The Impact of Mobile Advertising Cue Types on Consumer Response Behaviors: Evidence from a Field Experiment

Abstract

This study investigates how different mobile advertising cues (WOM, product, and price cues) affect consumer responses in terms of advertisement clicks and purchases. A large-scale field experiment was conducted on a mobile online learning platform with 45,000 users representing different customer life cycle stages, in which users were randomly assigned to one of three mobile advertisement types. Behavioral data on clicks and purchases were collected, and the dual-system processing model was used to analyze mediating effects. Consumers were more likely to click on adverts featuring WOM and price cues than product cues, but less likely to purchase. Purchasing experience moderated this effect: experienced consumers showed higher purchase probabilities for WOM and price cues. Affective processing mediated click behavior, while cognitive processing mediated purchases. This study advances cue theory in the mobile context by identifying distinct psychological and behavioral mechanisms driving consumer engagement and conversion. It highlights the importance of tailoring mobile advert strategies based on cue type and user experience.

1. Introduction

Recent data indicate that the global mobile advertising market was valued at $175.62 billion in 2023 and is projected to expand at an annual growth rate of 21.8% between 2024 and 2032 [1]. In contrast to the conventional internet, the mobile internet environment enables users to access the internet and interact with content without temporal and spatial constraints, leading to distinct user behaviors [2,3]. This increased flexibility of mobile internet introduces consumers to more complex patterns of network usage and online shopping contexts, which also shapes their responses to mobile advertising [4,5]. Accordingly, this change has driven significant adaptations in marketing strategies and content design for mobile platforms [6].

As the mobile internet enables convenient information access and allows consumers to shop online in fragmented moments, advertisements on this channel must focus on delivering essential information efficiently within limited timeframes [7]. Previous research suggests that mobile advertisements often adopt a streamlined informational approach, prioritizing a single content attribute [8]. This concentrated presentation of content poses a critical challenge for firms in determining the most salient information to communicate effectively to consumers [6].

Despite the rapid growth of mobile advertising, firms continue to face ambiguity regarding the optimal type of information to convey. Cue theory posits that each piece of information functions as a consumption cue [9,10], influencing consumers’ purchase decisions [9]. Among these, product cues, word-of-mouth (WOM) cues, and price cues are the most significant and widely utilized [11]. These cues, respectively, represent intrinsic product value, value derived from social influence, and value influenced by marketing strategies [8,12]. However, determining the optimal type of cue for firms to emphasize in mobile advertising remains further investigation.

Furthermore, mobile advertising elicits varying consumer responses. For instance, some advertisements may primarily enhance consumer awareness or interest, while others may primarily influence purchase decisions [13]. Identifying the appropriate type of consumption cue to convey for an advertisement still need to be explored.

To address this important and prevalent issue, we collaborated with a prominent Chinese online education platform and conducted a large-scale online field experiment involving more than 45,000 users preparing for the CPA certificate exams. By leveraging naturally occurring consumer browsing and usage data, we examine the effects of three types of mobile advertisements distinguished by their cue content. Specifically, our study investigates how different advertising cues influence consumer responses, how prior purchase experience moderates these effects, and what underlying mechanisms may account for these behavioral differences.

This study unveils the importance of understanding cue effectiveness in mobile advertising, where attention is often selective. Our findings also highlight the distinct roles of cognitive and affective mechanisms in shaping advertising outcomes, thereby extending existing theoretical frameworks [14]. These insights provide practical implications for marketers aiming to design more effective mobile advertising strategies in app-based environments, where digital engagement plays an increasingly central role.

2. Literature Review

2.1. Cue Theory and Advertising Communication

Cue theory posits that consumers interpret product value through informational signals, especially under conditions of information asymmetry [9]. Advertising serves as a key vehicle through which firms convey both intrinsic and extrinsic cues to guide consumer evaluation [15]. Intrinsic cues reflect the inherent attributes of the product, such as taste or quality, and are immutable without altering the product itself. Extrinsic cues, such as price and reputation, are marketing-controlled signals that influence perceptions of value and trust [16,17].

Prior research indicates that intrinsic cues deliver predictive value by reflecting product quality, while extrinsic cues offer confidence value, helping consumers confirm or justify evaluations, especially in online contexts where direct product interaction is limited [18]. Consumers often rely more heavily on extrinsic cues when time is limited or product knowledge is low.

Among extrinsic cues, price and WOM are particularly salient [8,19,20]. For instance, studies found that consumers integrated different cue types (e.g., price, comments) when forming judgments [16]. Cross-cultural research further suggests variation in cue reliance: Eastern consumers emphasize social evaluation cues, whereas Western consumers prioritize intrinsic product information [21].

Cue prominence also matters. Previous studies show that focused, salient cues (e.g., price-only) elicit stronger consumer attention than mixed or diluted cue formats [22,23]. Dominant cues tend to override weaker ones in shaping consumer evaluations [24].

