Consumers’ Preferences for Digital Corporate Content on Company Websites: A Best–Worst Scaling Analysis

Abstract

Digital content marketing (DCM) complements traditional marketing communication approaches and is a major focus of research. Uses and gratifications research posits that DCM only unfolds positive effects if it provides valuable content to consumers. However, there is limited evidence on what constitutes gratifying digital corporate content on company websites. This study aimed to elicit consumers’ preferences for key characteristics of digital corporate content on company websites and whether preferences differ among consumer subgroups. Best–worst scaling (BWS) was used to reveal preferences. To obtain BWS data, a cross-sectional survey was employed. The study sample comprised 1527 consumers from Germany, Switzerland, and Austria. Data were analyzed using counting analysis and conditional logit modeling. Subgroup comparisons were performed with t-tests and one-way ANOVA. The results consistently show that consumers prioritize information value as the most important content characteristic, followed by value in use, entertainment value, process value, and social value. Subgroup comparisons revealed generally similar priorities among consumers, with the greatest heterogeneity being found in assessments of the importance of social value. The study also suggests that consumers prioritize digital corporate content characteristics on company websites differently than they do on social media. These findings contribute to the evolving literature on DCM and provide insights that could help set evidence-based priorities in DCM practice.

1. Introduction

In recent years, content marketing has grown in importance and now complements traditional marketing communication approaches [1]. It refers to creating and sharing brand-related content with customers and other target groups to achieve business objectives [1,2,3,4]. Unlike traditional advertising, which is usually geared towards persuading or even pushing the target group to take action [5], content marketing intends to add value by, e.g., informing, helping to solve problems, or entertaining [4]. Content marketing generally employs both printed corporate media and digital platforms [1]. The latter play a central role in an environment of ubiquitous digital interactions between consumers and companies, making digital content marketing a major focus of research [2].

An important research domain is the effectiveness of digital content marketing, i.e., its capacity to trigger cognitive, emotional, and behavioral consumer engagement in the customer journey and foster a positive brand attitude [2]. In this research context, regular reference is made to uses and gratifications research, which assumes that digital content marketing can only unfold positive effects if it provides the target group with content that is perceived as valuable [4]. Recent empirical research on content marketing has revealed that consumers consider digital corporate content to be gratifying, for example, if it provides informative, entertaining, functional, or social value [6]; causes enjoyment [7]; or is rewarding because it offers access to promotions or similar benefits [8]. However, it has to be acknowledged regarding these findings that current studies primarily focus on social media content marketing, while corporate content on other digital platforms, especially branded content on company websites, has been studied only to a limited extent [9]. The broader literature on company websites outside the content marketing domain also offers only limited evidence on what constitutes gratifying digital corporate content from a consumer perspective. Extant research in this area has examined how consumers perceive and respond to company websites depending on website-related factors, such as usability, interactivity, design, and content [10,11,12,13], but without specifically addressing the hybrid journalistic-persuasive content typical of content marketing [4]. This is noteworthy since there is considerable consensus in the literature that branded content on company websites is a core pillar of digital content marketing [2] and because industry studies have found that the associated investments account for a substantial share of total content marketing investments [14]. Since the relative importance of digital content characteristics is assumed to be platform-dependent [15], it seems likely that consumers show different preferences for various characteristics of digital corporate content on company websites compared with social media. Thus, findings on gratification effects of digital corporate content presumably cannot be transferred from social media content marketing to content marketing on company websites. There is a need to investigate the preferences for digital corporate content context specifically [2,16].

It can thus be stated that there is limited evidence on what constitutes valuable, i.e., gratifying, digital corporate content on company websites from a consumer perspective. We address this knowledge gap in this paper. First, we aim to elicit consumers’ preferences for key characteristics of digital corporate content on company websites using best–worst scaling (BWS). BWS is an innovative stated preference method [17] that has been shown to provide more accurate predictions than rating or ranking approaches [18] and to be particularly helpful to validate priorities from a consumer perspective [19]. Second, we also aim to address how these preferences might differ across subgroups of consumers. A better understanding of the relative importance of digital corporate content characteristics and related differences among groups of consumers could help marketers improve their websites and content marketing initiatives and could contribute to the content marketing literature.

2. Literature Review

2.1. Gratifying Characteristics of Digital Corporate Content

Uses and gratifications theory [20,21] provides a theoretical basis to elicit consumers’ preferences for digital corporate content on company websites since it can help to unveil and structure potential key characteristics of digital corporate content that consumers may perceive as gratifying and valuable and that may be preferred to varying degrees. Uses and gratifications theory is a seminal theory for the study of media and media content, encompassing a large corpus and a long history of well-documented work [20,21,22]. It has provided the basis for various studies in leading communications and marketing journals [23] and has been used in studying traditional as well as digital media [24,25]. Uses and gratifications theory posits that consumers are aware of their needs and preferences and select media content in response to them and that the use of content leads to specific gratifications [21,26].

Recently, Hollebeek and Macky [2] developed a digital content marketing framework guided by uses and gratifications theory. Based on prior research in the uses and gratifications domain, this framework proposes specific consumer needs and motives for interacting with digital content. Hollebeck and Macky [2] contended that consumers’ decisions to interact with digital corporate content are driven by a triad of functional, hedonic, and authenticity motives. The functional motive denotes a consumer’s utilitarian desire for brand-related information or learning through digital corporate content [22,27]. The hedonic motive refers to consumers’ emotional desire for brand-related entertainment, diversion, fun, transportation, or relaxation through digital corporate content [22,28]. The authenticity motive, in turn, relates to consumers’ desire for content that conveys aspects such as brand-related continuity, integrity, or credibility [22,29].

