Drivers of Liking for Oat Milk

ABSTRACT

Oat milk is one of the most popular plant‐based milk alternatives in today's market. The objective of this study was to identify the sensory drivers of like and dislike for oat milk. Twenty‐eight commercial oat milks were collected in duplicate lots. A highly experienced trained panel (n = 7) identified and documented attributes and definitions for the oat milks. Ten representative oat milks were selected for consumer acceptance testing with oat milk consumers (n = 157). Each consumer evaluated the 10 oat milks across 2 days. External preference mapping and penalty lift analyses were then applied to identify drivers of like and dislike. By external preference mapping, opacity, sweet aromatic and cooked cereal flavors, viscosity, and residual mouthcoating were drivers of liking for oat milk consumers, whereas thin/watery texture and cardboard flavor were drivers of dislike. Two clusters of oat milk consumers were identified which differed primarily by degree of liking for all oat milks. The determination of a sensory lexicon specific for oat milks and drivers of liking and dislike can be used by product developers for formulating the ideal oat milk and to improve current products.

1. Introduction

The plant‐based (PB) milk category is the most developed PB category, even surpassing PB meat, and is valued at $2.9 billion (Good Food Institute ²⁰²⁵). PB milk accounts for 36% of the total PB food market, making it the largest PB category (Good Food Institute ²⁰²⁵). Over the past decade, the PB food market has experienced significant growth in the United States. Insights released by the Good Food Institute (GFI) and the Plant Based Foods Association (PBFA) show that in 2023, the US retail PB food market was worth $8.1 billion (Good Food Institute ²⁰²⁵). The global PB milk market size was valued at $2.8 billion in 2022 and is projected to reach $7.3 billion by 2032, growing at a CAGR of 10.3% from 2023 to 2032 (Allied Market Research ²⁰²⁵).

PB milks, also called PB milk alternatives (PBMAs), are water‐based extracts made from nuts, seeds, and other foods. Usually, PBMAs are classified on the basis of their primary ingredients: cereal‐based (oat, rice, etc.), legume‐based (soy, pea, etc.), nut‐based (almond, cashew, coconut), seed‐based (hemp, flax, sesame), and pseudocereal‐based (quinoa) (Sethi et al. 2016). Oat, soy, almond, and coconut milk are some of the widely recognized PBMAs. The percentage change of the total US retail sales of PBMA in 2021–2023 was −3.7 for almond milk, −0.2 for soy milk, 32.4 for coconut milk, and 115 for oat milk: making oat milk the fastest growing category (Mills 2024). It is forecasted that by 2029, retail sales of oat milk will rise to a total of $1 billion, a 10.3% increase compared to 2023 sales (Mills 2024).

The emergence of PBMAs can be attributed to a number of factors, including an increase (or perceived/self‐diagnosed increase) in lactose intolerance and milk protein allergies, as well as health concerns, sustainability, and the health benefits of PB foods (Jaeger and Giacalone 2021; Tziva et al. ²⁰²⁰). Consumers consider PB proteins and PBMAs to be more sustainable and environmentally benign than dairy proteins (Keefer et al. 2024; Schiano et al. ²⁰²²; Yang and Dharmasena ²⁰²¹). PBMAs provide a variety of options for consumers. Oat milk particularly provides an option for consumers who are looking for an allergen‐free milk alternative. It is important to stipulate that across all types of milk, dairy milk remains the category leader ahead of all PBMAs. When the actual market shares of dairy milk and PB milk are compared, cow's milk still constitutes 87% of the market. Annual sales of dairy milk are $24.9 billion, whereas PB milk comprises $2.2 billion (Ramsing et al. 2023).

Despite interest in PB foods, consumers are also concerned with cost and sensory quality (Moss et al. 2022). Extant literature indicates that consumer dairy preference poses a serious barrier to consumer acceptance of PBMAs (Cardello et al. 2022; Collier et al. ²⁰²²; Schiano et al. ²⁰²²). PBMAs typically cost more than dairy milk (Ramsing et al. 2023; Redan et al. ²⁰²³). Moss et al. (2022) and Tonsor and Wolf (2019) found clear evidence that concern for farm animals did not translate to consumers paying a price premium for PBMAs. The absence of desirable sensory qualities is also frequently cited as reasons for dislike of PBMAs (Cardello et al. 2022, 2024; Jaeger et al. ²⁰²⁴). Beany, earthy, salty, and bitter are a few of the off‐flavors attributed to PBMAs (Cardello et al. 2024). Consumer perception and drivers for PBMAs as a category have been studied (Moss et al. 2022; Jaeger and Giacolone ²⁰²¹; Jaeger et al. ²⁰²⁴), and drivers of liking specifically for soy milk have been investigated (Lawrence et al. 2016). However, to our knowledge, no studies have examined oat milks in particular. The objective of this study was to determine the sensory attributes that drive liking and dislike of oat milk.

2. Materials and Methods

2.1. Experimental Overview

A category survey with trained panel profiling was conducted on 28 commercial oat milks (Figure 1). Representative oat milks (n = 10) were subsequently evaluated by 157 oat milk consumers. Both internal and external preference mapping were applied to identify consumer likes and dislikes for oat milks. Descriptive analysis and consumer testing were conducted in accordance with the North Carolina State University (NCSU) Institutional Review Board for Human Subjects guidelines (IRB exempted protocols 3481 and 26974).

FIGURE 1.

Experimental overview of oat milk liking study. A number of oat milks and panelists are represented by k and n, respectively.