Consumer characteristics also influence cue processing. Experienced or trusting consumers may place greater weight on extrinsic cues for reassurance, while less experienced consumers seek intrinsic information for quality inference [25]. Individual factors such as purchase intention, search behavior, and situational context also play important roles [10].

Although extensive work has examined cue effects in traditional and online advertising, most studies are based on desktop environments and rely on qualitative or lab-based methods. Little is known about how mobile-specific contexts shape cue effectiveness or how consumers process advertising cues in real-world mobile settings. This study addresses these gaps by investigating how cue types influence mobile ad performance using quantitative field evidence.

2.2. Advertising Communication Within the Mobile Internet

Compared to desktop-based environments, mobile advertising is embedded in dynamic, often distracting user contexts [25,26]. The mobile medium enables real-time, location- and time-sensitive advertising, which enhances relevance but also requires greater precision in content delivery.

Existing literature demonstrated that location and timing significantly affect ad performance. Consumers exposed to time-relevant promotions (e.g., same-day movie offers) and those geographically closer to the promoted service are more likely to respond [4]. It also confirmed that location-based ads increase both short- and long-term sales effects [27]. Similarly, message-context congruence enhances advertising effectiveness which showed that advertisements received at home referencing family yield higher engagement than unrelated content [25].

External environmental factors also influence mobile ad outcomes. Based on behavioral restriction theory, studies show that users in crowded settings (e.g., subways) focus more on their phones, increasing ad receptivity [26]. Weather conditions can also affect conversion rates, with sunnier weather linked to higher ad effectiveness [28].

User traits and motivations further shape mobile ad responses. Existing literature highlight that mobile shopping continuance is influenced by hedonic and utilitarian values within a confirmation framework [29]. Previous study also found that mobile ads are more persuasive for practical, high-involvement products, suggesting that consumers’ pre-existing knowledge and involvement levels moderate ad effectiveness [30]. And empirical evidence pointed out that customization enhances behavioral intention, especially when mediated by technology-related perceptions [31].

While these studies offer valuable insights into the contextual and behavioral dimensions of mobile advertising, most focus on external or environmental factors, with less attention given to the content of the advertisements themselves. In mobile environments, where consumer attention is more selective, understanding which types of cues are most effective becomes critical. This study contributes by shifting the focus toward cue content in mobile ads and examining how such cues interact with consumer characteristics to influence engagement and purchase behavior.

3. Theoretical Background and Hypotheses

Cue theory suggests that product attributes can be manifested through various cues that can significantly influence consumers’ purchasing behavior [20,32,33]. One of the earliest offline applications of cue theory was in terms of brick-and-mortar bakeries. The bakers released the newly baked bread aroma at the store entrance, as an olfactory cue reflecting product quality to attract consumers [9]. Cues about the relevant attributes of products are commonly delivered through advertising [8,16].

Research has indicated that consumers are drawn to cues that provide more information if they are unfamiliar with a product. Auxiliary cues related to product information can serve as the basis for more considered judgments and distinctions [9]. In a study of television advertising, it found that price-oriented messages, rather than those emphasizing product features, were more likely to be recalled and evaluated positively by consumers [34]. Examining advertisements in both online and offline shopping contexts, the study found that consumers’ decision-making processes were more strongly influenced by core attributes related to third-party endorsements than by product attributes [35]. The cue of prices depicted in adverts can also increase consumers’ awareness of products, leading them to engage in further thought processes. Therefore, we propose the following hypothesis:

H1: 

Consumers are more likely to click on mobile adverts featuring (a) price cues and (b) WOM cues than those featuring product cues.

Product cues can involve either predictive value, which reliably predicts product quality, or confidence value, which indicates the level of confidence consumers have in their judgments [9]. Cues that focus on product attributes can be either intrinsic or extrinsic. Intrinsic cues refer to the inherent physical characteristics of the product that cannot be altered without changing its essence, such as its shape, taste, or quality. Extrinsic cues refer to external attributes of the product that can be modified through marketing, including price, WOM, and brand names [16,17,18].

Consumers tend to prioritize the intrinsic cues that are related to the core quality and value of the product, as these offer higher predictive value. Extrinsic cues such as price and WOM offer higher confidence value and assist consumers in making judgments [8]. Thus, intrinsic cues have a greater effect on purchasing decisions than extrinsic cues. Consumers with a higher willingness to purchase will also place greater importance on intrinsic cues related to the product or brand, as they can bring about the utility associated with the product [17,35]. Therefore, we propose the following hypothesis:

H2: 

Consumers are less likely to purchase based on mobile adverts featuring (a) price cues and (b) WOM cues than those featuring product cues.

