be supported
Vocational 0% 42.0% 70.4%
by campaigns promoting awareness, understanding and
100.0% 0% 0%
Higher education
use of the symbol, and the effects of such campaigns have
not been subject to scientific scrutiny. In this study, we
N 528 281 602
analysed the effects of a campaign aimed at promoting the
use of the Keyhole symbol in Denmark, a campaign that
campaign week as opposed to before the campaign for was specifically aimed at the target group men > 35 with a
those respondents who had short education but were not short education. In order to detect possible campaign ef-
part of the special target group. It also increased for the fects, we collected sales data from six retail stores from
special target group, while it slightly decreased for respon- three different retail chains and analysed the share of sales
dents with a longer education. The likelihood was smaller of products with a Keyhole before, during and after the
for ready meals as compared to cereals, and in the dis- campaign. Our results suggest that the Keyhole campaign
count chain B as compared to chain A. These effects did had an impact on the share of sales of Keyhole-labelled
not differ between the three segments. products. The effect during and after the campaign varied
across the retail chains: The retail chain with the lowest
share of Keyhole-labelled products showed an increase in
Table 5 Measures for in-store behaviour the sales of those products during the campaign which
Before the Keyhole During the Keyhole continued after the completion of the campaign, whereas
campaign campaign
the retail chain with the highest share of Keyhole-labelled
Time until purchase Mean 21.73 21.92
(sec)
products actually experienced a drop of sales of Keyhole-
labelled products during the same period.
Looked at front no 15.7% 13.7%
of pack We can only speculate about the reasons for the de-
cline of sales of Keyhole labelled product in chain B. As
yes 84.3% 86.3%
noted, this was the chain with the highest share of
Health as a purchase no 81.9% 78.5%
motive
transactions with Keyhole labelled products in the pre-
campaign period. At the same time, it is (in contrast to
yes 18.1% 21.5%
chains A and B) a discount chain and may attract fewer
Looked at nutrition no 73.7% 62.9% health-conscious shoppers, which is supported by the
info
finding that the likelihood of shoppers there having had
yes 26.3% 37.1%
health as a shopping motive was considerably lower com-
Look for Keyhole no 43.3% 31.8% pared to chain A. The campaign may, for some shoppers
(only shoppers who
looked for nutrition in chain B, have had the effect of alerting them to the
info) presence of the Keyhole on many products in these shops
yes 56.7% 68.2% and may have resulted in actual avoidance behaviour.
Total N 708 703
Since the Keyhole was more prominent in chain B com-
pared to chains A and C, avoidance behaviour would be
Mørk et al. BMC Public Health (2017) 17:239 Page 10 of 11

Table 6 Effect of campaign on purchase behaviour in store

Dependent: Health as one of the purchase motives
Class 1: Class 2: Class 3: p* (main effect) p* (difference
Longer education Short education, Men > 35 years with between classes)
not men >35 years short education
Coefficient Odds ratio Coefficient Odds ratio Coefficient Odds ratio
Intercept −0.755 −1.362 −1.094 .00 .07
Period: Campaign (vs. pre-campaign) −0.215 .806 0.749 2.114 0.247 1.280 .00 .03
Coefficient Odds ratio
Product: Ready meals (vs. cereals) −0.698 .497 .00
Retail chain: B (vs. A) −0.398 .671 .00
Dependent: Looked for nutrition information
Intercept −.944
Period: Campaign (vs. pre-campaign) .469 1.598 .00
Product: Ready meals (vs. cereals) ns
Retail chain: B (vs. A) .286 1.331 .01
*Wald statistic

easier to implement there, with a decrease in transactions the increasingly more common practice of multichannel
involving Keyhole labelled products as a result. Recent re- retailing [27].
sults from an eye-tracking study suggest indeed that con- Several limitations of the study apply. First, the results
sumers from our focal target group (men > 35 with a short are based on analysis of a few selected stores. Second,
education) may try to avoid looking at the Keyhole when the observation/interview part of the study is limited by
making choices [25]. the selection of product categories. The frozen ready
To substantiate the results from the analysis of the meal counter was not very well visited and might be one
transaction data, we carried out observations and inter- of the categories where a nutrition label could even have
views in four out of the six stores. 1411 store interviews a reverse effect. Thirdly, the studies were carried out
were carried out before and towards the end of the cam- within a limited time frame, which means that we did
paign in order to analyse changes in shopping behaviour not measure the long-term effect of the campaign.
with regard to two selected product categories. Results Though the sales data covered a period of 3 weeks after
indicated some increase in having health as a shopping the campaign had ended, this says little about long term
motive and in looking for nutrition information in gen- effects, and no observations and interviews were carried
eral, though not in looking specifically for the Keyhole. out after the end of the campaign. Finally, this was a
The effect on having health as a shopping motive was field study without the possibility of having a control
limited to shoppers having a short education, which may group, which weakens the possibility to make causal
be due to the fact that their likelihood of having health inferences, also our interpretation is backed by the com-
as a shopping motive is lower in the first place. The ef- parison to sales data from the same period the year
fect on looking for nutrition information did not differ before (where there was no campaign).
significantly between consumer segments.