2.2. Descriptive Analysis

Duplicate lots of 28 commercial oat milks (original/unflavored, regular, and extra creamy) were purchased locally or online approximately 4 weeks apart (Table 1). The 28 profiled oat milks included original version (n = 12); extra creamy/full fat (n = 7); less sweet/zero‐sugar/unsweetened (n = 5); organic (n = 4); barista (n = 2); shelf‐stable (n = 1); and low‐fat (n = 1). Oat milks were evaluated no less than 10 days before the expiration date. Milks were stored in the dark at 4°C. An experienced descriptive analysis panel (n = 7, 5 females and 2 males, ages 23–53 years) evaluated the oat milks. Each panelist had a minimum of 50 h of previous experience with sensory profiling of products, including plant proteins and PB milks, using a 0‐to‐15‐point universal intensity scale consistent with the Spectrum descriptive analysis method (Meilgaard et al. 2006). An additional 22 h of calibration was dedicated specifically to oat milks. During panel calibration (11 sessions, approximately 2 h each), panelists tasted oat milks and other PB milks and identified sensory attributes relevant to oat milks (Table 2). A lexicon that consisted of 2 visual attributes, 12 aromatics, 5 texture/mouthfeel attributes, and 5 basic tastes was identified. Following lexicon identification, calibration sessions focused on panelist and panel consistency in identifying attributes and attribute intensities in oat milks. Panelist repeatability was monitored to confirm that individual panelist standard deviation was less than 10% of the scale range (1.5 points on the 15‐point scale) (ASTM International ²⁰¹⁸). During final panel calibration sessions, evaluation of preliminary data demonstrated that the panel was able to consistently document statistical differences among samples for the identified attributes (Drake and Civille 2003; Chambers et al. ²⁰⁰⁴). The consistent detection of differences (p < 0.05) among samples indicated a well‐performing panel and readiness for formal data collection.

TABLE 1.

List of oat milks used in the study.

Blind code	Category	Total sugar (g/240 mL)
1	Original	3
2	Full fat	7
3	Zero sugar/Unsweetened	0
4	Original	7
5	Original	4
6	Zero sugar	0
7	Extra creamy	3
8	Barista and shelf stable	0
9	Unsweetened	1
10	Original	3
11	Organic	5
12	Original	7
13	Extra creamy	4
14	Extra creamy	7
15	Original	5
16	Barista	7
17	Unsweetened	0
19	Original	8
20	Low fat	7
21	Extra creamy	4
22	Extra creamy/Organic	7
23	Original	8
25	Original and organic	7
27	Extra creamy	7
29	Original	7
30	Less sweet	9
31	Original/Organic	4
32	Original	12

Note: Samples selected for consumer acceptance testing are indicated in bold. The listed oat milk samples were categorized on the basis of product packaging/labels.

TABLE 2.

Sensory lexicon for oat milks.

Category	DA term	Definition/Reference	Example/Preparation
Visual	Color intensity a	Visual term referring to the intensity of the color of the rehydrated solution from light to dark
	Opacity a	Visual term referring to the degree of opacity of the solution	Water = 0 Whole fat fluid milk = 12
Aromatics	Overall aroma intensity b	The overall orthonasal aroma impact	Evaluated as the lid is removed from the cupped sample
	Sweet aromatic c	Sweet aromatics associated with white cake mix or vanillin	Dilute 5 mg of vanillin in skim milk
	Overall flavor impact	The overall intensity of flavor in the sample, from mild to strong
	Beany d	Aromatics characteristic of beans	Canned great northern beans
	Cardboard a	Aromatics associated with wet cardboard and brown paper	2 cm × 2 cm piece of brown paper bag boiled in water for 30 min
	Cooked cereal	Aromatics associated with cooked steel cut oats in water	1 cup of oats with 5 oz of water cooked at medium heat for 5 min
	Fatty/Potato c	Canned white potato slices	Remove the sliced potatoes from the broth
	Green grassy b	Aromatics characterized by cut grass and unripe or green fruit	Hexanal, 200 ppm in a sniff jar
	Nutty/Roasted a	Aromatics associated with roasted nuts	Roasted, unsalted soynuts
	Oily fish d	Aromatic associated with fish oil as found in mackerel, canned sardines, or cod liver oil	Dried bonito shaving, canned tuna in oil
	Oily veg d	Overall perception of heated oil aromatics commonly associated with products containing oil or fat	Ritz cracker
	Raw oat	Characteristic aromatic from raw steel cut oats
Basic taste	Sweet taste c	Basic taste elicited by sucrose	5% sucrose solution
	Salty taste c	Basic taste elicited by NaCl	2% NaCl solution
	Bitter taste b	Basic taste elicited by various compounds including caffeine and quinine	0.5% caffeine solution
	Umami taste b	Chemical feeling factor elicited by monosodium glutamateb	1% monosodium glutamate in water
	Aftertaste intensity	The overall intensity of taste after swallowing the sample, from mild to strong
Texture/Mouthfeel	Astringent mouthfeel b	Chemical feeling factor characterized by a drying or puckering of the oral tissues	Soak 6 black tea bags (Lipton) in 500 mL water for 10 min
	Viscosity b	Attribute evaluated in the mouth, place product in mouth (approx. 1 tsp), evaluate the rate of flow across the tongue	Water = 1 Heavy cream = 3 Sweetened condensed milk = 12
	Chalky a	The degree to which fine particles are perceived in the mouth	Whole fat fluid milk = 0 Sour cream with instant cream of wheat cereal added = 5
	Gritty e	Degree to which gritty, sandy texture is perceived (these are small fine particles)	Fresh or canned pears (gritty or sandy)
	Residual mouth coating f	The amount of residue remaining in the mouth after expectorating the sample

^a Russell et al. (2006).