The impact of advertising cues on purchasing decisions is also influenced by consumer heterogeneity. Different consumers may engage with and utilize either intrinsic or extrinsic cues for the same product [36]. The experiences and focus of consumers regarding a product or service may vary according to whether they are situated in the pre-purchase or post-purchase stage of the product life cycle [35,37,38]. Consumers with previous purchasing experience may not be persuaded to find out more about a product through extrinsic cues like price and WOM. Thus, the attractiveness of these cues diminishes for experienced consumers [17]. In light of this, we propose the following hypothesis:

H3: 

Previous purchasing experience negatively moderates the likelihood of consumers clicking on mobile adverts that feature price and WOM cues.

Consumers with purchasing experience will have higher levels of familiarity with the product’s intrinsic attributes when making subsequent purchases. Consequently, they will place higher importance on the confidence value of external cues, such as brand and reviews, which will affect their judgments [34,38]. Therefore, we suggest that the impact of advertising cues on purchase behavior will be moderated by the consumer’s position in the life cycle. We further propose the following hypothesis:

H4: 

Previous purchasing experience positively moderates the likelihood of consumers purchasing based on mobile adverts featuring price cues and WOM cues.

Behavioral research has shown that when individuals receive different types of information, they may apply either cognitive or affective processing systems [39]. The cognitive processing system involves using cognitive resources to analyze and think about information systematically and thus is a complex and relatively slow type of processing that is relatively rational. Conversely, the affective processing system responds to information intuitively and with more instinctive emotional reactions [39]. It is therefore simpler and faster and represents a more emotional mode of information processing [24].

In terms of attentional focus, non-sensory or numerical attributes are processed cognitively, whereas perceptible and surface attributes are processed affectively, and the study further demonstrated that preferences are driven specifically by one of these two systems [14]. The affective system can generate a strong, intuition-based preference, which is then supported by cognitive processing. However, if the affective system focuses on surface characteristics without generating a strong preference, the cognitive system guides decision-making by conducting attribute-based evaluations. Preferences are then mainly determined by the cognitive system.

Earlier study applied dual-process theory to explain differences in decision-making strategies across print magazine advertisements and shopping websites, and found that when advertisements emphasized product attributes, consumer preferences were predominantly shaped by the cognitive system [40]. In mobile advertising, price and WOM information tends to be more prominent than product information, and is thus more effective in activating individuals’ affective processing systems [41]. Therefore, we propose the following hypothesis:

H5: 

Through the mediating effect of the affective processing system, consumers are more likely to click on mobile adverts featuring price and WOM cues than those featuring product cues.

When consumers consider making a purchase, they become more closely involved with the product [38]. Advertising cues typically offer more complex information related to the specifics of the product. The higher the complexity of information, the more cognitive resources must be invested in thinking and understanding, leading to greater processing by the cognitive system [23]. The consumer’s cognitive processing system then takes precedence. Based on this, we propose the following hypothesis:

H6: 

Through the mediating effect of the cognitive processing system, consumers are more likely to purchase based on mobile adverts featuring product cues.

Figure 1 shows the framework of the research.

4. Experiment Design and Model Setup

4.1. Experiment Design

To empirically test our hypotheses, we conducted a field experiment in collaboration with a leading Chinese online education platform specializing in accounting courses. This platform, with its extensive and engaged mobile user base, provided an ideal setting for real-world testing of advert effectiveness across different content cue types, while minimizing potential interference from other product categories or user behaviors. We designed three types of mobile in-app banner advertisements based on distinct information cues: price cues, product cues, and WOM) cues. These were deployed in a natural consumer environment, allowing us to observe user–advertisement interactions.

(1)

Experiment Overview. The experiment took place over a one-week period beginning 23 August 2019. During this time, the platform displayed banner adverts on both the homepage (i.e., the app’s landing interface) and the secondary advert content page (accessed via banner clicks). Adverts were presented continuously throughout the day to maximize user exposure.

(2)

Sample. To capture user behavior across different customer life cycle stages, we employed a stratified random sampling method. A total of 45,000 platform users were included in this study. To examine user responses across different customer life cycle stages, we implemented a stratified random sampling procedure in three steps:

First, we first categorized all platform users into three distinct groups based on their prior engagement history: unregistered users, registered users without purchase history and registered users with purchase history. From each group, we randomly selected 15,000 users, resulting in a pool of 45,000 users in total.

Second, from each of the three groups above, we further randomly sampled 5000 users and combined them to form a new experimental group consisting of 15,000 users. This process was repeated three times, producing three new experimental groups, each comprising 5000 users from each of the original engagement types. This ensured that every experimental group contained a balanced representation of all three user types. Then, each of the three newly formed groups was randomly assigned to receive one of the three advertisement cue treatments: price cue, product cue, or WOM cue.

(3)

Experiment Process. Users first encountered a top-level banner featuring one of the three cue types. Clicking the banner directed them to a second-level advert page, where cue-aligned content was reinforced via matching copy and small cue-specific icons. After viewing, users could choose whether to proceed with a purchase. All other experimental elements, such as course details, layout, and purchase mechanisms, were held constant to isolate cue effects.

