**1. Introduction**

Food companies need to continue to innovate products to sustain market leadership. Current markets are overloaded with product offerings; thus, the challenge is to innovate new products and update existing products to gain new consumers [1]. The innovation of new products has a positive effect on the economic growth of companies [2]. Innovation helps to develop new market segments, expand current market segments and product portfolios, provide positive image building, and bring new consumers to food companies [3]. The rapid changes in technology, market trends, and consumer expectations (e.g., specific dietary, health, environmental sustainability, and packaging) is keeping the food industry under tremendous pressure to spend large amounts of money on new food product development (NPD) to either increase profits or survive [3–6].

Broadly, NPD consists of four stages, namely opportunity identification, development, optimization, and launch [1,7]. The success of NPD is directly related to several factors: (1) a unique product idea or opportunity; (2) large-scale predevelopment research; (3) superior

knowledge of the market; and (4) a cross-functional team (management, scientist, marketing and launch) collaboration [2,7]. The combination of the first three factors truly determines the quality of the opportunity identification. At this stage, the idea and product developers unearth new areas of opportunities to fulfill the unmet needs of consumers [8,9]. Food companies use three primary sources for new product idea generation, i.e., the marketplace, within the company, and the environment outside the marketplace [1]. Global markets can be excellent places to explore new product ideas because those markets often provide products unknown to the developers [10].

Globalization has integrated regions, companies, markets, and societies from different countries and continents. The internationalization of markets has removed barriers for food availability and consumption and has allowed companies to explore foreign markets for product innovation and idea generation [10]. Food companies have successfully developed global food products by generating ideas and products and one country and moving those ideas or products to other countries; for example, beverages (e.g., Coca-Cola and Pepsi), tea (e.g., Lipton), coffee (e.g., Nescafe), cigarettes (e.g., Marlboro), or chewing gums (e.g., Wrigley). The inclusion of international markets in NPD for generating new product opportunities offers a great diversity of products, customers, and consumers. Food companies use data (consumer involvement, food trends, and environmental factors) most frequently in the opportunity identification and product design stage of NPD [11]. Thus, researchers and food companies need to find both novel and quality opportunities from the market [12]. These gaps (white spaces) could be potential unmet consumer needs that can be filled by developing products for these identified consumer needs [12].

The main task of NPD is to develop products that deliver desired benefits to their intended consumers. Developing consumer-centric products involves great risks and failures [2,3,13]. Fuller [1] identified two main early-stage risk components in NPD: (a) wrong investments in new products that would later fail in the market, and (b) overlooking a potentially successful new product, termed an opportunity loss. Dijksterhuis [14] explained five factors for a high number of new product failures: (1) the uncoordinated efforts of many different functions working on different aspects of consumer and product development; (2) lack of understanding of consumer behavior; (3) usage of outdated research models; (4) lack in seriousness towards behavioral sciences; and (5) high reliability on the notion that good-quality products automatically lead to high sales. Even after producing a large amount of literature on NPD, the failure rate is still very high. Between 2011 and 2013, 76% of the newly launched consumer goods did not survive one year on the market [15], 45% of products remained on the market for less than half a year [14], 75 to 95% of newly developed food and beverage products failed within one year of launch [16].

To increase the odds of NPD success, many researchers recognized the need to consider consumer behavior and choice-based ideas from external global markets [9,17–22]. Sensory science and consumer research provide techniques to identify white spaces in NPD, support research and development, and contribute to minimizing the decision uncertainty [23].