^b Liu et al. (2021).

^c Drake et al. (2003).

^d Cherdchu et al. (2013).

^e Oliver et al. (2018).

^f Drake and Delahunty (2017).

Oat milks (80 mL) were dispensed into 118 mL transparent souffle cups with 3‐digit blinded codes and lidded. Oat milks were tempered to 15°C for descriptive analysis. Each lot of each oat milk was evaluated in triplicate (six replications per oat milk). Panelists evaluated six to seven samples of oat milk per session. Fourteen sessions were required to evaluate all samples. These sessions were spread across 2 weeks with two sessions each day (morning and afternoon). Each panelist evaluated samples in a randomized order. A 3‐min rest was enforced between every sample. Panelists cleansed their palates by taking a bite of unsalted cracker between each sample and rinsing their mouths with bottled spring water. Data were collected using an electronic ballot on the NCSU secure server.

2.3. Consumer Testing

On the basis of the examination of descriptive analysis principal component (PC) biplots and evaluation of market share, 10 representative oat milks were selected for consumer acceptance testing (Figures 2 and 3). Consumers were recruited from a database of more than 10,000 consumers maintained by the Sensory Service Center at NCSU. Consumers aged 18–64 years that consumed PB dairy alternatives at least a few times per month and consumed oat milk at least once in the last 4 months were selected for the study. The test was conducted across 2 days with a partial presentation of 5 samples per day. The design for sample presentation was randomized and balanced across both days. Oat milks (80 mL) were served at 4°C in 118 mL lidded transparent souffle cups. A 3‐min rest was enforced between samples, and consumers were provided with unsalted water crackers and bottled spring water for palate cleansing. Compusense Cloud version 23.0.3 (Compusense Inc., Guelph, Canada) was used to collect the data.

FIGURE 2.

Principal component biplot of sensory attribute means of oat milks by trained panel profiling (PC 1 and PC 2). Numbers represent oat milks. The samples highlighted were selected for consumer acceptance testing. PC 1—opacity, sweet aromatic, raw oat, cooked/cereal, cardboard, sweet taste, bitter taste, astringency, and residual mouthfeel. PC 2—overall aroma intensity, overall flavor impact, salty taste, and aftertaste intensity.

FIGURE 3.

Principal component biplot of sensory attribute means of oat milks by trained panel profiling (PC 1 and PC 3). Numbers represent oat milks. The samples highlighted were selected for consumer acceptance testing. PC 1—opacity, sweet aromatic, raw oat, cooked/cereal, cardboard, sweet taste, bitter taste, astringency, and residual mouthfeel. PC 3—nutty/roasted flavor.

The ballot consisted of questions about appearance, color, overall liking, flavor, sweetness, texture, thickness, and aftertaste liking, as well as quality, expectations, and purchase intent of the oat milks. Consumers also described oat milks using a check‐all‐that‐apply (CATA) question. Liking questions were scored using a 9‐point hedonic scale. Just about right (JAR) questions utilized a 5‐point scale where 1 and 2 = too light/not enough, 3 = JAR, and 4 and 5 = too dark/too much. Consumers evaluated appearance liking and color JAR questions prior to tasting oat milks. Consumers then evaluated overall liking. After that, flavor liking and flavor JAR were asked followed by sweetness liking and sweetness JAR questions. Texture/mouthfeel liking, thickness liking, and thickness JAR were then asked. Consumers were then asked to select all applicable descriptors from a list of attributes to describe the sample (CATA). The CATA attribute list was developed on the basis of the DA sensory lexicon (Table 2) and modified to include more consumer‐friendly terminology (i.e., “vanilla flavor” instead of “sweet aromatic”). Consumers then evaluated the aftertaste pleasantness of the samples. Next, consumers scored the quality of the oat milk on a 5‐point scale where 1 = extremely low quality, 2 = low quality, 3 = neither high nor low quality, 4 = high quality, 5 = extremely high quality followed by a purchase intent question. Purchase intent was scored on a 5‐point scale where 1 = definitely would not buy, 2 = probably would not buy, 3 = maybe/maybe not buy, 4 = probably would buy, 5 = definitely would buy.

Final questions addressed usage and attitudes towards oat milk. One question addressed usage of oat milks: consume oat milk on its own, add to cereal/oatmeal, add to smoothie, add to coffee/tea, use as an ingredient, or I would not use this oat milk. Questions also addressed brand as a CATA and as a single select. Finally, consumers were asked to select the factors that influence their choice of dairy alternative milks/PB milk using a CATA question with the following options: Appearance, clean label, cost, flavor, health/nutritional value, more varieties, package type, protein content, sustainability, texture, and other (specify). A total of 157 consumers completed the test and received a $30 Amazon e‐gift card.