(4)

Mobile Advert Design. The promoted product was a CPA pre-exam intensive training course. The top banner advertisement slogans displayed on this mobile app that represented price cues, product cues, and WOM cues, respectively, were as follows: “2019 Pre-exam Spotlight Intensive Training Class, $300 off per subject, and another 20% off for 2 or more subjects, grab it now!” “2019 Pre-Exam Spotlight Intensive Training Classes, taught by famous teachers, 45-h crash course to help you easily get to 60+, grab it now!” and “2019 Pre-Test Point Close Training Class, highly recommended by previous students, the majority of candidates’ choice, join the study immediately, grab it now!”

When clicking on the top banner advertisement, a consumer was directed to the detailed advertisement page. On this page, a circular floating icon of the simplified advertisement was shown at the middle right side of the page, ensuring continued visibility as users browsed. On this floating icon, corresponding mobile taglines were prominently displayed—“Save $300 now”, “Helps you get to 60+” and “Highly recommended”.

Both the banner and the floating advertisement icon were designed with a dark-blue background, which contrasted sharply with the app’s overall white interface, thereby drawing consumer attention. Within the banner, the emphasized cue words—“$300 OFF”, “GET to 60+”, and “MAJORITY of CANDIDATES’ CHOICE”—were highlighted in bold fluorescent red, while the remaining text appeared in non-bold orange. This color contrast further enhanced the salience of the emphasized cues. The banner remained persistently displayed at the top of the app’s entry screen, ensuring that all users were exposed to it upon app entry, while the floating icon provided additional visibility on the specific advertisement page.

4.2. Descriptive Statistics and Variables

Throughout the experiment, the advertisement click rate was 14.93%, calculated as (number of click/number of browsing) × 100%, while the purchase conversion rate, calculated as (number of purchases/number of clicks) × 100%, amounted to 19.01%. Figure 2 and Figure 3 illustrate the model-free results of the click data and purchase data for mobile advertisements pertaining to the three distinct cue types, respectively.

Figure 2 and Figure 3 reveal that product-cue adverts exhibit a lower click-through rate than price-cue and WOM-cue adverts. The results for purchase rates show an opposite pattern, with purchases of product-cue adverts surpassing those of price-cue adverts and WOM-cue adverts. These descriptive statistics lend support to H1 and H2 from a data-driven perspective.

We now define the variables in our study, as outlined in

Table 1. Variables and Definitions.

Title 1	Variable	Definition
Dependent Variable	PURCHASE	Purchase or not: 1 = Purchase; 0 = otherwise
	CLICK	Click on the ad or not: 1 = click; 0 = otherwise
Independent Variable	PRICE	Is it a price-cue ad? 1 = yes; 0 = otherwise
	WOM	Is it a WOM-cue ad? 1 = yes; 0 = otherwise
Moderator	BUY	Registered users with purchase records: 1 = yes; 0 = otherwise
	SIGN	Registered users without purchase records: 1 = yes; 0 = otherwise
Control Variable	LEARN	Learning other courses in progress: 1 = yes; 0 = otherwise
	DUR1	Time spent on the first-level page
	DUR2	Time spent on the ad-specific content page

.

Dependent Variables. We identified two dependent variables, CLICK and PURCHASE, which we assessed using dummy variables. They took a value of 1 when the consumer chose to click/purchase and a value of 0 if the consumer did not choose to do so.

Independent Variables. The mobile application adverts analyzed in this paper encompassed three types based on the consumption cues of price, WOM, and product. We took a dummy variable measurement approach for the independent variables, based on the assumptions outlined in the preceding section. The mobile adverts reflecting the product cue served as the baseline for comparison with the other two types of adverts, which were then set as dummy variables, i.e., price-cue mobile adverts (PRICE) and WOM cue mobile adverts (WOM).

Moderator. As we examined the moderating effect of consumers’ buying experience on purchases, we categorized consumers into three groups: unregistered with no purchase, registered with no purchase, and registered with a purchase. We considered the previous assumptions and designated the scenario of being unregistered and having no purchase as the baseline, while the other two scenarios were designated as dummy variables. Thus, a consumer who registered before viewing the advert but had no record of purchase was denoted as SIGN, whereas a consumer who both registered and had a record of purchase was denoted as BUY.

Control Variables. To further understand the click and purchase behavior of consumers, we controlled for other variables that may potentially influence their decisions. First, we controlled for whether consumers were participating in an ongoing course during the experiment period (LEARN), which served as a measure of individual heterogeneity. Second, in our exploration of consumers’ purchasing behavior, we accounted for the duration of stay on the first-level interface (DUR1) and on the advert content page (DUR2), which can both be used to assess consumer behavior throughout the browsing and purchasing process.

Before conducting regression analyses of the data and variables, we performed correlation analyses for the selected key variables. The results are given in

Table 2. The correlation of variables.