However, this classification of consumer motives for interacting with digital corporate content and the related gratifications may not sufficiently capture the gratifications sought with digital corporate content on company websites. This becomes evident when comparing the triad with other media gratification typologies developed in the past decades in uses and gratifications research. Based on a review of 20 uses and gratifications studies that contained gratification typologies for major media including digital media, Sundar and Limperos [24] determined a canon of certain categories of gratifications. The information category they identified corresponds to Hollebeek and Macky’s [2] information and learning dimension, as the escape, entertainment, and emotion categories they identified correspond to Hollebeek and Macky’s [2] hedonic gratification dimension. However, while Hollebeek and Macky [2] limit the social gratification dimension of digital content to the aspect of authentic interactions with a brand, the analysis by Sundar and Limperos [24] suggests that this dimension should rather be defined more broadly by including further aspects, such as the value derived from interacting with others about content, from sharing content, or from content that enhances social status. Beyond that, the work by Sundar and Limperos [24] points to an additional potentially gratifying characteristic of digital corporate content in terms of process gratifications, which can result from the process of using digital corporate content, such as convenience, technological pleasure, interactivity, or multimedia experiences.

Since Sundar and Limperos’ [24] review was not specific to digital content, it is mandatory to also consider potentially gratifying content characteristics that may have emerged from previous research specifically related to digital content. In this domain, scholars have identified many gratifications related to different digital platforms [7,8,15,30,31,32,33,34] but have tended to refrain from elaborating a theory-based classification of gratifications. Lou and Xie [6], however, recently presented four gratifying value dimensions of digital corporate content based on advertising value theory [35] and the theory of consumption values [36]. In this classification, the informative value refers the digital content’s utility of providing new, timely, and valuable brand-related information [6]. The entertainment value describes the capacity of digital corporate content to meet the entertainment and emotional needs of consumers [6,36], such as diversion, relaxation, or escape [37,38]. Further, the social value of content pertains to gratifications that stem from its capacity to gain social benefits, such as gaining popularity among others, interacting with others about specific content, sharing content, connecting with others, or self-expression [6,24]. Finally, the functional value relates to the perceived benefit that consumers obtain from corporate content in terms of its utilitarian performance [36]. This means how well it fulfills functional needs in a reliable way [6], such as “How To Content” that helps resolve problems. The dimensions of informative value, entertainment value, and social value resemble to a large extent the gratification categories mentioned above. Beyond that, however, Lou and Xie’s [6] research points to another potentially important gratification category in the form of the functional or utility value of digital content.

In summary, considering the different theoretical perspectives mentioned above, the literature thus points to five gratifying characteristics of digital corporate content where consumer preferences might differ with respect to company websites:

• Information value refers to digital corporate content that is perceived to be interesting, relevant, up-to-date, and of high quality.
• Entertainment value refers to digital corporate content that is entertaining, surprising, and relaxing and puts consumers in a good mood.
• Value in use refers to digital corporate content that offers functional, utilitarian performance in terms of inspiration, ideas, advice, tips, or help in everyday life or purchasing decisions.
• Social value stems from interacting with others about specific content, sharing content, or content that serves to better present or understand oneself.
• Process value stems from a gratifying process of using digital corporate content and may be related to, e.g., multimedia experiences, ease of use, or specific technologies that users perceive as pleasurable to use.

2.2. Best–Worst Scaling

This study employs case 1 (object case) BWS to determine consumers’ preferences regarding the identified digital corporate content characteristics on company websites. BWS is grounded in random utility theory assuming that a consumer’s relative preference for an object A over object B depends on the relative frequency at which A is preferred to B [17]. BWS thus aims to infer the strength of preference for certain objects by asking individuals to make choices as in real life. In case 1 (object case) BWS tasks, study participants respond to a series of choice tasks. Each choice task presents a subset of objects from all objects of interest and asks participants to choose the best or most important and worst or least important object on an underlying subjective scale of interest [17]. Simultaneously asking study participants for the most and least important object in each choice set yields greater information about the overall ranking of the objects than if only the first choice were collected. An analysis of the responses to the presented choice tasks permits inferring the underlying relative importance of the investigated objects and establishing a ranking of the objects [39].

3. Materials and Methods

3.1. Construction of Choice Sets

Prior to creating the BWS choice sets, the five characteristics of digital corporate content derived from the literature were once more reviewed for their relevance to company websites, and the corresponding descriptions were also checked to ensure their clarity for the study participants. Furthermore, it was re-examined whether the five content characteristics adequately covered all potentially salient aspects. This was accomplished in an expert think tank that gathered two representatives from academia, two senior content marketers, and one senior representative from an agency specialized in this field (N = 5). The think tank was run digitally and combined presentations and structured discussion to facilitate agreement. Unanimous agreement was required for characteristics to be included in the BWS. In this process, the literature-driven selection of content characteristics to be studied was reconfirmed, no additional categories were added, and only minor stylistic refinements were made to the descriptions.

Next, to construct the BWS choice sets for the five characteristics of digital corporate content, a balanced incomplete block design (BIBD) was generated using RStudio (2021.09.0 + 351), R (4.1.2), and the bwsTools package (1.2.0). A BIBD ensured that all choice sets had the same number of items, that each characteristic occurred equally often across all choice sets, and that pairwise comparisons among all characteristics appeared the same number of times, thus preventing unintended signals from being given to study participants (e.g., inference to a putative study objective) [17]. A design was chosen in which the v = 5 content characteristics of interest were shown in a series of b = 5 choice tasks, each with k = 4 content characteristics. Each content characteristic was presented a total of r = 4 times across all choice tasks and coappeared with each other content characteristic λ = 3 times. The position of the content characteristics within the choice sets and the order of choice sets were randomized to mitigate any order effects. Each study participant responded to all five choice tasks. Figure 1 displays an example of a BWS choice set.

3.2. Survey Structure and Implementation

For obtaining data, we relied on a structured questionnaire comprising four sections. First, the potential participants were provided with introductory explanations regarding digital corporate content on company websites to ensure a uniform understanding of this study’s topic. As part of this, the participants were also presented with six practical cases to illustrate the subject of the study. The cases covered various sectors and were drawn from the pool of winners of the Best of Content Marketing Award, the largest competition for content-driven communication in the German-speaking area. The cases presented included the company websites or parts of the websites of BMW.com (automotive), redbull.com (soft drinks), euronics.de (consumer electronics retail), migros.ch (food retail), tui.com (tourism), and union-investment.de (capital investments).