2.4. Data Analysis

Analysis of variance (ANOVA) with means separation (Fisher's least significant difference [LSD]) was applied to trained panel data to identify differences among oat milk attributes. PC analysis (PCA) was used to identify the primary multivariate “big picture” patterns in the trained panel data, and biplots were generated to visualize these sensory differences among samples. Consumer data were analyzed by univariate and multivariate analysis. ANOVA was conducted on overall liking scores with means separation (LSD). Penalty analysis was performed on JAR scores with K‐proportions and the Marascuilo procedure to determine statistical letterings. The Kruskal–Wallis test was used to evaluate 5‐point questions. Penalty lift analysis was applied to the consumer attribute CATA data with overall liking scores to determine the impact of checked attributes on overall liking. Consumers were grouped into clusters by agglomerative hierarchical clustering (AHC) with k‐means on overall liking scores. Drivers of liking were identified by external preference mapping of trained panel data and consumer liking scores using the PrefMap procedure of XLSTAT. CATA data were also analyzed using correspondence analysis (CA). All statistical analyses were performed with XLSTAT (version 2023.3.1, Addinsoft Inc., New York, NY, USA) and were carried out at a 5% significance level.

3. Results and Discussion

3.1. Descriptive Analysis

The identified lexicon comprised 2 visual attributes, 12 aromatics, 5 basic tastes, and 5 mouthfeel/texture attributes (Table 2). Of these attributes, 19 were documented in all oat milks evaluated, whereas other attributes were only documented in a few oat milks. Many of these attributes were previously documented in other plant milks or plant proteins (Lawrence et al. 2016; Liu et al. ²⁰²¹; Nishku ²⁰²⁰). Oat milks were differentiated using the identified lexicon (p < 0.05), and PCA was applied to visualize differences among the 28 oat milks (Figures 2 and 3). A total variability of 54% was explained by three dimensions. PC 1 explained 27.9% of the variability and comprised opacity, sweet aromatic, raw oat, cooked/cereal, cardboard, sweet taste, bitter taste, astringency, and residual mouthfeel, whereas PC 2 explained 16.6% of the variability and comprised overall aroma intensity, overall flavor impact, salty taste, and aftertaste intensity. PC 3 with 9.7% variability comprised nutty/roasted flavor. Oat milks within each category (Table 1) had some expected similarities along with some differences. Oat milks with no added sugar (Samples 3, 6, 9, 17, and 30) tended to be lower in sweet taste compared with other oat milks (Table 3). Extra creamy milks (Samples 7, 13, 14, 21, 22, and 27) were expected to have higher viscosity and residual mouthcoating intensities (Table 3), but other oat milks (17, 23, and 32) also had high viscosity and mouthcoating as did milks 4, 25, and 29 for oat milk which may explain variability among the samples. Consequently, most of the oat milks were spread throughout the sensory space (Figures 2 and 3). Ten representative oat milks were selected for consumer testing based on the variability explained by the trained panel sensory attributes and market share (Figures 2 and 3).

TABLE 3.

Trained panel mean attribute intensities for each oat milk.

Sample	Color intensity	Opacity	Overall aroma intensity	Sweet aromatics	Overall flavor impact	Card‐board	Cooked cereal	Fatty/Potato	Nutty/Roasted	Raw oat	Sweet taste	Salty taste	Bitter taste	Umami taste	Aftertaste intensity	Astringent mouthfeel	Viscosity	Chalky	Residual mouth coating
1	6.9	8.4	2.0	1.5	2.9	0.8	2.8	0.0	0.7	1.0	2.0	1.9	ND	1.0	1.4	1.8	2.1	ND	1.4
2	4.2	11.6	3.0	2.0	3.7	0.6	2.7	1.2	0.5	0.7	2.7	2.0	ND	1.0	1.7	2.0	2.2	ND	2.0
3	2.3	8.7	1.5	0.9	2.1	ND	2.0	ND	0.7	1.8	1.6	1.1	ND	1.1	1.1	1.9	1.9	ND	1.2
4	6.2	10.8	1.8	1.7	2.4	ND	2.1	1.1	1.1	1.8	2.9	2.0	ND	1.2	1.3	1.6	2.1	ND	2.1
5	4.4	9.3	2.1	0.7	2.6	1.0	2.0	1.0	0.6	2.5	1.9	1.9	ND	1.3	1.4	1.8	2.0	1.0	1.6
6	3.9	8.9	3.6	0.7	4.0	1.5	1.1	ND	1.0	2.9	1.4	1.7	ND	1.1	2.4	2.1	1.9	ND	1.1
7	4.8	9.1	2.9	1.8	3.0	ND	2.1	ND	1.1	1.6	2.0	2.0	ND	1.4	2.2	2.5	1.9	0.5	1.1
8	3.1	10.0	2.0	0.6	3.2	2.1	1.1	ND	ND	1.2	1.5	1.9	ND	1.4	2.1	3.0	2.0	ND	1.1
9	6.0	9.0	1.3	0.6	3.2	1.9	1.1	ND	0.9	2.4	1.8	1.8	1.0	1.5	2.0	2.5	1.9	1.5	1.1
10	5.4	8.6	2.1	2.0	2.1	ND	1.6	ND	1.3	1.4	1.8	1.5	ND	1.0	1.5	2.0	1.9	0.7	0.9
11	2.0	9.3	1.5	1.0	2.0	1.5	1.5	ND	ND	1.5	2.3	1.4	0.9	1.0	1.4	2.2	1.8	0.9	1.1
12	5.7	9.0	2.1	1.8	1.9	ND	2.4	0.8	1.0	1.5	2.0	1.3	ND	1.1	1.5	1.7	1.9	0.6	1.5
13	5.5	8.3	2.3	1.2	2.0	ND	1.9	ND	1.0	1.6	2.1	1.5	ND	1.0	1.5	1.9	2.1	ND	1.2
14	2.1	11.5	1.9	1.8	2.1	ND	2.3	ND	1.3	2.1	2.0	1.4	ND	1.1	1.4	2.0	2.1	ND	2.0
15	6.4	5.5	2.1	1.4	2.1	1.7	1.1	ND	0.5	2.2	2.0	2.1	1.0	1.1	2.0	2.0	1.8	0.5	1.1
16	5.9	11.6	2.0	1.2	2.2	ND	1.6	1.1	0.5	1.5	2.1	1.6	ND	1.1	1.9	1.7	2.0	ND	1.2
17	7.6	9.8	1.2	1.8	1.8	ND	2.1	ND	0.6	1.7	1.5	1.4	ND	1.1	1.4	1.6	2.2	ND	1.4
19	2.9	10.6	2.3	1.8	2.5	ND	2.4	ND	1.3	1.6	1.9	1.6	ND	1.1	1.4	2.0	2.0	0.5	1.5
20	7.6	5.2	2.0	2.1	2.1	ND	2.0	ND	1.2	1.6	2.1	1.5	ND	1.1	1.6	2.1	1.9	ND	1.3
21	6.1	13.9	2.9	1.0	3.4	ND	2.9	ND	1.5	2.1	1.9	2.0	ND	ND	1.6	2.1	2.1	0.9	1.6
22	3.5	13.6	2.3	2.5	2.9	ND	1.1	ND	ND	1.1	2.9	1.9	ND	1.3	2.2	1.7	2.1	ND	1.9
23	3.7	14.0	3.0	1.1	3.1	1.4	1.2	ND	ND	1.9	1.9	2.1	ND	1.1	1.9	2.0	2.2	0.7	1.4
25	4.1	13.8	2.1	2.5	2.5	ND	1.5	ND	ND	1.1	3.1	1.7	ND	ND	1.5	1.8	2.1	ND	2.0
27	3.1	14.3	2.0	2.5	3.0	ND	2.7	ND	1.3	1.2	1.9	1.8	ND	ND	1.6	1.7	2.2	ND	2.1
29	3.4	13.9	2.0	1.6	2.7	ND	2.4	ND	1.6	1.0	1.8	2.1	ND	ND	1.6	2.0	2.1	0.6	2.1
30	6.3	13.3	2.1	1.1	2.1	0.8	1.5	1.0	ND	2.0	2.0	1.5	0.6	ND	1.2	2.0	2.1	1.0	1.3
31	3.1	8.9	2.0	0.6	2.5	1.5	1.1	ND	ND	1.6	1.6	1.7	1.0	ND	1.6	2.1	1.7	ND	1.1
32	6.6	12.1	3.0	2.9	3.5	ND	1.7	1.0	1.2	1.0	3.7	1.6	ND	ND	2.0	1.9	2.3	1.1	1.5
LSD	0.2	0.2	0.2	0.2	0.2	0.2	0.2	0.2	0.2	0.2	0.2	0.2	0.2	0.2	0.2	0.3	0.2	0.2	0.2