	PURCHASE	PRICE	WOM	BUY	SIGN	LEARN	DUR1	DUR2
PURCHASE	1.000
PRICE	−0.001	1.000
WOM	−0.008 *	−0.500 *	1.000
BUY	0.029 *	−0.000	0.000	1.000
SIGN	−0.014	0.000	0.000	−0.500 *	1.000
LEARN	0.016 *	−0.006	−0.002	0.828 *	−0.414 *	1.000
DUR1	0.003	−0.001	0.003	0.070 *	−0.040 *	0.070 *	1.000
DUR2	0.028	−0.002	0.006	0.008	−0.001	0.011	0.008	1.000
CLICK	0.001 *	0.027 *	0.005	0.018 *	−0.010 *	0.010	0.014 *	0.024 *

* p < 0.1.

, and indicate significant correlations among the key variables, which serves to support the subsequent analysis.

4.3. Empirical Model

To investigate our research hypotheses, we used a discrete logit model for regression analysis [42] to explore the impact of the independent and moderating variables on consumer behavior. In line with the hypotheses, mobile adverts of the product cue type were used as the baseline, resulting in the establishment of two dummy variables: mobile adverts for the price cue and the WOM cue.

When examining the effect of mobile advert types on click behavior, we assumed that the utility value of consumer i at moment t when choosing to click on a certain type of adverts could be represented as follows:

$$U_{it}=V\left({\alpha }_i+{\beta }_1{PRICE}_{it}+{\beta }_2{WOM}_{it}+{\beta }_3{BUY}_{it}+{\beta }_4{SIGN}_{it}+{\beta }_5{LEARN}_{it}+{\mathsf{\Gamma }}_t\right)+{\epsilon }_{it}$$

(1)

When investigating the impact of mobile advert types on purchasing behavior, we proposed the following regression equation for consumer i at moment t when opting to purchase via a specific type of advertisement:

$$U_{it}=V\left({\alpha }_i+{\beta }_1{PRICE}_{it}+{\beta }_2{WOM}_{it}+{\beta }_3{BUY}_{it}+{\beta }_4{SIGN}_{it}+{\beta }_5{LEARN}_{it}+{\beta }_6{DUR1}_{it}+{\beta }_7{DUR2}_{it}+{\mathsf{\Gamma }}_t\right)+{\epsilon }_{it}$$

(2)

To examine the moderating effect of consumers’ experience on clicks, we introduced interaction terms into regression Equation (1) to assess the impact of consumers’ purchasing experience. This regression equation was as follows:

$$\begin{gathered} U_{it}=V\left({\alpha }_i+{\beta }_1{PRICE}_{it}+{\beta }_2{WOM}_{it}+{\beta }_3{BUY}_{it}+{\beta }_4{BUY}_{it}\times {PRICE}_{it}+{\beta }_5{BUY}_{it}\times {WOM}_{it}+{\beta }_6{SIGN}_{it} \\ +{\beta }_7{SIGN}_{it}\times {PRICE}_{it}+{\beta }_8{SIGN}_{it}\times {WOM}_{it}+{\beta }_9{LEARN}_{it}+{\mathsf{\Gamma }}_t\right)+{\epsilon }_{it} \end{gathered}$$

(3)

Similarly, we also use the interaction term to test the moderation effect of consumers’ experience on purchase.

According to the utility maximization principle, the probability formula for consumer i to choose whether to click on the mobile application advert or make a purchase through the mobile ad-pushed content at moment t was:

$$P_{it}={\displaystyle \frac{exp\left(V_{it}\right)}{1+exp\left(V_{it}\right)}}$$

(4)

where ${\alpha }_i$ is individual-related and only varies with the individual but not with time, such as customer gender, while ${\mathsf{\Gamma }}_t$ does not vary with the individual but only with time.

5. Results

5.1. The Impact of Cue Types on Clicks and Purchase

To assess the impact of mobile advertisement cues on clicks, we applied the regression model in Equation (1). First, we checked for any multicollinearity using the variance inflation factor (VIF) test. We find that the VIF values are all below 10, with an average of 3.07, thereby confirming the absence of multicollinearity. The regression results of the model are presented in

Table 3. The impact of mobile ads on click choice.

	DV: CLICK
	Model (1)	Model (2)	Model (3)
PRICE	0.097 ***		0.098 ***
	(0.000)		(0.000)
WOM		0.833 ***	0.833 ***
		(0.001)	(0.000)
BUY	2.121 ***	2.120 ***	2.120 ***
	(0.000)	(0.000)	(0.000)
SIGN	0.002	0.002	0.002
	(0.977)	(0.930)	(0.957)
LEARN	−0.132 **	−0.133 **	−0.133 *
	(0.002)	(0.002)	(0.003)
LR chi2	10.09	12.63	14.73
Log-Likelihood	−127.10	−125.75	−124.08
Observation	45,000	45,000	45,000

*** p < 0.01, ** p < 0.05, * p < 0.1.

. Model (1) solely incorporates the independent variable of whether the mobile adverts include price cues; model (2) solely includes whether the mobile adverts include WOM cues; and model (3) incorporates both the independent variables of price cues and WOM cues.