In the second section, potential participants were asked screening questions to verify the inclusion criteria and allow the stratification of the sample. All questions were formulated in accordance with established national surveys. The third and main section consisted of the BWS choice sets. In the final section, we collected background information on personality traits and lifestyle of the study participants to be able to conduct additional subgroup analyses on consumers’ preferences. Personality traits and lifestyle were included because they were frequently found to be associated with digital content use in previous studies [40,41,42,43]. The questionnaire was implemented online using EFS Survey software (version Fall 2022). Prior to fielding the survey, a pre-test was realized. Data were collected in December 2022.

3.3. Participants

Consumers between 16 and 65 who lived in Germany, Austria, or Switzerland and generally used digital corporate content were eligible to participate in the study. We targeted the general population to obtain results that are reasonably generalizable. Participants were recruited with Gapfish, the largest ISO-certified online access panel in the German-speaking area, to facilitate data collection. Online access panels are widely used in research, and data from online panel sampling have been shown to be of adequate quality provided that certain recommendations are followed [44,45]. Thus, to mitigate related concerns, potential study participants were asked screening questions on residence, age, gender, education, employment, and use of digital corporate content on company websites to verify the inclusion criteria. Additionally, attentiveness checks were applied. To alleviate potential nonresponse bias, reminders were sent to encourage completion of the survey. We also controlled the socioeconomic composition of the sample using quotas and offered a fair but not excessively appealing incentive for participation to minimize economic self-selection bias [44]. To obtain a sample that corresponds with the target population, non-interlocking quotas were applied to the screening criteria. We used mandatory answering to prevent missing data [46].

3.4. Study Size

There is a dearth of formal guidelines to determine the target sample size for BWS studies [17,19], which frequently leaves the choice of sample size unjustified [47]. To derive the minimum sample size for our study, we considered three perspectives.

First, as a heuristic, we recurred to the sample size commonly chosen in BWS studies. According to a recent review, the average sample size of BWS studies published between 2017 and 2021 was 472 respondents [47].

Second, we adhered to recommendations to use sample size methods developed for multinomial proportions data, since BWS measures are derived from multinomial frequency counts or proportions [17]. Following Thompson’s [48] method and assuming that the probability should be at least 1 − $\alpha$ = 0.95 that all estimated proportions are within a distance of d = 0.05 of true population proportions, the required sample size would have had to be 510 respondents, irrespective of the number of choice options.

Third, we considered what minimum sample size was required to obtain sufficient statistical power to detect differences in preferences among various subgroups of consumers. Since unpaired t-tests for comparisons between two groups and one-way ANOVA for comparisons among up to five groups of consumers were planned, corresponding a priori power analyses were performed with G*Power (version 3.1). For the t-tests, a priori power analysis with error probability $\alpha$ = 0.05 indicated a required total sample size of 788 participants (394 per group) to detect small size effect d = 0.20 with target power 1 − $\beta$ = 0.80. For the ANOVA with five consumer groups, a priori power analysis indicated that a total sample size of 1200 participants (i.e., 240 per group) was required to detect small size effects ${\eta }^2$ = 0.01 with target power 1 − $\beta$ = 0.80 and error probability $\alpha$ = 0.05.

Considering all three perspectives outlined above, we defined a minimum required sample size of 1200 participants for this study (which is the maximum of the values resulting from the different perspectives).

3.5. Statistical Analyses

3.5.1. Processing of the Sample

First, we verified that all respondents met the eligibility criteria and attention checks described in Section 3.3. Participants not meeting these criteria were removed from the sample. Subsequently, the structure of the study sample was described. SPSS Statistics (version 29) was used for these procedures.

3.5.2. Counting Analysis

Next, we analyzed the BWS response data with the counting approach [17,49,50]. The total number of times that each item was chosen as “most important” (best choice, B) and “least important” (worst choice, W) across all choice sets and study participants was calculated. For each content characteristic, the best–worst scores were calculated as the differences between these frequencies (BW = B − W). Standardized best–worst scores were calculated for each item by dividing the best–worst scores by the number of times that each item was shown to participants in the choice sets (r = 4), and the total number of participants in the study (standardized BW = BW/Nr). The standardized best–worst scores ranged from −1 to +1, with higher scores indicating higher perceived importance. To assess if choices were consistent across study participants, the standard deviation (SD) of standardized best–worst scores was calculated with higher values representing greater heterogeneity [51]. We generated 95% confidence intervals (CIs) for the standardized scores using bootstrapping [52] with 5000 bootstrap draws, and the bias-corrected and accelerated (BCa) method. Overlapping intervals were regarded as an indication that the difference between scores was not significant. To infer the relative importance of the content characteristics, the square root of the ratio of best counts to worst counts (sqrt BW = $\sqrt{B/W}$) was also calculated for each item [17]. To facilitate interpretation, these values were standardized with the highest square root value and scaled to a maximum of 100 (standardized sqrt BW = 100 × sqrt BW/maximum sqrt BW), so that the most important content characteristic was indexed to 100 [53].

All the above analyses, which concern the counting approach, were performed using RStudio, R, and the support.BWS package (0.4–4). The only exception was bootstrapping, which was performed using SPSS.

3.5.3. Modeling Analysis

Next, we estimated a conditional logit model to analyze the BWS responses [17,49,50]. It was assumed that respondents answered the BWS questions based on the maxdiff model [39], so a suitable dataset with dummy-coded item variables was used for analysis [49]. Since our study included five content characteristics of interest and the item variables were dummy-coded, four variables were entered into the model as independent variables, with the remaining variable serving as reference. The content characteristic that was found to be the least important in the counting analysis was chosen as the reference. The estimated model coefficients and related statistics are reported in the Results section. The coefficients indicate relative utilities, i.e., they show how important the content characteristics that entered the model as independent variables are compared with the reference characteristic, which has a coefficient of zero [49,50]. Based on the model estimates, we also calculated the shares of preference for all content characteristics to aid interpretation. The preference shares measure the extent of relative importance between the content characteristics and sum to 100 [49,54]. To evaluate the goodness-of-fit of the model, McFadden’s R-squared (adjusted rho-squared) was used. Values ≥ 0.10 were considered to be indicative of acceptable model fit, and values ≥ 0.20 were regarded to represent excellent model fit [55]. For assessing consistency between the counting and modeling approach, Pearson’s correlation between the best–worst scores and the conditional logit estimates was calculated. A correlation coefficient ≥ 0.90 was considered to indicate acceptable consistency [56].

