The Interplay of Inflated Expectations, (Dis)Confirmation, and Emotional Spillover: Implications for Post‐Purchase Loyalty in Mystery Deals

Brodschelm, Florian; Vetter, Sebastian; Hüttl‐Maack, Verena; Lazarovici, Isabel‐ Sophie; Schumann, Jan Hendrik Article — Published Version The Interplay of Inflated Expectations, (Dis)Confirmation, and Emotional Spillover: Implications for Post‐Purchase Loyalty in Mystery Deals Psychology & Marketing Provided in Cooperation with: John Wiley & Sons Suggested Citation: Brodschelm, Florian; Vetter, Sebastian; Hüttl‐Maack, Verena; Lazarovici, Isabel‐ Sophie; Schumann, Jan Hendrik (2025) : The Interplay of Inflated Expectations, (Dis)Confirmation, and Emotional Spillover: Implications for Post‐Purchase Loyalty in Mystery Deals, Psychology & Marketing, ISSN 1520-6793, Wiley, Hoboken, NJ, Vol. 42, Iss. 11, pp. 2883-2901, https://doi.org/10.1002/mar.70017 This Version is available at: https://hdl.handle.net/10419/330169 Standard-Nutzungsbedingungen: Die Dokumente auf EconStor dürfen zu eigenen wissenschaftlichen Zwecken und zum Privatgebrauch gespeichert und kopiert werden. Sie dürfen die Dokumente nicht für öffentliche oder kommerzielle Zwecke vervielfältigen, öffentlich ausstellen, öffentlich zugänglich machen, vertreiben oder anderweitig nutzen. Sofern die Verfasser die Dokumente unter Open-Content-Lizenzen (insbesondere CC-Lizenzen) zur Verfügung gestellt haben sollten, gelten abweichend von diesen Nutzungsbedingungen die in der dort genannten Lizenz gewährten Nutzungsrechte. Terms of use: Documents in EconStor may be saved and copied for your personal and scholarly purposes. You are not to copy documents for public or commercial purposes, to exhibit the documents publicly, to make them publicly available on the internet, or to distribute or otherwise use the documents in public. If the documents have been made available under an Open Content Licence (especially Creative Commons Licences), you may exercise further usage rights as specified in the indicated licence. https://creativecommons.org/licenses/by/4.0/ Psychology & Marketing RESEARCH ARTICLE The Interplay of Inflated Expectations, (Dis)Confirmation, and Emotional Spillover: Implications for Post‐Purchase Loyalty in Mystery Deals Florian Brodschelm 1 | Sebastian Vetter 1 | Verena Hüttl‐Maack 2 | Isabel‐Sophie Lazarovici 1 | Jan Hendrik Schumann 1 1 Chair of Marketing and Innovation, University of Passau, Passau, Bavaria, Germany | 2 Chair of Marketing and Consumer Behavior, University of Hohenheim, Hohenheim, Baden‐Wuerttemberg, Germany Correspondence: Isabel‐Sophie Lazarovici ([email protected]) Received: 20 December 2024 | Revised: 12 July 2025 | Accepted: 14 July 2025 Keywords: affective reaction | customer loyalty | disconfirmation effect | mystery deals | positive affective spillover ABSTRACT Mystery deals, a marketing tactic used to boost sales, involve withholding the exact version of the product a customer will receive until after purchase. While prior research has examined consumers' pre‐purchase reactions to mystery deals, their post‐ purchase responses remain underexplored. Addressing this gap, we contribute to the mystery marketing literature by investigating how consumers respond after the purchase when all details of the deal are revealed. Across six empirical studies (1 field, 3 online, 1 lab that uses IA‐based facial expression recognition software, 1 supplementary), we disentangle the complex interplay of curiosity‐induced inflated expectations, (dis)confirmation, and positive affective spillover effects generated during the pre‐purchase phase. The results highlight the crucial role of positive affect in substantially enhancing customer loyalty during the post‐purchase phase—compared to traditional deals where all details are known in advance. These affective processes even counterbalance the negative effects on loyalty when consumers receive their non‐preferred outcome. For managers, our findings provide insights into how mystery deals impact customer retention, expanding beyond prior research that has primarily focused on customer acquisition. Thus, mystery deals can serve as a viable strategy to sustain customer loyalty, even when individual preferences cannot be fully met. 1 | Introduction Many retailers employ mystery deals (MDs) as a marketing tactic, where the key product or service remains unknown before purchase (Guo et al. 2024; Hill et al. 2016; Kovacheva et al. 2022; Kovacheva and Nikolova 2023). MDs aim to boost demand, optimize capacity, or clear inventory (e.g., Fay and Xie 2008). For instance, Burger King offered a “mystery box” with one of eight possible burgers at a discount, without revealing which one customers will receive in advance (mein‐ deal.com 2019). Similarly, Pura Vida Bracelets, an online retailer, promotes surprise bracelets and rings online (Pura Vida 2024). More examples appear in Appendix A, Table A.1. Unlike other mystery marketing tactics—such as mystery discounts, mystery gifts, or mystery teaser advertisements— mystery deals (MDs) incorporate uncertainty and surprise directly into the core offering itself. In these cases, the main component of the purchase (i.e., the product or service the consumer is actually buying) remains unknown at the point of purchase, rather than being a supplementary or promotional element. This distinguishes MDs from other tactics where the mystery is peripheral or additive, such as receiving a surprise gift alongside a known product (Guo et al. 2024). A clear classification and differentiation of mystery deals from other mystery marketing tactics can be seen in Figure 1. This is an open access