Researchers have identified the early stages in NPD as the most important activities for both product success and failure [1,24]. The early stages of NPD have sometimes been termed as the "fuzzy front-end" because they are looking to take vague ideas and provide some clarity in understanding actual needs. Unfortunately, they also have been called "fuzzy" for reasons such as ill-defined processes, ambiguities, confusion, and ad hoc decisions [18,24]. The early involvement of sensory and consumer research in NPD is recommended as an important success factor [2,23,25,26]. Thus, there is a need for a structured sensory science-based framework in the early stages of NPD for idea generation [3]. The use of techniques such as interviews, focus groups, behavioral observation, ethnography and other such qualitative measures plays an important part of the process for determining and documenting consumer needs [27–33]. In addition, quantitative measures of consumer understanding, attitudes, behaviors, and emotions as they relate to products provide additional information that may be critical to discerning potential product requirements [34–38].

A sensory method called projective mapping (PM) or "napping" is used as a tool to categorize products and discover white spaces among product groups. In PM, assessors position the products (samples) on a two-dimensional space according to the similarities and differences of product characteristics [5,39–42]. PM has been described as a natural, holistic, and spontaneous way for people to describe products. It has been successfully applied to various food products, e.g., orange juice [43], red sufu [44], wine [45], pork [46], peas, and sweetcorn [47]. The influence of extrinsic factors on a consumer's perception of foods such as smoked bacon [48], fermented dairy products [49], and chicken meat [50] as well as packaging [51] also has been studied with PM. Over the years, PM or "napping" has been shown to be efficient, timely, and cost-effective, to obtain a "big picture" overview of a category and is considered a rapid method for gauging some descriptive sensory attributes. The application of PM as a sensory tool for rapid product categorization and characterization for a large number of products is common [52].

The early stage of NPD includes brainstorming and ideation by looking at consumer and market trends. To develop new concepts, researchers and food companies obtain information from competitive food products in the market and then develop concepts for new products. Using descriptive sensory analysis gives an edge to the researcher in a better understanding of competitive products, and of the marketplace where the potential new product will be placed [53]. Descriptive sensory analysis is a classic sensory method used in NPD to profile products on all of its perceived sensory properties [54,55]. It involves the discrimination and description of both quantitative and qualitative sensory attributes by trained sensory panelists [53]. The descriptive analysis offers various applications such as help in understanding the relationship between sensory and instrumental measurements, the relationship between descriptive sensory and consumer preference measurements, product optimization and validation, product profiling, quality control (product comparison), sensory mapping and product matching, shelf life and packaging effect, etc. [53,56–60].

The descriptive profiling of foods helps to identify the main sensory attributes of food products which can be manipulated: (a) to create a profile of desirable sensory characteristics to help in the development and (b) to define early-stage specifications for a new product [53]. The key sensory attributes that are identified help to distinguish the importance of "tangible" product characteristics that form the basis of technical product specifications [18,25]. Sensory characteristics are measurable and can be manipulated, and therefore, characteristics obtained from a wide range of products can encourage the researcher to create a product with different and multiple sensory profiles [53,61]. Descriptive profiling methods have been used to profile many products including products such as bread [62], fresh and dried mushrooms [63] snacks and snack-like foods [64], potato varieties [65], mate tea [66], ground beef [67], and smoked food products [68]. Many sensory studies combined descriptive analysis results with consumer hedonics to determine why food products are liked by consumers [25,69,70]. The combination also helps to identify consumer segments and their specific sensory preferences for certain product characteristics, and also give insight into possible gaps in the marketplace [71,72].

Consumers describe a product's benefits by perceived intrinsic and extrinsic characteristics (e.g., the crispiness of potato chips [73], creaminess in dairy products [74,75], "health, good taste and convenience" [17]. Principal components analysis (PCA) plots generated on descriptive sensory profiling data provide an opportunity to access the positioning and comparison of products in the market space [53]. Using PCA plots, several white spaces (the open space between products) and product clusters can be identified with their identifying main sensory attributes [53]. Those sensory attributes are reported to be directly experienced by consumers to assess products' evaluation and significantly influence consumer product appraisal [76]. The "white spaces" suggest areas where new products could be developed to meet unmet needs [10,77–79]. However, the presence of white space does not necessarily mean that (a) products do not exist in that space but only that they were not part of the study, (b) just because a product is made to fill the space

that the product will succeed, or (c) it is impossible to develop a product that fits the white space based on current technology.