All aforementioned analyses concerning the modeling approach were performed with RStudio, R, and the support.BWS package.

3.5.4. Subgroup Analyses

The above analyses focused on average preferences. To explore preference heterogeneity among consumers, subgroup analyses were performed based on standardized best–worst scores [49]. Differences in standardized best–worst scores among subgroups were tested with unpaired t-tests for comparisons between two groups and with one-way ANOVA and post hoc tests (Tukey’s and Games–Howell) for comparisons of more than two groups. Four analyses were performed based on gender, age, personality, and lifestyle.

Respondents’ personality was assessed using the Big Five model [41]. We used the BFI-10 scale to capture the five personality traits, as it is a highly economic scale and has shown adequate psychometric properties in prior research [57]. Based on the Big Five traits, study participants were clustered into five personality types by administering Ward’s method, followed by the k-means clustering algorithm commonly used in the literature [58,59]. To classify the clusters, the mean z-scores of the Big Five factors for each cluster were visually compared with the mean z-scores of different personality types from the literature [59]. In this way, resilient, overcontroller, undercontroller, reserved, and vulnerable–resilient personality types were obtained for the subgroup analyses [59].

Respondents’ lifestyles were categorized using the LFT typology [60,61]. The LFT structures lifestyle according to two dimensions: social position, reflected in one’s economic and cultural capital, and biographical route, referring to openness in life and modernity of values. Both dimensions were measured using the 14-item LFT lifestyle diagnostic statement scale [62]. Based on reference values [62], study participants were assigned to the five lifestyle types of Social Elite, Traditionalists, Middle Class, Avantgarde, and Underprivileged–Modernized [63], which were used for the subgroup analyses.

SPSS was used for all statistical procedures related to the subgroup analyses. Cohen’s d (for t-tests; 0.20 = small, 0.50 = medium, 0.80 = large effect) and ${\eta }^2$ (for ANOVA; 0.01 = small, 0.06 = medium, 0.14 = large effect) was used as effect size measure. A p-value of <0.05 was considered significant.

4. Results

4.1. Participants’ Data

In total, 1638 consumers participated in the survey. After removing participants that failed to meet the inclusion criteria or attentiveness checks, the sample for the analyses comprised N = 1527 consumers. Half of the sample consisted of men (51.8%), and half, of women (48.2%), with the 50–65 age group (32.5%) being the largest (

Table 1. Sociodemographic characteristics of the sample and the target population.

	Sample		Target Population
	n	%	%
Gender
Female	736	48.2	49.4
Male	791	51.8	50.6
Age (years)
16–29	369	24.2	23.7
30–39	338	22.1	20.1
40–49	324	21.2	19.0
50–65	496	32.5	37.2
Level of education
Low	334	21.9	25.7
Medium	572	37.5	35.3
High	621	40.7	39.0
Employment status
Employed (incl. actively seeking work)	1189	77.9	76.9
Education/training	127	8.3	10.3
Pension	104	6.8	5.5
Housemen/-wives	107	7.0	7.3

Note: N = 1527.

). Consumers with higher (40.7%) and medium levels of education (37.5%) were present in the sample in similar proportions, while consumers with lower levels of education constituted a smaller share (21.9%). Most participants were employed (77.9%).

The sample aligned well with the target population; only 50–65-year-old consumers and persons with lower education were slightly underrepresented [64,65,66].

4.2. Counting Analysis

The results of the data analysis with the counting approach are presented in

Table 2. Results of the counting analysis.

Content Characteristics	B	W	BW	Mean Standardized BW	SD	BCa 95% CI		Sqrt BW	Standardized Sqrt BW	Rank
						LL	UL
Information value	2972	484	2488	0.407	0.438	0.386	0.430	2.478	100.0	1
Value in use	1920	791	1129	0.185	0.409	0.165	0.205	1.558	62.9	2
Entertainment value	1217	1433	−216	−0.035	0.449	−0.058	−0.012	0.922	37.2	3
Process value	797	1691	−894	−0.146	0.420	−0.168	−0.124	0.687	27.7	4
Social value	729	3236	−2507	−0.410	0.540	−0.438	−0.383	0.475	19.2	5

Note: N = 1527; B = total count best choice (most important); W = total count worst choice (least important); BW = B − W; mean standardized BW = BW/Nr with N = 1527 and r = 4 (number of times each item was shown to participants in the choice sets); SD = standard deviation of mean standardized BW; LL/UL = lower and upper limits of the bias-corrected and accelerated bootstrap (BCa) 95% confidence intervals of mean standardized BW; sqrt BW = $\sqrt{\mathrm{B}/\mathrm{W}}$; standardized sqrt BW = 100 × sqrt BW/maximum sqrt BW.

. Overall, consumers considered information value (mean standardized BW (SD) = 0.407 (0.438), BCa 95% CI [0.386; 0.430], standardized sqrt BW = 100.0) to be the most important characteristic of digital corporate content on company websites. The value in use of digital corporate content (mean standardized BW (SD) = 0.185 (0.409), BCa 95% CI [0.165; 0.205], standardized sqrt BW = 62.9) was stated as the second most important content characteristic by consumers. The signs of the mean standardized BW scores were positive for both content characteristics, i.e., they were each selected more frequently by consumers as the most important than as the least important characteristics. Since the confidence intervals did not overlap, consumers considered information value to be statistically significantly more important than value in use. The standardized sqrt BW values for the two characteristics of 100.0 and 62.9, respectively, imply that information value was regarded as 1.6 (100.0/62.9) times as important as value in use.