article under the terms of the Creative Commons Attribution License, which permits use, distribution and reproduction in any medium, provided the original work is properly cited. © 2025 The Author(s). Psychology & Marketing published by Wiley Periodicals LLC. 2883Psychology & Marketing, 2025; 42:2883–2901 https://doi.org/10.1002/mar.70017 A defining characteristic of MDs is the intentional creation of an information gap, which consumers are motivated to bridge by completing the purchase. Unlike traditional deals (TDs), where all details are known upfront, MDs involve uncertainty and elevate expectations while introducing risk, as buyers cannot be certain they will receive their preferred option. In Burger King's deal, for example, customers had a one‐in‐eight probability of obtaining their preferred burger. Despite their growing prevalence, little is known about how consumers evaluate MDs after purchase and what psychological and behavioral mechanisms shape their final judgment. Surprisingly, most research on MDs focuses on pre‐purchase behavior, leaving post‐purchase reactions largely unexplored. Studies suggest that mystery purchases are especially appealing when potential outcomes vary in attributes such as taste or style, rather than objective value (Buechel and Li 2023), and in experiential contexts (Urumutta Hewage and He 2022). Demographic factors, such as gender, also influence consumers' openness to mystery formats (Kovacheva et al. 2022). Studies have also identified several mechanisms by which MDs enhance purchase motivation, including stimulating curiosity (Hill et al. 2016), fostering innate optimism (Goldsmith and Amir 2010), triggering a desire for surprise, and eliciting positive affect (Buechel and Li 2023; Laran and Tsiros 2013). Research in operational and strategic management further highlights how retailers benefit from MDs in the short term, particularly through price discrimination and inventory management (e.g., Fay and Xie 2008). However, little is known about how consumers react in the post‐purchase phase, particularly when they receive an outcome that does not align with their preferences (for a summary of existing research, see Table A.2 in Appendix A). This study seeks to address this gap by examining the impact of MDs on customer loyalty, defined as a consumer's attitude and behavioral intention to maintain a relationship with a firm by making repeat purchases, either of the same or different products (Sirdeshmukh et al. 2002). We propose two key effects that influence customer loyalty in the post‐purchase phase. First, a (dis)confirmation effect that arises as consumers compare the actual outcome of the MD to their inflated expectations and preferred outcome. Second, we introduce the concept of positive affective spillover, wherein the positive emotions elicited during the pre‐purchase phase extend into the post‐purchase phase, influencing consumers' loyalty (Brown and Kirmani 1999; Zillmann 1971). These two effects likely interact, and the result of their combination has not been studied, raising the question of their overall impact on customer loyalty. Our research pursues three primary objectives: (1) to examine consumer reactions in the post‐purchase phase when the mystery is revealed; (2) to test the interplay of the (dis)confirmation effect and positive affective spillover effect; and (3) to assess the overall impact of MDs on customer loyalty, especially when consumers do not receive their preferred outcome. To achieve these goals, we present six studies: three online experiments, one laboratory experiment, one field study, and one supplementary study. To our knowledge, this is the first study to systematically investigate how mystery deals—where the uncertainty pertains to the core component of the purchase rather than an ancillary benefit—affect customer loyalty in the post‐purchase phase. By moving beyond the commonly examined domain of pre‐ purchase excitement and curiosity, our research shifts the analytical focus to the long‐term consequences of mystery deals, specifically their impact on sustained customer engagement and brand loyalty. We establish that post‐purchase reactions are shaped by (1) the (dis)confirmation effect, where loyalty strengthens when expectations are met and remains stable even when they are not (and surprisingly, disconfirmation effects have not been studied in the context of MDs), (2) positive FIGURE 1 | Classification of and differentiation from varying mystery marketing tactics. The aim of this figure is not to provide a comprehensive review of the literature, but rather to offer an overview of different branches and types of mystery marketing tactics and to position the present research within this context. (Alavi et al. 2015; Dhar et al. 1999; Elsen et al. 2016; Gupta et al. 2020; Mazar et al. 2017; Schroll and Grohs 2019; Sevilla and Meyer 2020). 2884 Psychology & Marketing, 2025 affective spillover, which enhances emotional responses and mitigates negative effects of disconfirmation, and (3) inflated expectations, which make disconfirmation likely but do not necessarily harm loyalty. This advances MD research by demonstrating how resolving uncertainty impacts retention—an aspect overlooked in prior research focused on pre‐purchase responses (Buechel and Li 2023; Goldsmith and Amir 2010; Hill et al. 2016). Furthermore, our findings also contribute to research on affect by showing that pre‐purchase emotions persist into the post‐purchase phase, shaping satisfaction and loyalty (e.g., Lee and Qiu 2009). Beyond theoretical contributions, our findings provide actionable managerial insights: MDs not only drive attention and sales but also foster loyalty— especially when designed to maximize positive pre‐purchase affect. Retailers can boost engagement through visually appealing deals, curiosity‐driven messaging, and exclusive mystery incentives, making MDs a strategic alternative to traditional offers across various industries and retail contexts. 