The campaign studied here used a range of different Conclusions
materials, including some to be used in-store. While pro- Overall, the results contain some encouraging signs that
motional material to be used in the shops was a major the Keyhole campaign indeed had impacts on shopping
element of the Keyhole campaign, it turned out that actual behaviour, and that it is possible to address target groups
use of this material in the shops where the study was car- not usually responsive to campaigns on healthier eating,
ried out was limited. This makes it unlikely that the effects if the campaign is especially designed to address this tar-
came about only as a result of what happened in the get group. While this study is specifically on the Key-
stores. On the contrary, it is more likely that the measured hole, the results have implications for the promotion of
effects are the result of the multi-media campaign, in front-of-pack health symbols in general. Apart from
which different communication channels worked together. their attention getting properties, the use of health sym-
Many successful social marketing campaigns with differ- bols is highly dependent on top-down processes linked
ent aims acknowledge the cumulative effects of multi- to the salience of the health motive at the point of
channel communication [26], which are also in line with purchase, and our results suggest that a multi-channel
Mørk et al. BMC Public Health (2017) 17:239 Page 11 of 11

campaign can make the health motive more salient, at 7. Grunert KG, Bolton LE, Raats MM. Processing and acting on nutrition
least for consumer groups where the base rate of health labeling on food. In: Micj DG, Pettigrew S, Pechmann C, Ozanne JL, editors.
Transformative consumer research for personal and collective well-being.
motive salience is low. Further research in this area New York: Routledge; 2012. p. 333–51.
should address more long-term effects, and should 8. Berning JP, Chouinard HH, McCluskey JJ. Do positive nutrition shelf labels
ideally involve a design where transaction data can be affect consumer behavior? Findings from a field experiment with scanner
data. Am Jf Agr Econ. 2010;93:364–9.
linked to individual data. 9. Steenhuis I, van Assema P, van Breukelen G, Glanz K. The effectiveness of
nutrition education and labeling in Dutch supermarkets. Am J Health Prom.
Acknowledgements 2004;18:221–4.
We would like to thank Tino Bech-Larsen, Lise Bundgaard, Liisa Lähteenmäki 10. Sutherland LA, Kaley LA, Fischer L. Guiding stars: the effect of a nutrition
and Gitte Lundberg Hansen for comments on a previous version of this navigation program on consumer purchases at the supermarket. Am J Clin
manuscript. We would like to thank the participating retail chains for Nutrition. 2010;91:1090S–4S.
providing the sales transaction data. 11. Aachmann K, Grunert KG. Mærkningselementer på fødevarer - forståelse og
oplevet relevans blandt forbrugere. Aarhus: DCA - Nationalt center for
Funding fødevarer og jordbrug, (DCA Rapport; No. 009). 2012. Can be downloaded
The research described in this paper has been commissioned by the Danish from http://web.agrsci.dk/djfpublikation/index.asp?action=show&id=1144.
Ministry of Food, Agriculture and Fisheries and is a part of the “Contract 12. Darmon N, Drewnowski A. Does social class predict diet quality? Am J Clin
between Aarhus University and Ministry of Food, Agriculture and Fisheries Nutr. 2008;87:1107–17.
for the provision of research-based policy advice, etc., at Aarhus University, 13. James WT, Nelson M, Ralph A, Leather S. Socioeconomic determinants of
DCA – Danish Centre for Food and Agriculture, 2013–2016”. health. The contribution of nutrition to inequalities in health. Brit Med J.
1997;24;314:1545.
Availability of data and materials 14. Diderichsen F, Andersen I, Manuel C. Ulighed i sundhed - årsager og
The datasets used in this study are available from the corresponding author indsatser. Copenhagen: Department of Public Health, University of
upon reason able request. Copenhagen; 2011.
15. Christensen AI, Davidsen M, Ekholm O, Pedersen PV, Juel K. Danskernes sundhed
Authors’ contributions - Den nationale sundhedsprofil. Copenhagen: Sundhedsstyrelsen; 2014.