Information value and value in use were followed by entertainment value (mean standardized BW (SD) = −0.035 (0.449), BCa 95% CI [−0.058; −0.012], standardized sqrt BW = 37.2) in consumers’ importance evaluation. The latter was found to be statistically significantly less important than the former characteristics given the non-overlapping confidence intervals of the mean standardized BW scores. The mean standardized BW score near zero reflects that entertainment value was selected by consumers as the most important and least important feature of digital corporate content on company websites with similar frequency. A comparison of standardized sqrt BW values signaled that information value was 2.7 (100.0/37.2) times more important to consumers than entertainment value, and that value in use was 1.7 (62.9/37.2) times more important.

Further, the value of a gratifying usage process of digital corporate content (mean standardized BW (SD) = −0.146 (0.420), BCa 95% CI [−0.168; −0.124], standardized sqrt BW = 27.7) was found to be statistically significantly less important than its informational, use, and entertainment value. The ratios of the standardized sqrt BW values indicated that particularly information value (100.0/27.7 = 3.6) and use value (62.9/27.7 = 2.3) were found to be more important than process value, whereas this was less pronounced for entertainment value (37.2/27.7 = 1.3).

Finally, the counting analysis showed that the social value of digital corporate content was considered least important to consumers (mean standardized BW (SD) = −0.410 (0.540), BCa 95% CI [−0.438; −0.383], standardized sqrt BW = 19.2). The analysis also revealed that the variance in social value (SD = 0.540) was larger than that in the other content characteristics (SD 0.409 to 0.449). Thus, the importance of social value differed to a greater degree among consumers than the importance of the other content aspects, indicating greater heterogeneity.

4.3. Conditional Logit Analysis

Table 3. Results of the conditional logit model.

Content Characteristics	β	SE	p	SP	Rank
Information value	1.480	0.027	<0.001	36.9	1
Value in use	1.082	0.026	<0.001	24.8	2
Entertainment value	0.677	0.025	<0.001	16.5	3
Process value	0.473	0.025	<0.001	13.5	4
Social value	reference			8.4	5

Note: N = 1527; β = coefficient/relative preference compared with reference; SE = standard error; SP = share of preferences; $R_{McFadden}^2=$ 0.11; $R_{McFadde{n}_{adj}}^2=$ 0.11; correlation between best–worst scores of the counting analysis and conditional logit estimates: r = 0.99, p < 0.001.

reports the results for the conditional logit model. Since “social value” emerged as the least important content characteristic in counting analysis, it was omitted from the model and utilized as the reference.

Consistency between the results of counting analysis and modeling was high, with a strong positive correlation between standardized best–worst scores and estimated model coefficients, as well as consistent rank orders of importance. The adjusted McFadden’s R-squared value of 0.11 indicated an acceptable overall model fit, while also implying some heterogeneity in consumer preferences.

All variables had statistically significant and positive coefficients, indicating that the four content characteristics included as independent variables in the model were more important to consumers than the social value of digital corporate content, which was the base item having a coefficient of zero. Among them, information value (ß = 1.480 (0.027), p < 0.001, SP = 36.9) was found to be the most important characteristic of digital corporate content on company websites, followed by value in use (ß = 1.082 (0.026), p < 0.001, SP = 24.8), entertainment value (ß = 0.677 (0.025), p < 0.001, SP = 16.5), and process value (ß = 0.473 (0.025), p < 0.001, SP = 13.5).

4.4. Subgroup Analyses

The results of the subgroup analyses based on mean standardized best–worst scores are presented in

Table 4. Results of the subgroup analyses.

		Information Value		Value in Use		Entertainment Value		Process Value		Social Value		N
		M	SD	M	SD	M	SD	M	SD	M	SD
Gender	female	0.373	0.413	0.233	0.427	−0.010	0.458	−0.191	0.442	−0.405	0.538	736
	male	0.440	0.458	0.140	0.386	−0.059	0.439	−0.105	0.393	−0.416	0.543	791
	t-value, p, d	−3.00, <0.01, −0.153		4.42, <0.001, 0.227		2.15, <0.05, 0.110		−3.97, <0.001, −0.204		0.39, 0.699
Age (years)	16–29	0.341 a	0.445	0.124 1	0.406	−0.018 1	0.442	−0.152	0.450	−0.295 a	0.552	369
	30–39	0.345 a	0.467	0.106 1	0.395	0.008 1	0.461	−0.150	0.416	−0.309 a	0.552	338
	40–49	0.418 a,b	0.429	0.184 1	0.411	0.009 1	0.438	−0.172	0.421	−0.440 b	0.528	324
	50–65	0.491 b	0.405	0.284 2	0.400	−0.107 2	0.445	−0.122	0.398	−0.546 c	0.498	496
	F-value, p, ${\eta }^2$	11.74, <0.001, 0.022		17.25, <0.001, 0.033		6.64, <0.001, 0.013		1.02, 0.381		21.47, <0.001, 0.040
Personality type	Resilient	0.449 1	0.437	0.248 a	0.391	−0.041 1,2	0.481	−0.202 1	0.424	−0.453 1	0.537	308
	Overcontroller	0.427 1	0.449	0.211 a	0.415	−0.004 1,2	0.440	−0.106 1,2	0.402	−0.528 1	0.524	186
	Undercontroller	0.304 2	0.435	0.088 b	0.388	0.004 1	0.433	−0.132 1,2	0.423	−0.263 2	0.536	489
	Reserved	0.498 1	0.426	0.250 a	0.383	−0.105 2	0.424	−0.107 2	0.423	−0.537 1	0.523	287
	Vul.-resilient	0.439 1	0.417	0.202 a	0.458	−0.048 1,2	0.464	−0.180 1,2	0.410	−0.413 1	0.522	257
	F-value, p, ${\eta }^2$	11.33, <0.001, 0.029		11.93, <0.001, 0.028		2.93, <0.05, 0.008		2.98, <0.05, 0.008		16.29, <0.001, 0.041
Lifestyle	Social Elite	0.388 1	0.447	0.138 1	0.402	−0.047 1,2	0.429	−0.135	0.408	−0.344 a	0.536	529
	Traditionalists	0.498 2	0.412	0.336 2	0.390	−0.097 1	0.458	−0.094	0.364	−0.643 b	0.451	152
	Middle Class	0.431 1,2	0.430	0.202 1	0.418	−0.048 1,2	0.472	−0.155	0.442	−0.429 a	0.537	475
	Avantgarde	0.377 1,2	0.426	0.167 1	0.412	−0.002 1,2	0.438	−0.168	0.425	−0.373 a	0.554	241
	Und.-Modernized	0.350 1	0.468	0.173 1	0.378	0.069 2	0.436	−0.181	0.435	−0.412 a	0.570	130
	F-value, p, ${\eta }^2$	3.11, <0.05, 0.008		7.39, <0.001, 0.019		3.02, <0.05, 0.008		1.12, 0.343		12.41, <0.001, 0.025