Note: Means within a column that differ by LSD are different (p < 0.05). All attributes were scored using a 0‐to‐15‐point universal intensity scale (Meilgaard et al. 2006).

Abbreviations: LSD, least significant difference; ND, not detected.

3.2. Consumer Testing

Of the 157 consumers who participated in the test, 33% were male and 65% were female. The majority of consumers (79%) were 18–44 years, and 21% were 45–64 years. Nearly all consumers (97.5%) reported that they consumed dairy products (milk, cheese, yogurt) at least occasionally. More than 90% of consumers recruited for this study reported that they consumed oat milk at least once a month. These consumers also reported that flavor (83.4%) followed by nutrition (75.2%) was the most influencing factors when choosing a PBMA.

Similar to the trained panel, consumers also differentiated oat milks (Figure 4). Consumer overall liking scores varied from 4.1 to 7.1 on the 9‐point hedonic scale (Figure 4, Table 4). Oat milks 27 and 29 scored the highest score in overall liking along with flavor, sweetness, texture/mouthfeel, thickness, and appearance liking. Oat milks 6 and 31 received the lowest scores for these attributes. Consistent with liking scores, oat milks 29 and 27 received high percentages of Just‐About‐Right scores for all JAR attributes (color, flavor, sweetness, thickness) with no penalties (Table 4). Samples 6 and 31, the lowest scored oat milks, were penalized for flavor JAR in addition to sweetness, thickness, and color. Seven of the 10 oat milks received a significant penalty for the Flavor JAR question (Table 4), which confirms the dominance of flavor to overall liking. Previous studies have confirmed that flavor/taste are the drivers of liking for PBMAs (Cardello et al. 2022; Jaeger et al. ²⁰²⁴; Lawrence et al. ²⁰¹⁶; Moss et al. ²⁰²²), although this is generally true for most food products.

FIGURE 4.

Overall liking scores of oat milks for all consumers (n = 157) and the two identified consumer clusters: Cluster 1 (n = 64); Cluster 2 (n = 93). Overall liking was scored on a 9‐point hedonic scale where 1 = dislike extremely and 9 = like extremely. Means labeled with different letters along each plot line signify significant differences (p < 0.05).

TABLE 4.

Consumer acceptance scores for oat milks.