TM, MF and KGG designed the study. TM and KGG did most of the writing 16. Verplanken B, Aarts H. Habit, attitude, and planned behaviour: is habit an
of the manuscript. TM coordinated the field work. MF and HJJ analysed the empty construct or an interesting case of goal-directed automaticity? Eur
sales transaction data. GT analysed the observation/interview data and Rev Soc Psych. 1999;10:101–34.
contributed to the writing of the paper. All authors commented on the 17. Bettman JR, Luce MF, Payne JW. Constructive consumer choice processes. J
manuscript and approved the final version of it. Consumer Res. 1998;25:187–217.
18. Gigerenzer G, Gaissmaier W. Heuristic decision making. Ann Rev Psych.
Competing interest 2011;62:451–82.
The authors declare that they have no competing interests. 19. Payne JW, Bettman JR, Coupey E, Johnson EJ. A constructive process view
of decision making: multiple strategies in judgment and choice. Acta
Consent for publication Psych. 1992;80:107–41.
All data were anonymized immediately after data collection and no data are 20. Vyth EL, Steenhuis IH, Vlot JA, Wulp A, Hogenes MG, Looije DH, Brug J,
traceable to any individuals. Seidell JC. Actual use of a front-of-pack nutrition logo in the supermarket:
consumers’ motives in food choice. Public Health Nutr. 2010;13:1882–9.
21. Alkerwi AA, Crichton GE, Hébert JR. Consumption of ready-made meals and
Ethics approval and consent to participate
increased risk of obesity: findings from the Observation of Cardiovascular
There is no review body either at the national Danish level or at Aarhus
Risk Factors in Luxembourg (ORISCAV-LUX) study. British J Nutr.
University that provides ethical approval for this type of study, as only
2015;113:270–7.
biomedical studies need ethical approval (see https://komite.regionsyddan
22. Devi A, Eyles H, Rayner M, Mhurchu CN, Swinburn B, Lonsdale-Cooper E,
mark.dk/wm368978). The study was reviewed internally for compliance with
Vandevijvere S. Nutritional quality, labelling and promotion of breakfast
the ethical guidelines of the MAPP Centre. These guidelines are available
cereals on the New Zealand market. Appetite. 2014;81:253–60.
from the authors upon request and also available at www.mapp.au.dk. In the
23. Julia C, Kesse-Guyot E, Ducrot P, Péneau S, Touvier M, Méjean C, Hercberg S.
interview part of the study, informed consent was (orally) obtained from
Performance of a five category front-of-pack labelling system–the 5-colour
respondents prior to initiation of the interview.
nutrition label–to differentiate nutritional quality of breakfast cereals in
France. BMC Public Health. 2015;15(1):1.
Received: 6 September 2016 Accepted: 25 February 2017
24. Van der Horst K, Brunner TA, Siegrist M. Ready-meal consumption:
associations with weight status and cooking skills. Public Health Nutr.
2011;14:239–45.
References 25. Orquin JA, Lahm E, Redere L, Bagger MP, Tsalis G, Grunert KG.
1. Campos S, Doxey J, Hammond D. Nutrition labels on pre-packaged foods: a Forbrugeradfærd ved valg af nøglehulsmærkede produkter i detailhandlen.
systematic review. Public Health Nutr. 2011;14:1496–506. DCA rapport, nr. 87, 2016. Can be downloaded from http://web.agrsci.dk/
2. Cowburn G, Stockley L. Consumer understanding and use of nutrition djfpublikation/index.asp?action=show&id=1226
labelling: a systematic review. Public Health Nutr. 2005;8:21–8. 26. Evans WD. Social marketing campaigns and children's media use. Future
3. Grunert KG, Wills JM. A review of European research on consumer response Child. 2008;18:181–203.
to nutrition information on food labels. J Public Health. 2007;15:385–99. 27. Dholakia RR, Zhao M, Dholakia N. Multichannel retailing: a case study of
4. Hersey JC, Wohlgenant KC, Arsenault JE, Kosa KM, Muth MK. Effects of front- early experiences. J Interact Mark. 2005;19:63–74.
of-package and shelf nutrition labeling systems on consumers. Nut Rev.
2013;71(1):1–4.
5. Hodgkins CE, Raats MM, Fife-Schaw C, Peacock M, Gröppel-Klein A,
Koenigstorfer J, Wasowicz G, Stysko-Kunkowska M, Gulcan Y, Kustepeli Y,
Gibbs M. Guiding healthier food choice: systematic comparison of four
front-of-pack labelling systems and their effect on judgements of product
healthiness. Brit J Nut. 2015;113:1652–63.
6. Grunert KG, Celemín LF, Storcksdieck genannt Bonsmann SS, Wills JM.
Motivation and attention are the major bottlenecks in nutrition labelling.
Food Sci Tech. 2012;26:19–21.