Note: M = mean standardized best–worst scores; SD = standard deviation; mean values in the same column and sub-table where the subscript is not identical differed statistically significantly using Tukey’s test (subscripted digits) or Games–Howell test (subscripted letters).

. Regarding the rank order of the five content characteristics, the results were largely consistent across all the subgroups studied. The only difference was found in the Traditionalists, who considered process value to be the third (instead of the fourth) most important characteristic and entertainment value to be the fourth (instead of the third) most important characteristic.

Beyond the observed consistency in importance ranks, t-tests and ANOVAs revealed information on differences in preferences among subgroups of respondents. Concerning gender, we found that women considered the use value (M (SD) 0.233 (0.427) vs. 0.140 (0.386), p < 0.001, d = 0.227) and entertainment value (M (SD) −0.010 (0.458) vs. −0.059 (0.439), p < 0.05, d = 0.110) of digital corporate content to be significantly more important than men did. Compared with men, women attributed significantly less importance to information value (M (SD) 0.373 (0.413) vs. 0.440 (0.458), p < 0.01, d = −0.153) and process value (M (SD) −0.191 (0.442) vs. −0.105 (0.393), p < 0.001, d = −0.204). No statistically significant difference was found between genders regarding the importance of the social value of digital corporate content (M (SD) −0.405 (0.538) vs. −0.416 (0.543), p = 0.699).

Regarding consumers’ age, the importance of information value (Welch’s F = 11.74, p < 0.001, ${\eta }^2$ = 0.022), value in use (F = 17.25, p < 0.001, ${\eta }^2$ = 0.033), entertainment value (F = 6.64, p < 0.001, ${\eta }^2$ = 0.013), and social value (Welch’s F = 21.47, p < 0.001, ${\eta }^2$ = 0.040) differed significantly among age groups, but there was no statistically significant difference in the importance of process value (Welch’s F = 1.02; p = 0.381). Post hoc analysis revealed that 50–65-year-old participants considered information value significantly more important than the younger groups of 16–29 and 30–39 years old (both p < 0.001). The 50–65-year-old participants also regarded use value as significantly more important than all younger age groups (p < 0.001 to < 0.01). In contrast, the 50–65-year-old participants found entertainment value and social value to be significantly less important than the younger age groups (p < 0.001 to < 0.05). Further, social value was deemed significantly more important by the two youngest groups compared to those aged 40–49 (p < 0.01 and 0.05, respectively).

In terms of consumer personality, the analyses showed significant differences in the importance of all five content characteristics (information value: F = 11.33, p < 0.001, ${\eta }^2$ = 0.029; value in use: Welch’s F = 11.93, p < 0.001, ${\eta }^2$= 0.028; entertainment value: F = 2.93, p < 0.05, ${\eta }^2$ = 0.008; process value: F = 2.98, p < 0.05, ${\eta }^2$ = 0.008; social value: F = 16.29, p < 0.001, ${\eta }^2$ = 0.041). Post hoc, it was found that the undercontrollers attributed statistically significantly less importance to information value and use value but more importance to social value than the other personality types (p < 0.001 to < 0.01). Undercontrollers also considered the entertainment value of digital corporate content to be more important than reserved consumers did (p < 0.05). Furthermore, reserved consumers attached greater importance to process value than resilient consumers did (p < 0.05).

Considering consumer lifestyle, the importance of information value (F = 3.11, p < 0.05, ${\eta }^2$ = 0.008), use value (F = 7.39, p < 0.001, ${\eta }^2$ = 0.019), entertainment value (F = 3.02, p < 0.05, ${\eta }^2$ = 0.008), and social value (Welch’s F = 12.41, p < 0.001, ${\eta }^2$ = 0.025) differed significantly among lifestyle types, but there was no statistically significant difference in the importance of process value (F = 1.12; p = 0.343). According to the post hoc tests, the Traditionalists attributed significantly higher importance to the use value of digital corporate content and less importance to its social value than consumers of all other lifestyle groups did (all p < 0.001 to < 0.01). In addition, information value was more important to them than to consumers from the Social Elite and Underprivileged–Modernized groups (both p < 0.05). The Traditionalists also attributed less importance to the entertainment value of content than consumers from the Underprivileged–Modernized segment did (p < 0.05).

In terms of effect size, the subgroup comparisons revealed small differences among groups (Cohen’s d < 0.50 in each case, ${\eta }^2$ < 0.06 in each case), with the greatest heterogeneity being found in assessments of the importance of social value, in line with the above counting analysis results.

5. Discussion

Scholars regard digital corporate content on company websites as a cornerstone of digital content marketing [2]. Its conceptualization, creation, and distribution accounts for a substantial share of companies’ total content marketing investments [14]. While there are studies that have investigated digital corporate content on social media [6,8,9], there is a scarcity of studies that have examined the preferences of consumers regarding the characteristics of digital corporate content in the context of company websites. This large study, which included over 1500 participants, quantified and prioritized the preferences of consumers regarding potentially gratifying key characteristics of digital corporate content on company websites using best–worst scaling as an innovative and theory-driven prioritization method. By providing evidence on what constitutes gratifying digital corporate content on company websites from a consumer perspective and whether differences exist across subgroups, this study contributes to the evolving literature on digital content marketing and provides relevant insights to help set evidence-based priorities in digital content marketing practice.