A goal of this project was to highlight one strategic framework to find white spaces in the marketplace and then develop new snack texture concepts to fit the sensory concepts identified as white spaces. The specific objectives were to (a) find the new texture and flavor gaps in several large-scale markets; (b) identify key sensory texture characteristics of the Japan (JP) and South Korea (SK) snack foods; and (c) to demonstrate how unfamiliar marketplaces can be used in NPD for ideation. This study is a continuation and expansion of earlier work [10].

#### **2. Materials and Methods**

#### *2.1. Materials*

One hundred and twenty-three packaged snacks from Seoul and Busan, SK, and ninety-five packaged snacks from Kyoto, JP, were purchased in-country and shipped to the Center for Sensory Analysis and Consumer Behavior, Kansas State University (KSU), United States (US). Although a wide range of products other than those thought of as traditional snack foods are eaten as snacks [80], fresh fruits, candies and confectionary products often eaten as desserts were excluded from this study to focus on foods that were made and marketed primarily as snack foods. Trained sensory scientists and product developers from the US, China, India, and SK purchased snacks for this study, following a product procurement strategy recommended by Murley [10] to help ensure that the wide range of snack foods and types was represented. Package guidelines were followed for storage and handling.

#### *2.2. Snacks Data Bank*

Information related to each snack type such as product name, product description, manufacturer, package size, number of packages, ingredient list, and pictures (front and back) were collected to develop a snack data bank for each country (See the JapaneseSnacksDataBank.xlsx and SouthKoreanSnacksDataBank.xlsx at https://krex.k-state.edu/ dspace/handle/2097/40897) (accessed on 2 January 2021). The collected data helped in product identification, product cataloging, and most importantly in knowledge generation about various snack foods such as packaging data, as well as the ingredient and nutritional data. Several authors concluded that knowledge generation on market products, and its proper integration with organizational learning are important aspects of NPD [3,13].

#### *2.3. Projective Mapping*

PM was used in its original concept as described by its authors with few modifications [42,81]. A subset of purchased snacks with the most diversified texture profile, new ingredients, and novel concepts were selected from each country for PM. Fifty-one snacks from JP and sixty-six from SK were included in the PM. The modalities used for PM were texture (hard to soft texture perception) and flavor (savory to sweet flavor perception). Snack foods were sorted for similarities and dissimilarities on the aforementioned modalities. The panel determined the key aspects for placement. The snacks were tasted blind with only a two-digit code and sorted into groups by six trained sensory scientists with prior experience in snack food evaluation.

Two 1 h training sessions were held to orient the panelists with products; training included tasting samples. PM was performed on a rectangular des covered k; the center of the desk was labelled for axis interaction and extreme ends were labeled exactly the same as represented in Figure 1. Panelists evaluated one sample at a time, discussed and reached consensus on positioning the samples. The number of samples evaluated in each session was restricted to ten samples. Additional sessions were held after a wait of at least 1 h. When all the samples for a country had been tested, discussed, and placed on the desktop, the panelists reviewed the placement and made any final modifications. At that point, the "x" and "y" coordinates of the desk were measured to provide the specific data for each

sample. Water and unsalted crackers were used as palate cleansers. The products were grouped subjectively (based on the perceived texture and flavor evaluation).

**Figure 1.** Projective mapping plot of the fifty-one JP snacks showing nine product groupings and outlying products (snacks are coded with 2-digit numbers and snacks with the same color are in the same group). The products' grouping was subjective.