Table 3 clearly reveals that with mobile app adverts featuring product cues as the baseline, the inclusion of price cues yields a significant and positive effect on clicks compared with product-cue adverts (β1 = 0.098, p < 0.001) and WOM-cue adverts (β2 = 0.833, p < 0.001). These results confirm H1. Thus, the data confirm that consumers are more likely to click on mobile adverts featuring price cues and WOM cues than on those featuring product cues.

We then tested the data based on the regression model in Equation (2). We first checked the multicollinearity problem of the model (all VIF values are below 10, with an average of 4.18). The results are shown in Table 3, in which model (1) solely incorporates the dummy variable of price cues, model (2) solely includes the dummy variable of WOM cues, and model (3) incorporates both variables. Here, we also considered the control variable of duration.

Table 4. The impact of mobile adverts on purchases.

	DV: PURCHASE
	Model (1)	Model (2)	Model (3)
PRICE	−0.613 ***		−0.656 ***
	(0.004)		(0.004)
WOM		−1.370 ***	−1.356 ***
		(0.002)	(0.003)
BUY	2.261 ***	2.258 ***	2.256 ***
	(0.000)	(0.000)	(0.000)
SIGN	0.004	0.004	0.004
	(0.969)	(0.981)	(0.987)
LEARN	−0.770	−0.770	−0.772
	(0.091)	(0.092)	(0.092)
DUR1	0.000	0.000	0.000
	(0.824)	(0.813)	(0.810)
DUR2	−0.099	−0.098	−0.098
	(0.846)	(0.841)	(0.834)
LR chi2	10.16	13.63	15.37
Log-Likelihood	−147.30	−145.57	−144.70
Observation	45,000	45,000	45,000

*** p < 0.01.

shows that with product cues as the baseline, the inclusion of price-cue adverts has a significant and negative effect on purchases (β1 = −0.656, p < 0.001), as does the inclusion of WOM-cue adverts (β2 = −1.356, p < 0.001). These results support H2, which suggests that consumers are less likely to purchase adverts featuring price cues and WOM cues than those featuring product cues.

5.2. The Moderator of Consumer Experience on Clicks and Purchases

We then used the model in Equation (3) to examine the moderating effect of consumers’ purchasing experience on their click behavior toward mobile adverts with different cue types. The test for multicollinearity indicates the absence of covariance in the model (VIF = 2.08). The empirical test results are presented in

Table 5. The moderating effect of consumer experience on clicks.

	DV: PURCHASE
	Model (1)	Model (2)	Model (3)
PRICE	0.092 ***	0.091 ***	0.091 ***
	(0.000)	(0.000)	(0.000)
WOM	0.802 ***	0.802 ***	0.803 ***
	(0.001)	(0.001)	(0.000)
BUY	2.088 ***	2.088 ***	2.092 ***
	(0.000)	(0.000)	(0.000)
BUY * PRICE	−0.027 ***		−0.026 ***
	(0.000)		(0.000)
BUY * WM	−0.019 ***		−0.019 ***
	(0.000)		(0.000)
SIGN	0.002	0.002	0.002
	(0.893)	(0.860)	(0.805)
SIGN * PRICE		0.000	0.000
		(0.523)	(0.694)
SIGN * WOM		0.000	0.000
		(0.482)	(0.749)
LEARN	−0.131 **	−0.131 **	−0.132 *
	(0.002)	(0.001)	(0.002)
LR chi2	10.11	12.97	15.03
Log-Likelihood	−122.62	−121.08	−119.77
Observation	45,000	45,000	45,000

*** p < 0.01, ** p < 0.05, * p < 0.1.

.

Table 5 shows that model (1) includes only the corresponding interaction terms for consumers with purchase records, while model (2) includes the interaction terms for consumers who have registered but have no purchase record. Model (3) incorporates the variables and interaction terms for both types of consumers. Model (1) indicates that the interaction term for consumers with purchase records is significant and negative (β4 = −0.027, p < 0.001; β5 = −0.019, p < 0.001). Conversely, in model (2), the interaction term is not significant for consumers who have registered but have no purchasing experience (β7 = −0.000, p = 0.523; β8 = −0.000, p = 0.482). Thus, we find no moderating effect when consumers have no purchasing experience, unlike for those who have not registered but have made purchases on the platform. These findings provide support for H3, indicating that consumers with rather than without purchasing experience are less likely to click on mobile adverts with price cues and WOM cues.

We then examined the moderating role of consumers’ position in the life cycle on their purchasing behavior. Testing for multicollinearity confirms the absence of any covariance in the model (VIF = 2.08).

Table 6. The moderating effect of consumer experience on purchases.