5.1. Theoretical Implications

This work, first, demonstrates that consumers consider the information value of digital corporate content to be the most important characteristic when it comes to company websites. This finding is generally in line with prior research on company websites in the wider literature outside the content marketing domain, which found that consumers’ use of company websites is driven by information seeking to a considerable degree [67]. Likewise, Ageeva et al. [68] concluded that perceived information value is a key parameter of website favorability. However, the margin by which we found information value to be most important seems remarkable. The preference for informative digital corporate content stood out clearly, as information value was considered 1.6 times more important than the next aspect and even 2.7 to 5.2 times as important as the other investigated characteristics. Consumers’ epistemic curiosity, denoting the desire for knowledge either in terms of the pleasure associated with discovering new things (interest-type curiosity) or the need for specific answers (deprivation-type curiosity) [69] may provide a useful theoretical frame to explain this marked importance of information value. Future work could assess the relevance of the two types of epistemic curiosity to the observed importance of information value drawing on research on consumer curiosity [70].

Second, our study provides evidence that digital corporate content that offers use value in terms of inspiration, ideas, advice, tips, or help in everyday life is relatively strongly preferred on company websites. The use value of digital corporate content emerged to be 1.7 to 3.3 times more important than its entertainment value, process value, or social value. This finding is basically in line with the results of other research that examined the role of the use value of digital corporate content in contexts like social media content marketing on YouTube [6], digital content marketing in tourism [7], and B2B [71], and found it to be important. However, these prior studies took a company-centric perspective by examining the benefits that companies can derive from providing useful and functional content, and they did not specifically address company websites. This work, instead, adopted a clearly consumer-centric perspective and determined consumers’ relative preference for content with use value in the context of company websites. The theory of planned behavior [72] may offer a useful theoretical lens to explain the relatively high importance consumers were found to attribute to use value in this study. According to this perspective, digital corporate content with high use value could be comparatively strongly preferred, as it would be considered as a means to achieve higher perceived behavioral control in life. Other work could further scrutinize whether this explains our result.

Third, our study demonstrated that consumers consider the entertainment value of digital corporate content on company websites to be clearly less important than its information value and use value. This is, on the one hand, consistent with previous studies in the broader literature on company websites outside the content marketing domain. Pollach [67], e.g., found informational and use value related benefits to be a stronger motivating factor for consumers’ company website use than emotive and hedonic aspects. On the other hand, our result contrasts markedly with recent studies in the field of social media content marketing that showed higher importance of the entertainment value of digital corporate content than its information value or use value for these platforms [6,30]. This work, therefore, substantiates that the relative importance of digital corporate content characteristics is platform dependent [2,15,16]. Specifically, this study reinforces that consumers weigh the importance of digital corporate content characteristics on company websites differently than they do on social media.

Fourth, this work casts some new light on the role of a gratifying usage process of digital corporate content in the context of company websites. Our results suggest that consumers attach clearly less importance to the process value of digital corporate content than to its information value and use value, and slightly less importance than to its entertainment value. We do not consider this finding to contradict studies that have demonstrated the relevance of process value-related aspects, such as usability, interactivity, aesthetics, or multisensory experiences, to company websites [11,13]. Rather, our findings help to better contextualize these studies by demonstrating the importance of process value from the consumer’s perspective relative to other characteristics of digital corporate content. In doing so, we come to a conclusion similar to that by Ageeva et al. [68], who found process value-related aspects such as visual design, navigation, or usability to be less important to the favorability of a company website than information value. The results of the present study go beyond this, however, in that we elicited the importance of process value not only relative to information value but also to use value, entertainment value, and social value.

A noteworthy fifth finding is that consumers considered social value to be the least important characteristic of digital corporate content on company websites. Lou and Xie [6] reached a similar conclusion in their study on digital corporate content on YouTube. In contrast, referring to WeChat, Chen et al. [30] found the social gratification aspect of digital corporate content to be highly relevant, especially in terms of corporate content enabling brand interactions. These heterogeneous results corroborate once again that the relative importance of digital corporate content characteristics is platform-dependent, and our findings specifically suggest that consumers, at least on average, seem to prefer digital corporate content with social value on other digital platforms rather than on company websites. However, while content priorities were generally similar among the analyzed subgroups of consumers, our analyses also revealed that it was in the stated importance of social value that the greatest heterogeneity was found among consumers. Younger people, consumers beyond traditional working-class and lower middle-class cultures, and the undercontrollers attributed at least slightly higher importance to the social value of digital corporate content on company websites. The result for the undercontrollers seems to be particularly noteworthy here, as these consumers are usually judged to be less sociable [59]. Their higher preference for the social value dimension could be due to perceiving interacting with other consumers and brands about branded content as a social compensation mechanism (compensation theory) [73] or as a means to aggressively communicate and complain [74] in line with their disposition [59]. Future research could further explore the relevance of these possible explanations.

5.2. Practical Implications

The results of this study also have practical implications. First, content marketers using company websites in their content marketing activities are advised to design the content architecture of their company websites considering the relative preferences of consumers for the different characteristics of digital corporate content as revealed in this work. This implies the need to provide a content offering that is appropriately balanced and attaches particular weight to content with information and use value, as these two content characteristics have been shown to be the top preferences overall and across all subgroups of consumers studied. Content focusing on entertainment value and social value may be better distributed on digital platforms other than the company’s own website. Naturally, these decisions must also adequately consider aspects that are beyond the scope of the present study, such as avoiding excessive social media platform dependency [75,76].