#### *2.4. Snacks Sensory Description*

After PM the entire set of products and examining the results, 20 snacks from each country were selected to represent the entire map and were screened for descriptive sensory profiling. To increase the product pool size, 12 new snack products from each country were also added for descriptive profiling. The parameters used to screen snacks included the coverage of the map surface and the selection of diversified and novel textures, new ingredients, and novel concepts. The screened snack foods are listed in Table 1 (for JP) and Table 2 (for SK). In addition, three snacks (Stacy's pita original, Lay's classic potato chips and Tostitos original corn chips) widely available around the world also were included in the test to provide a "reference" set of products that could help anchor the maps. This also allows other researchers to help better understand the similarities and differences shown on the map, particularly because many would never have seen or tasted the products tested.


**Table 1.** List of the Japan (JP) snacks screened for descriptive profiling.

<sup>1</sup> Products without a manufacturer listed are snacks either sold "on the street" or in local "snack" shops in packages without a label. \* PM code is a 2-digit number used in the projective mapping plot as an identifier for snack samples. \*\* Group numbers are provided to identify the snack sample association in projective mapping grouping. New = a product added to this study; Anchor = a common U.S. product used for comparison purposes.


**Table 2.** List of South Korea (SK) snacks screened for descriptive profiling.

<sup>1</sup> Products without a manufacturer listed are snacks either sold "on the street" or in local "snack" shops in packages without a label. \* PM code is a 2-digit number used in the projective mapping plot as an identifier for snack samples. \*\* Group numbers are provided to identify the snack sample association in projective mapping grouping. New = a product added to this study; Anchor = a common U.S. product used for comparison purposes.

#### *2.5. Descriptive Profiling*

Consensus methodology was used to develop sensory attributes, definitions, and references [55,82]. Panelists and the sensory analysts determined attributes for further rating by consensus. The final list of attributes was kept consistent for both JP and SK snacks. The snacks were profiled for flavor, amplitude, appearance, and texture attributes. However, because the flavors of many snack foods can be easily changed based on consumer preferences and many of the snacks tested come in many different flavors, only

appearance and texture attribute data were considered in this analysis and are shown in this paper. The texture terms used in descriptive profiling were adopted from the snacks texture lexicon published by Kumar and Chambers [64].

Panelists used a scale ranging from 0 to 15.0 with 0.5 increments where 0 represents none and 15 extremely strong to profile snack samples. Each panelist independently allocated intensities to the attributes and then the intensities were discussed within the panel to reach a single consensus score for each attribute for each product. Three samples were evaluated in each session. Panelists cleaned their palates between samples with freshly cut cucumbers, mozzarella cheese (manufactured by Kroger, Cincinnati, OH, USA), hot water, and a washcloth for the cleaning of lips and hands. The descriptors list, definitions and reference standards are provided in Supplementary Table S1. Similar methodology has been used in other recent studies for the sensory profiling of various foods, e.g., [62–66,82,83].

#### *2.6. Sample Preparation*

The snacks used were all ready to eat and needed no preparation; they were served as they were. The samples were blind coded with three-digit codes, served in 8 oz (Styrofoam) and 3.25 oz (plastic) cups (based on the size and shape of the snacks) covered with a lid. One sample at a time was served to panelists in a randomized order.

#### *2.7. Panelists*

For the PM, six sensory analysts with experience in snack food evaluation served as the panel for the study. All of the analysts had training in PM techniques and worked as a group to produce a single joint map of snacks for each country. The panelists were trained to specifically focus on the texture and flavor stimuli. The assessors were tasked to screen the large pool of samples, position them on a 2-dimensional space by reaching a consensus on the general differences on texture perception. The objective of PM using a trained panelist was to layout an overall product space rather than generate data through scaling differences. After PM, the descriptive study was planned to identify the subtle difference and quantify descriptors among different panelists.

For the descriptive analysis, six highly trained descriptive sensory panelists were used for this study. Each panelist had more than 120 h of training in descriptive panel training and more than 1000 h of descriptive testing experience with various types of foods and beverages, including extensive testing on different snack type products. The panelist worked on evaluation techniques for appearance, texture, and flavor perception. The panelist received 9 h of additional orientation with both the JP and the SK snacks. The number of highly trained panelists who participated in this study was sufficient to differentiate the samples in the descriptive analysis [84–87] and similar panels have been used in other studies [66,88–92].