		Sample 1	Sample 2	Sample 3	Sample 6	Sample 9	Sample 12	Sample 23	Sample 27	Sample 29	Sample 31
Overall appearance liking		5.5d	7.1ab	6.9bc	6.6c	5.3d	6.7c	6.7c	7.3a	6.9bc	4.7e
Color JAR	Too light	14.0%ef	5.1%efg	16.6%e	3.2%efg	5.1%efg	2.5%fg	8.3%efg	10.8%efg	1.3%g	9.6%efg
	Just about right	45.9%f	88.5%e	79.6%e	76.4%e	29.3%fg	79.0%e	77.7%e	84.7%e	85.4%e	21.0%g
	Too dark	40.1%d	6.4%efg	3.8%g	20.4%de	65.6%c	18.5%ef	14.0%efg	4.5%fg	13.4%efg	69.4%c
Overall liking		5.7de	6.2bc	5.1f	4.1g	5.5ef	6.3b	5.9cd	7.1a	7.1a	4.1g
Overall flavor liking		5.6cd	6.0bc	4.9e	3.9f	5.5d	6.4b	5.8cd	7.3a	7.2a	4.1f
Flavor JAR	Not enough flavor	35.7%ef	16.6%fg	53.5%e	28.7%f	32.5%ef	18.5%fg	21.0%fg	9.6%g	7.6%g	28.7%f
	Just about right	53.5%ef	58.0%e	34.4%fg	26.8%g	51.6%ef	65.6%de	52.9%ef	79.0%d	79.6%d	31.2%gf
	Too much flavor	10.8%g	25.5%fg	12.1%g	44.6%f	15.9%g	15.9%g	26.1%fg	11.5%g	12.7%g	40.1%f
Sweetness liking		5.6cd	6.0bc	4.6ef	4.2f	5.5d	6.1b	5.8bcd	6.9a	7.0a	4.6e
Sweetness JAR	Not sweet enough	42.7%ef	36.9%f	72.0%d	63.7%de	42.7%ef	35.7%f	31.8%f	7.6%g	10.2%g	49.0%ef
	Just about right	54.8%def	61.1%cde	28.0%g	35.0%fg	53.5%def	61.8%cde	62.4%cde	79.6%c	75.2%cd	43.9%efg
	Too much sweet	2.5%efg	1.9%fg	0.0%g	1.3%fg	3.8%efg	2.5%efg	5.7%efg	12.7%ef	14.6%e	7.0%efg
Texture/Mouthfeel liking		5.8cd	6.6b	5.9c	5.5d	5.7cd	6.8b	6.7b	7.4a	7.2a	4.9e
Thickness liking		5.4d	6.6c	5.5d	5.4d	5.5d	6.7bc	6.5c	7.2a	7.1ab	4.5e
Thickness JAR	Too thin	48.4%bc	7.6%a	37.6%b	33.8%b	46.5%bc	12.7%a	14.0%a	6.4%a	11.5%a	62.4%c
	Just about right	47.1%ab	79.6%de	57.3%bc	63.7%bcd	51.6%ab	84.1%e	77.7%cde	87.3%e	82.2%de	34.4%a
	Too thick	4.5%a	12.7%a	5.1%a	2.5%a	1.9%a	3.2%a	8.3%a	6.4%a	6.4%a	3.2%a
Aftertaste	Yes	38.2%bc	58.0%ab	33.1%c	62.4%a	44.6%abc	47.1%abc	49.0%abc	40.8%abc	42.7%abc	54.8%abc
	No	61.8%ab	42.0%bc	66.9%a	37.6%c	55.4%abc	52.9%abc	51.0%abc	59.2%abc	57.3%abc	45.2%abc
Aftertaste pleasantness		3.1bc	3.0bc	2.8cd	2.1f	2.6de	3.2b	2.9bcd	3.9a	3.7a	2.4ef
Quality		3.0d	3.6b	2.9d	2.6e	2.9d	3.6b	3.3c	4.0a	4.0a	2.5e
Expectations		2.9cd	3.2b	2.6e	2.1f	2.8de	3.3b	3.1bc	3.9a	3.9a	2.1f
Purchase intent		2.9cd	3.2b	2.5e	2.0f	2.7de	3.4b	3.1bc	4.0a	3.9a	2.0f

Note: Data represent 157 consumers. Different letters in rows following means signify significant differences (p < 0.05). Liking attributes were scored on a 9‐point hedonic scale where 1 = dislike extremely and 9 = like extremely. Statistical analysis: ANOVA with Fisher's LSD. JAR questions were scored on a 5‐point scale where 1 or 2 = too light/not enough/too thin, 3 = just about right, and 4 or 5 = too dark/too much/too thick. Statistical analysis: K‐proportions with Marascuilo procedure. Numbers in bold indicate significant penalties via penalty analysis (p < 0.05). Aftertaste question shows the percentage of people who selected “yes” or “no,” respectively. Statistical analysis: K‐proportions with Marascuilo procedure. Aftertaste pleasantness question was only shown to those who selected “yes” for the aftertaste question. It was scored on a 5‐point scale where 1 or 2 = unpleasant, 3 = neutral, and 4 or 5 = pleasant. Statistical analysis: Kruskal–Wallis. Quality was scored on a 5‐point scale where 1 or 2 = low quality, 3 = neither high nor low quality, and 4 or 5 = high quality. Statistical analysis: Kruskal–Wallis. Expectations were scored on a 5‐point scale where 1 or 2 = worse than expected, 3 = about the same as expected, and 4 or 5 = more than expected. Statistical analysis: Kruskal–Wallis. Purchase intent was scored on a 5‐point scale where 1 or 2 = would not buy, 3 = may or may not buy, and 4 or 5 = would buy. Statistical analysis: Kruskal–Wallis.