	DV: PURCHASE
	Model (1)	Model (2)	Model (3)
PRICE	−0.656 ***	−0.628 ***	−0.656 ***
	(0.001)	(0.000)	(0.000)
WOM	−1.346 ***	−1.315 ***	−1.315 ***
	(0.000)	(0.000)	(0.000)
BUY	0.314 ***	0.314 ***	0.315 ***
	(0.000)	(0.000)	(0.000)
BUY * PRICE	0.104 ***		0.105 ***
	(0.000)		(0.000)
BUY* WM	0.162 ***		0.162 ***
	(0.000)		(0.000)
SIGN	−0.002	−0.003	−0.001
	(0.903)	(0.912)	(0.978)
SIGN * PRICE		0.000	0.000
		(0.332)	(0.973)
SIGN * WOM		0.000	0.000
		(0.122)	(0.996)
LEARN	−0.772	−0.771	−0.771
	(0.092)	(0.095)	(0.093)
DUR1	0.000	0.000	0.000
	(0.810)	(0.812)	(0.810)
DUR2	0.984 ***	0.985 ***	0.985 ***
	(0.000)	(0.000)	(0.000)
LR chi2	15.37	17.10	17.37
Log-Likelihood	−147.70	−157.71	−144.70
Observation	45,000	45,000	45,000

*** p < 0.01, * p < 0.1.

reports the results.

In Table 6, model (1) only includes the corresponding interaction terms for consumers with purchase records, while model (2) includes the interaction terms for consumers who have registered but have not made a purchase. Model (3) includes both variables and interaction terms for consumers at different stages in the life cycle. The results of model (1) indicate that the interaction terms for consumers with purchase records are significant (β4 = 0.104, p < 0.001; β5 = 0.162, p < 0.001). This suggests that internal cues do not play a prominent role for this group of consumers, as they already possess ample knowledge about the product due to their rich experience or because they are in the mature stage of the life cycle. In contrast, external cues, such as price and WOM, significantly influence the purchasing decisions of this group, with WOM cues having a greater effect than price cues.

Model (2) indicates that the interaction term is not significant (β7 = −0.000, p = 0.332; β8 = −0.000, p = 0.122) for consumers who have already registered but have no purchasing experience. This implies that although they have already registered, they still lack purchasing experience and therefore have limited knowledge of and familiarity with the product, so the internal cues of the product remain the main influencing factor in their purchasing decision-making. The purchasing behavior and sensitivity to advertising cues of this group also align with those of unregistered new users, but not with those who have already made a purchase. This suggests that the decision to register on the platform does not significantly impact the influence of the life cycle on purchasing behavior. Rather, whether the consumer is in the post-purchase life cycle stage is important. This analysis of the moderating effect supports H4, which states that consumers with purchasing experience have a higher probability of buying a new product via mobile adverts with price cues and WOM cues than via those with product cues. As shown in Table 6, once consumer experience heterogeneity is taken into account by including the interaction between different level of consumer experiences and purchase, the time spent on the specific advertisement content page (DUR2) exhibits a significant positive effect on purchase behavior. In other words, when consumer experience heterogeneity is explicitly modeled, the effect of advertisement viewing time on purchase decisions becomes more evident. This result suggests that viewing duration is a critical factor shaping the influence of advertising content on purchase behavior across different levels of consumer experience.

5.3. The Mediator of the Dual System

To examine the mediating role of the dual-system model in consumers’ click and purchase behaviors, we conducted a between-groups experiment. Each participant was exposed to one of three mobile advertisement cue types and asked to assess their click and purchase intentions. Participants also completed dual-system processing scales.

We used images of the actual adverts from the field study and randomly assigned participants into three groups, each exposed to one advert type. Click and purchase intentions were measured on a 7-point Likert scale (1 = “not at all” to 7 = “definitely”). Cognitive system processing was assessed using statements such as “The picture made me learn more about the product,” “made me more aware,” and “helped me evaluate the product” (α = 0.921). Affective system processing was measured by items like “The picture evoked numerous emotions,” “stirred various feelings,” and “touched my heart” (α = 0.905), also on a 7-point scale [23].

A significance test revealed differences across advert types (p = 0.003). Click intention was significantly lower for product cues (M = 3.98) than for price cues (M = 4.37, p = 0.008) and WOM cues (M = 4.29, p = 0.009). Conversely, purchase intention was higher for product cues (M = 4.27) than for price (M = 4.02, p = 0.009) and WOM cues (M = 4.10, p = 0.005), supporting H1 and H2.

We then conducted a mediation analysis using bootstrap tests for multi-categorical variables following [43] and using product-cue advertisement as the baseline.

For click behavior, Figure 4 shows the results. Comparisons between product and price cues (A and C) showed no significant mediation via the cognitive system (A: LLCI = −0.144; ULCI = 0.526), but a significant mediation via the affective system (C: LLCI = 0.218; ULCI = 0.829; indirect effect = 0.085). Similarly, between product and WOM cues (B and D), the cognitive system was non-significant (B: LLCI = −0.027; ULCI = 0.690), while the affective system had a significant mediating effect (D: LLCI = 0.209; ULCI = 0.851; indirect effect = 0.593). After controlling for the affective system, direct effects became non-significant (C: LLCI = −0.221; ULCI = 0.324; D: LLCI = −0.234; ULCI = 0.629), indicating full mediation and supporting H5.