An example of a digital corporate content offering that gives priority to information and use value without sacrificing the other content dimensions is the company website zukunftswaende.de of LBS Nord, a leading public credit institution for construction and real estate financing in Germany. This corporate content hub aims to provide guidance in the complexity of modern life by addressing everyday issues and topics related to housing, life, and the future, and seeks to empower consumers with related actionable solutions. Specifically, this is obtained by weekly publication of articles and videos on topics such as sustainability, interior design ideas, do-it-yourself instructions, home construction and renovation, and individual home stories. A typical feature of this company website is communication “around the brand and product”. The content is strongly tailored to the trend-based interests of the target group and offers information, inspiration, and orientation to trigger and strengthen the desire for one’s own home.

When it comes to supportive tools to establishing content offerings that prioritize content with information and use value, one recommendation could be to employ content marketing software platforms such as Optimizely [77] or Scompler [78] for content strategy development and content planning. Such platforms provide structured collaborative workflows, for example, for defining strategic content purposes, strategic topic planning, and prioritization, and for determining the different content formats and modalities to be used [79]. In this way, they may contribute to clarify within the organization that content with information and use value should be given priority, which umbrella topics are particularly suitable for providing information and use value to the respective target groups, and with which editorial formats (e.g., news items, reports, documentaries, fact-centered interviews, infographics, data journalism, advice and service journalism) and modalities (e.g., text, audio, video, AR, chatbots, interactive content assets) this shall be achieved. This, in turn, could help to align the work of the various teams involved in planning and creating content for a company’s website (e.g., the corporate newsroom, marketing, corporate communications, sales) towards the desired content architecture.

Further, the findings of the present study not only advocate for appropriately balancing the content architecture but also for further improving, in particular, the perceived information and use value of the content offered. One direction in this regard could be to leverage the rapidly developing capabilities of artificial intelligence systems like ChatGPT (for text) or DALL-E (for images) to create digital corporate content that provides high information or use value. This could be achieved, e.g., by using generative AI for automated news writing or the creation of personalized content for individual consumers based on their interests and preferences [80,81,82]. Current practical experiences of content marketers suggest that this works quite well for factual texts or summaries of specific areas of knowledge [83]. In addition, dialogs with systems such as ChatGPT may also assist the various teams involved in planning and creating content for a company’s website in finding ideas for content that could be interesting and relevant to consumers or provide them with a tangible benefit. For example, such a dialog could uncover the most important questions consumers have about a specific topic (e.g., financing a home) or product, and these questions could then be addressed precisely with the company’s own content. Another approach to using generative AI systems to identify and fill content gaps, according to content marketing practitioners, could be asking the AI system what to write about on a particular topic [84]. In this way, the system can point to missing items or details. Although the AI system cannot judge whether the identified missing aspects are relevant, it can provide a lens through which content marketers can discover missing content more easily. However, when using these AI technologies, the associated risks must also be carefully considered, such as the need to conscientiously review the generated content for quality and accuracy, and the possibility that the content may be biased [82,85].

Another approach to creating digital corporate content with high perceived information and use value could be to adopt principles of constructive journalism [86]. This implies, first, that those involved in content conception and creation (also) see their role as service providers and advisors to consumers [87] whose purpose is to provide insights into the recognition of life issues and advice on how to deal with them [88] but also to analyze broader cultural trends and lifestyles [86]. Second, it involves a specific positive and solution-oriented mode of addressing topics and audiences, characterized by framing stories in constructive ways and providing new ideas, solutions, practices, and knowledge [86]. This may also involve applying positive psychology techniques to story generation and corporate content production [89]. Third, at the level of individual content pieces, practical solutions to specific consumer problems may be provided in different ways [88]. One way to proceed could be for the company to define relevant consumers’ life issues and then involve experts to provide solutions, advice, tips, etc. Alternatively, consumers may be interactively involved in identifying and defining relevant life issues and providing potential solutions to these issues.

5.3. Limitations and Paths for Future Research

To elicit consumers’ preferences, best–worst scaling was used as an innovative and theory-driven method. We followed principles that can be considered current methodological standards in the design, implementation, and analysis of the study [47]. Nevertheless, this work has limitations.

First, only five major characteristics of digital corporate content were included. Though they were derived from the literature and subjected to expert assessment, this number is at the lower end of the range of items commonly used in BWS studies [47], which may have led to omissions. Future research could on the one hand explore the content characteristics included in this study in more detail. Following on from the discussion, information value, e.g., could be split into value from discovering new things (interest-type curiosity) or value from specific answers (deprivation-type curiosity). On the other hand, to further complement the list of content characteristics, bibliometric analyses using software such as Citespace [90] and Vosviewer [91] could be performed. There is also a risk that certain salient content characteristics have not been adequately addressed, as the topic of this study has only been studied to a limited extent in previous research, yet the content features examined were derived from the literature. To address this, future studies could use in-depth interviews, focus groups, or observational approaches on the consumer side, or interviews with content marketing experts from companies and agencies to explore whether there are further salient content characteristics.

Second, the analyses of the heterogeneity of consumers’ preferences were confined to a limited number of subgroups. There may be variations in preferences based on factors other than those considered in this work, and these should be explored in future studies. An example would be the relational orientation of Internet users, which reflects the general tendency to maintain relationships with websites, including the propensity to express one’s opinion and share experiences [92], which could affect the preference for the social value of digital corporate content on company websites.

Third, for the heterogeneity analyses, this work relied on stratification, i.e., differentiating consumer groups by factors expected a priori to affect preferences, which is the most common approach [47]. Future studies could use latent class analysis to uncover hidden segments of consumers and hidden heterogeneity among respondents [50].

Fourth, while the sample matched the population, which supports the generalizability of the results, a non-probability online panel sampling was used for data collection. Since panel respondents were paid to participate, economic self-selection could have potentially biased the panel sample. To alleviate this, the sample’s socioeconomic composition was controlled using quotas, and no overly appealing incentive was offered.

A fifth limitation is that the sample focused on consumers in Germany, Switzerland, and Austria. Previous research has pointed to cross-cultural differences in media preferences [93,94,95]. Another path for future research could, therefore, be to elicit the preferences for digital corporate content on company websites for other regions and assess whether there are differences between them.