The overall liking trend was consistent among the total population (n = 157) and the two identified consumer clusters (Figure 4). Cluster 1 (n = 64) and Cluster 2 (n = 93) were primarily differentiated by their degree/intensity of liking for oat milk. Cluster 1 (n = 64) consumers generally scored oat milks higher in liking than Cluster 2 consumers (n = 93). No differences were found in usage occasion of oat milk or demographics among the two clusters (p > 0.05). The current study does not have sufficient data to conclude that there are meaningful differences between the two identified consumer clusters, beyond their overall degree of liking or disliking as reflected in scale usage. Although the clusters were examined for potential differences in drivers of liking and penalty lift, their patterns did not differ meaningfully from each other or from the total consumer population. Consequently, these clusters are not used in subsequent discussions. This absence of distinct preference‐based clusters aligns with findings by Rovai et al. (2025), which reported that oat milk consumers were generally homogenous in their desired product attributes, differing only in their willingness to pay for these attributes.

Other liking attributes and JAR scores were generally consistent with overall liking scores (Table 4). Overall flavor liking (r = 0.997), flavor JAR (r = 0.976), sweetness liking (r = 0.974), sweetness JAR (r = 0.883), texture/mouthfeel liking (r = 0.948), thickness liking (r = 0.915), thickness JAR (r = 0.774), quality (r = 0.968), expectations (r = 0.994), and purchase intent (r = 0.992) were correlated with overall liking (p < 0.05), whereas appearance liking (r = 0.592) and color JAR (r = 0.544) were not correlated with overall liking (p > 0.05). This suggests that oat milks with higher overall liking scores tend to perform well across most attributes. Appearance and color may be exceptions, but good flavor and texture can overcome shortcomings in these visual attributes.

CA biplot provides a visual representation of CATA attributes selected by consumers applicable to each oat milk (Figure 5). Consumers differentiated the oat milks by CATA attributes. Oat milks 29 and 27, which were the most liked, were characterized by vanilla flavor, sweet taste, oatmeal cookie and cereal flavors, and tastes like real (dairy) milk along with thick, creamy, and smooth texture (Figure 5). Oat milks 2, 12, and 23 were also liked and were also characterized by these attributes to a lesser degree. Oat milk 31 was the most disliked followed by oat milk 6, and these oat milks were characterized by consumers by salty taste, watery texture, bitter taste, beany and cardboard flavors, astringency, oily mouthcoating, and chalky/gritty texture.

FIGURE 5.

Correspondence analysis (CA) biplot of check‐all‐that‐apply (CATA) attributes selected by consumers (N = 157) applicable to oat milks.

Penalty lift analysis, which is based on CATA data, was consistent with consumer overall liking. Sweet taste, smooth and creamy texture, mild flavor, and cereal flavor had positive influences on liking, whereas watery texture was disliked (results not shown). Consumers also selected usage occasion for each oat milk. CA of CATA data with usage occasion showed that mild flavor and creamy texture were associated with usage with cereal, in a smoothie, or in a recipe, whereas sweet taste, smooth and thick texture, vanilla and oatmeal cookie flavors/cereal flavor, and tastes like real (dairy) milk were associated with adding to coffee/tea and drinking on its own. Consumers indicated they would not use oat milks perceived as cardboard, bitter, salty, astringent, watery, beany, or having a chalky/gritty texture (results not shown).

Previous studies with PBMAs in general showed that consumers preferred a white color (Cardello et al. 2022; Moss et al. ²⁰²²), whereas others have suggested a preference for cream/brown color (Jaeger et al. 2024). The current study did not observe a correlation with appearance liking and overall liking although color intensity did vary among the milks by descriptive analysis (mean color intensity range of 2.0–7.6 on a 0‐to‐15‐point universal intensity scale), and consumers also documented differences in appearance liking of oat milks (Table 4). Oat milks 1, 9, and 31 received the lowest appearance liking scores. Oat milks 9 and 31 were penalized for being too dark in color. Oat milks 3 and 6 received high appearance liking score but were the least liked oat milks. Additionally, consumers differentiated oat milks on the basis of texture (Figure 5). Opacity was a driver of liking for oat milks in the current study and was correlated (r = 0.596, p < 0.05) with viscosity (not thin/watery texture) which is also a driver of liking. Consumers associated oat milks with residual mouthfeel as thick, creamy, and smooth in texture (Figures 5 and 6) and assigned higher scores for texture/mouthfeel and thickness liking for these oat milks (Table 4). Oat milks with higher overall liking scores also had a high percentage of JAR scores for thickness.

FIGURE 6.

External preference map of all consumers (n = 157). Trained panel attributes are vectors overlaid on consumer liking scores. Each color represents the percentage probability of consumers who liked the attributes and sample. Warm colors (red and orange) represent the most liked attributes and samples, whereas cool colors (blue) represent the least liked attributes and samples.

External preference mapping was conducted using consumer liking scores and trained panel data to further characterize liked and disliked attributes of oat milks (Figure 6). Trained panel attributes are vectors overlaid on consumer liking scores, which are represented by the colored background. Each color represents the percentage probability of consumers who liked the sample(s) and corresponding attribute(s). Warm colors (red and orange) represent the most liked attributes and samples, whereas cool colors (blue) represent the least liked attributes and samples. Consistent with consumer CATA and liking attributes, sweet aromatic flavor and sweet taste drove liking for all consumers, whereas raw oat flavor and bitter taste were disliked. Cooked/cereal, opacity, and viscosity were also liked attributes, whereas cardboard, astringency, and chalky were also disliked attributes. Both external preference mapping (Figure 6) and CA biplot (Figure 5) identified residual mouthfeel, smooth, and creamy texture as desirable oat milk attributes and watery or thin texture as undesirable. Other studies with PBMAs in general have also documented these desirable and undesirable attributes (Cardello et al. 2022; Jaeger et al. ²⁰²⁴; Moss et al. ²⁰²²).