For purchase behavior, we tested the cognitive system’s mediation. The outcomes of the mediating effect of the cognitive system are presented in Figure 5. In the comparison of product and price cues (E and G), the affective system was not significant (G: LLCI = −0.223; ULCI = 0.426), whereas the cognitive system showed a significant indirect effect (E: LLCI = 0.026; ULCI = 0.391; indirect effect = 0.197). A similar pattern held for the product vs. WOM comparison (F and H): affective system was non-significant (H: LLCI = −0.142; ULCI = 0.247), while cognitive mediation was significant (F: LLCI = 0.107; ULCI = 0.645; indirect effect = 0.266). In both cases, direct effects were non-significant (E: LLCI = −0.209; ULCI = 0.428; F: LLCI = −0.290; ULCI = 0.387), indicating full mediation by the cognitive system.

These findings confirm that affective processing mediates click behavior for price and WOM cues, while cognitive processing mediates purchase behavior, consistent with the dual-system model framework.

6. Discussion and Conclusions

6.1. Key Findings

This study investigates how different mobile advertising cues (price, product and WOM cues) affect consumer responses in the dynamic mobile Internet environment. Leveraging a randomized field experiment in partnership with a major Chinese online learning platform, we evaluated real consumer behavior by tracking click-through and purchase rates across advert types. The results reveal a consistent pattern: WOM and price cues significantly increase click rates, but these do not translate into higher purchase probabilities. In contrast, product cues yield lower click-through rates but lead to higher conversion rates.

Importantly, we found that consumer purchasing experience moderates these effects. Users with prior purchase history are more responsive to price and WOM cues, showing higher purchase intent under these conditions compared to product cues, an effect not observed among users without such experience.

Our mediation analysis further unpacks these behavioral differences through the lens of dual-system processing. Specifically, affective system activation mediates the relationship between price or WOM cues and clicks, explaining why these cues are effective in capturing immediate attention. Conversely, cognitive processing plays a central role in driving purchases, particularly when product-focused messages are used.

6.2. Theoretical Contributions

This study advances theory in several important ways. First, we extend the literature on advertising cues by shifting analytical attention from traditional and PC-based advertising to the mobile Internet context, where consumers experience content under more fragmented and real-time conditions [35,41]. Second, from a methodological perspective, this study moves beyond previous qualitative and survey-based approaches by implementing a randomized field experiment and analyzing behavioral data derived from actual consumer usage, thereby providing evidence from observed rather than self-reported behavior and contributing to enhancing the realism and robustness [27]. Third, we contribute to cue theory by differentiating the impacts of WOM, price, and product cues not just on consumer attention (clicks), but also on actual decision-making (purchases). This two-stage view of advertising response is particularly relevant in mobile contexts, where attention is scarce and decision windows are short [5]. Finally, our integration of the dual-system model provides theoretical insight into how affective and cognitive mechanisms operate distinctly within mobile advertising, bridging cognitive psychology and marketing communication [10,14].

6.3. Managerial Implications

From a practical standpoint, our findings offer actionable guidance for marketers aiming to optimize mobile advertising. First, WOM and price cues are highly effective at generating clicks, making them valuable tools for capturing early-stage consumer interest. However, since these cues do not necessarily lead to conversions, advertisers should transition users toward product-centered messages to increase the likelihood of purchase [5]. Second, our results underscore the importance of segmentation based on purchase history. Consumers with prior purchases are more likely to respond favorably to price and WOM cues, while inexperienced users may require more cognitive input, such as detailed product features, to convert. This suggests that tailoring ad content based on user behavior profiles can enhance overall campaign performance [29]. Third, the dual-system framework implies that effective mobile adverts should engage both the heart and the mind. Emotional elements like user testimonials or discounts should be paired with informative product attributes to activate both affective and cognitive pathways, thereby improving both engagement and sales [14].

6.4. Limitation and Future Research

While we randomized exposure to three cue types through field experiments, we could not fully control for potential variation in adverts tone or wording intensity, which may have introduced subtle bias. Future research should incorporate message calibration procedures or eye-tracking and engagement metrics to better isolate the effect of cue types. Additionally, exploring cross-cultural generalizability and long-term consumer behavior could further enhance our understanding of mobile advert efficacy. Moreover, our findings are derived from a single cultural context, which may limit their generalizability. Previous cross-cultural research suggests that Eastern consumers tend to emphasize social evaluation cues, whereas Western consumers prioritize intrinsic product information [21]. Future research should examine whether the effects we document hold across cultural settings and how cultural orientations moderate consumers’ responses to different advertising cues.