## *2.8. Data Analysis*

Correlation-based principal component analysis (PCA) and agglomerative hierarchical clustering (AHC) were performed on the sensory descriptive data using data analysis software XLSTAT 2019.3.2.61545. To prevent data redundancy, attribute correlations were analyzed by the data analytical software R-studio version 4.0.0 (R Foundation for Statistical Computing, Vienna, Austria; https://www.R-project.org/) (accessed on 10 January 2020). Note that for consensus profiling, because there is no variance in scores, "significant" differences are not determined [55,82]. Instead, the size of intensity differences deemed "important" is determined in advance by researchers. In this case, differences were deemed important if they varied by ≥0.5 points, a typical level used in such studies.

#### **3. Results**

The sequential use of sensory tools produced information on the main sensory descriptors, the snacks market categorization based on sensory descriptors, existing snacks market space, and white spaces (potential opportunities). All that information was produced by

the PM plots and subsequent PCA mapping along with the original data. The information can be used by a snack manufacturer to (a) have an overview of the snack markets (based on sensory parameters); (b) identify the major flavors, textures, and possible trends; (c) learn about a competitor's product positioning; (d) develop new concepts to bring to further sensory (including consumer) research; and (e) enhance their product snack portfolio. The results explain how this information can be generated using JP and SK snacks as examples.

#### *3.1. Projective Mapping*

The representative maps of the PM results are presented in Figure 1 (for JP) and Figure 2 (for SK). The snacks are coded with two-digit numbers for representation purposes.

**Figure 2.** Projective mapping plot of the sixty-six SK snacks showing nine product groupings and outlying products (snacks are coded with 2-digit numbers and snacks with the same color are in the same group). The products' grouping was subjective.

#### 3.1.1. Japanese Snacks

Fifty-one snacks with a variety of texture profiles were sorted into nine groups (Figure 1). The PM was primarily focused on the textural dimension from a hard to a soft texture. Because the snacks were seasoned with different types of flavors, sorting them based on flavor was much too difficult for a 2-dimensional space. The only flavor dimension that was considered was that from savory to sweet. All the products were analyzed visually, in the hand (tactile hand feel), and tested orally (for texture and flavor) by the sensory scientists who participated in the PM.

Out of 51 snacks, 33 snacks (64.71%) were considered as hard bite textures, ranging from moderately to extremely hard. The main texture descriptors were crispiness, crunchiness, sustained crispiness, sustained crunchiness, and hardness. The largest snacks group (group-1) had 14 products (for example, crackers, wafers, puffs, and rolls), representing 27.45% of total snacks. Similarly, group-6 had four snacks, grouped for extremely hard texture and strong savory flavor. Group-2 had six snacks (for example, corn trumpets, corn puffs, squid crackers, shrimp crackers, cheese-filled sticks, and unbranded grain crackers), representing a soft-bite texture with of the mild savory flavor category. A complete list of the JP snack food groups is presented in Table 3.

**Table 3.** Group identified in the projective mapping of the JP snacks. Group number, number of snacks in each group, texture, snack type, flavor, and snack names.


Group-1 represents the largest portion of JP snack foods from the selected snack pool. The results suggest that most JP snack foods are hard to bite texture snacks seasoned with various flavors such as savory, bland, and plain salt. Group-1 and -5 differed in terms of flavor intensities but were similar on textural dimensions. Collectively, snacks from groups-1, -5,-7,-8, and -9 formed a large hard texture block (highlighted with a red border) (Figure 1). The hard texture block accounted for 49% of the overall JP snacks market space. Hard texture snacks appeared to dominate the JP snacks market, which has a large number of existing products. The possible explanations could be (a) JP consumers prefer hard texture (crunchy and crispy) snacks; (b) our research team inadvertently collected more hard texture snacks and therefore limited the product pool; or (c) it is a true representative of the JP snacks market. Hence, for a new product developer, understanding the texture dimensions of JP snacks could be a potential framework or area of interest to explore either as copycat products (harder textures) or to create new textures (e.g., at the softer texture end of the spectrum). Of course, another niche area could be bringing new flavors into the existing texture spectrum where flavors may be lacking.