Many of these liked and disliked sensory attributes have been documented with other PBMAs. Sweet aromatic flavor and higher viscosity were drivers of consumer liking for soy milk (Lawrence et al. 2016). Many studies have established, perhaps not surprisingly, that sweet taste is a driver of liking for PB milks in general (Cardello et al. 2022; Jaeger et al. ²⁰²⁴; Lawrence et al. ²⁰¹⁶; Moss et al. ²⁰²²). Oat milks that were low in sugar (Table 1) were rated poorly on both sweetness and overall liking (oat milks 3, 6, and 9) (Table 4). Interestingly, consumers conceptually reported that they did not want sweeteners added to oat milk but concurrently reported that desirable flavor drove repeat purchase (Rovai et al. 2025).

Cereal flavor was identified in the current study as a positive attribute for oat milk (Figure 6). According to Cardello et al. (2022), consumers who were PB likers exhibited a more positive attitude towards PB flavors like cereal/oaty flavor, nutty flavor, and grain/wheat flavor. Jaeger et al. (2024) also showed similar results with cereal/oaty flavor, nutty, roasted nuts, and coconut flavor identified as positive drivers of liking for PBMAs. Lawrence et al. (2016) documented cereal flavor as a disliked attribute for soymilks. Lawrence et al. (2016) focused exclusively on soymilk, and the focus on soymilks may explain why cereal flavor was a disliked attribute. In the context of soymilks, cereal flavor may be a disliked attribute. Bitter taste, cardboard flavor, umami taste, chalky texture, and astringent mouthfeel have been identified as general drivers of dislike for PBMAs (Cardello et al. 2022; Jaeger et al. ²⁰²⁴; Lawrence et al. ²⁰¹⁶; Moss et al. ²⁰²²; Sethi et al. ²⁰¹⁶; Tangyu et al. ²⁰¹⁹) and were also identified as drivers of dislike for oat milks in the current study. It is perhaps not surprising that these sensory attributes appear to be universally disliked by consumers in PBMAs, perhaps because dairy milk remains the ideal reference for the consumer populations evaluated (in our study nearly all consumers also consumed dairy milk) or perhaps due to preferences/expectations for a mild flavor profile in a PB alternative beverage.

One unique contribution of this study is that consumers like cereal/cooked flavor of oat milk but not raw oat flavor (Figure 6). Some literature has suggested that nuttiness is a driver of liking for PBMAs (Cardello et al. 2022; Jaeger et al. ²⁰²⁴), but this attribute was not a highly cited attribute by consumers in CATA question (67.8% consumers reported “nuttiness” as absent in oat milks), and no such correlation was found in this study (p = 0.08). The trained panel distinguished the cereal/cooked attribute from nutty flavor but the consumer CATA attribute “cereal” in the current study may have a nutty component to it. Future studies should explore how this attribute is characterized by consumers as there is no evidence of what nuttiness means to the consumers.

It is intriguing to observe how oat milks were described by trained panelists compared to untrained consumers using the modified DA lexicon as a CATA list (Figures 5 and 6). Trained panelists scored Sample 2 high in viscosity and residual mouthfeel and Sample 31 low in these attributes (Figure 6); consumers also classified Sample 2 as having a thick and creamy texture, whereas Sample 31 was described as having a watery texture (Figure 5). Trained panelists rated Samples 29 and 27 high in sweet aromatic flavor (Figure 6), whereas consumers classified it as having vanilla flavor. Although trained panelists and consumers both identified sweet aromatic/vanilla flavors in these samples, none of them were actually “flavored” oat milks (all oat milks were labeled “original” flavor). Future work could examine the role of flavors (vanilla, chocolate), as these flavors can enhance liking of PB milks and can mask off‐flavors (Alsado et al. 2023). Another limitation is that the consumers were mostly young females who are based in Raleigh, North Carolina. It would be interesting to look at the perception of different demographics of the population. It would also be interesting to use exclusive PB milk consumers and see if there are any differences in expectations versus consumers that use both PB and dairy beverages (the majority of consumers in this study). A standard of identity for oat milk types would also help both manufacturers and consumers when labeling these products with terms such as “full fat,” “extra creamy,” “barista,” and “less sweet.” A standard of identity would ensure only specific ingredients within a certain range are used in formulations and provide consumers with a clear understanding of what to expect from the product in terms of nutrition and quality.

4. Conclusion

Sweet aromatic and cooked cereal flavors, sweet taste, and a smooth and creamy texture are the universal drivers of liking for oat milk. Watery/thin texture, bitter taste, chalky texture, and cardboard flavor are attributes that drive dislike in oat milk. The intrinsic attributes that drive the liking of oat milk can be utilized by processors to tailor a high‐quality and satisfactory oat milk.

Author Contributions

S. Gupta: investigation, writing – original draft, formal analysis, writing – review and editing. D. Rovai: investigation, formal analysis, data curation, supervision, writing – review and editing. P. D. Gerard: formal analysis, visualization, writing – review and editing. M. A. Drake: conceptualization, methodology, funding acquisition, writing – review and editing, project administration, supervision, formal analysis.

Conflicts of Interest

The authors declare no conflicts of interest.