Thirty-three snacks (65%) were savory, including snacks seasoned only with plain salt. Other flavors (for example, seafood, seaweed, prawns, squid, crab, and fish) also were

present in that grouping. Savory flavored snacks occupy the largest space in the JP snack market. Thus, for a product developer, a savory flavor could be an easy carry-over from one snack type to another, but also positions the product against a larger competitive set.

The broad range of textures and flavor, some of which were not found in tests conducted on snacks from other countries represent a new opportunity for manufacturers to transfer ideas from one country and culture to another. Taking ideas for new products from countries with a plethora of products often is an easy way to create new products for countries where existing products may be in more limited supply or exist in fewer sensory segments.

The gaps between the product grouping are the white spaces where no products were found to exist. Those empty spaces are potentially unexplored opportunities in the JP snack market and perhaps in other markets. The bottom half of the plot in Figure 1 represents the soft texture snacks space. More white space is available in soft texture snacks over hard texture snacks. This may be because (a) a smaller number of products are in the soft texture product pool (a potential opportunity), or (b) the JP consumer does not prefer soft texture snacks. If soft texture snacks are not as popular in various countries, they may not be a real opportunity. For JP, the further investigation of that snack segment is required in terms of consumer studies. For other countries, the opportunity for new snack development in the sweet category needs to be considered and further research with potential new products may be warranted. In addition, spaces that are not filled with many products also may be considered "white" spaces. For example, the space between group-1 and group-7 has only five products (i.e., group-5). Considering the number of products that exist in other areas of the map, more products could be developed to fill and position in this space.

The plot can be divided into four quadrants (Figure 1). The first quadrant (Q1) represents hard texture snacks with a sweet flavor, the second quadrant (Q2) is hard texture snacks with a savory flavor, the third quadrant (Q3) is soft texture snacks with savory flavor, and the fourth quadrant (Q4) is soft texture snacks with sweet flavor (Figure 1). Each quadrant produces different information. For example, Q4 and Q1 have the least number of snacks and more white spaces. A product developer can develop a wide range of new textures (hard to soft) with sweet flavors. The market space offered in these two quadrants is quite large. Similarly, other quadrants can be used to frame initial product concepts, either individually or in combination with other quadrants.

From a broader perspective, the plot can be divided into two halves. If a product developer is interested in new snack flavors, they can divide the plot on the vertical axis (Figure 1). For example, the left half of this plot, vertically divided, characterizes the savory flavor market space ranging from a hard to a soft texture. The right half of the plot represents the sweeter flavors market space with the same texture range from hard to soft. If the plot is divided into two halves on the horizontal axis, the top half contains all hard texture snacks with both sweet and savory flavors. The bottom half of the plot comprises all softer texture snacks spreading across savory and sweet flavor. There is a wide range of options that could be explored in soft texture with savory flavors. For example, there was no "soft texture, non-seafood" savory snack found in this study. Only 18 snacks were of soft texture, mainly groups-3 and -4. Group-3 consists of fish or seafood flavored soft chewy snack loaded with strong sour-savory flavors. In addition, group-4 snacks were soft textured sweet snacks but not chewy. Considerable white space is available across the savory-sweet flavor dimension with a soft texture profile that may help the developer in identifying additional products for the market.

One issue that must be considered is that many softer textured snacks were found when conducting the initial product search. However, many of those were in the form of freshly prepared "street snacks", such as fresh seafood or egg products that could not be sold in a shelf-stable manner given current technologies. Those products may be considered as inspiration for manufactured shelf-stable products but also represent a competitor that is not directly accounted for in this research.