2 | Theoretical Background 2.1 | How Mystery Deals Work—Creating and Resolving Information Gaps In general, MDs consist of a creation phase before purchase and a resolution phase afterwards. In the mystery creation phase, the consumer is confronted with a deal that involves deliberately created uncertainty or an information gap about the final outcome. In the case of the MD type we investigate, the exact product variant remains unknown at the time of purchase. Such information gaps, defined as the difference between what one knows and what one wants to know, induce the perception of information deprivation and curiosity (Loewenstein 1994). Consequently, this motivates consumers to close the information gap and pursue exploratory behavior to acquire the missing information (Hsee and Ruan 2016; Loewenstein 1994).Thereisconsensusintheliteraturethatsuch forms of positive uncertainty (i.e., which involves uncertainty about non‐negative outcomes only; Ruan et al. 2018)canstimulatepurchases and improve consumers' attitudes and affective responses (Lee and Qiu 2009;Ruanetal.2018). In the post‐purchase phase, the information gap induced by the MD is resolved by revealing the specific outcome. Unlike mystery advertisements, MDs require consumers to make a purchase—and invest money—to close the information gap. After the product is revealed, consumers will evaluate their individual utility of the outcome and will develop some level of satisfaction with the purchase and future intentions. In addition, the resolution of curiosity and uncertainty elicits affective responses that will also influence the outcome evaluations. In sum, we argue that creating and resolving a mystery for the consumer, through the MDs as defined earlier, leads to a complex interplay of effects that determine the outcome evaluation. The individual components of this overall effect are discussed in the following paragraphs. 2.2 | The Role of Inflated Expectations In the pre‐purchase phase, consumers will develop expectations about the outcome. However, these expectations will be inflated in thewaythattheyaremorepositivethanifnomysterywouldhave been involved, which can be attributed to two aspects. First, curiosity enhances expectations. It induces the anticipation of rewards in individuals, which leads them to expect to receive something valuable for their efforts undertaken to resolve it (e.g., Wiggin et al. 2019). In line with this, it has been reported that when consumers are in a state of curiosity, they tend to develop overly positive expectations about the outcome (e.g., Daume and Hüttl‐ Maack 2020). The second aspect is known under the term innate optimism (Bar‐hillel and Budescu 1995;GoldsmithandAmir2010). When consumers are confronted with uncertain alternatives, they overestimate the probability of positive as opposed to negative outcomes. In the context of mystery gifts, Goldsmith and Amir (2010) have even shown that consumers evaluate an uncertain outcome as valuable as the best possible outcome under certainty. In the same vein, we expect consumers to be overly optimistic when confronted with a mystery deal providing a set of equally likely (uncertain) product variants. Because of the uncertain outcome of the deal, consumers will form high expectations and expect their most preferred option. They will expect this best possible outcome eveniftheactualchanceofobtaining this outcome is as high as the chance to get one of the other options. We therefore assume: H1. Mystery deals will generate inflated expectations in consumers, such that the actual share of consumers expecting to receive the best possible outcome (i.e., their preferred option) will be higher than the predicted share of consumers that should receive it. 2.3 | Expectation‐(Dis)Confirmation Effects In the context of MDs, the expectation‐disconfirmation paradigm plays a crucial role due to the uncertain outcome. According to this paradigm, consumers evaluate the outcome of a purchase based on a comparison of their pre‐purchase expectations with the purchase outcome (Oliver 1980). The more a product meets consumers' initial expectations, the higher the customer satisfaction and loyalty (confirmation effect); the less it meets these expectations, the lower the customer satisfaction and loyalty (disconfirmation effect; e.g., Habel et al. 2016;Oliver1980). Given that in the case of MDs, expectations are inflated, as explained before, the risk of creating negative disconfirmation is extraordinarily high. Furthermore, the affectivereactionscausedbythedegreeofdisconfirmationplaya crucial role; it decreases the pleasure of a consumption situation, whereas confirmation increases it (e.g., Oliver 1980), which further influences consumers' satisfaction and loyalty intentions (e.g., Westbrook 1987). Immediately after the purchase, customers compare the outcome of the MD with their preferred alternative (Shen et al. 2019). We expect that consumers who receive a non‐ preferred outcome will experience a disconfirmation such that their positive affective reactions and loyalty intentions are lower than those of consumers getting a TD, in which they know from the beginning what they will receive. Furthermore, consumers who receive their preferred option when the MD is revealed will experience a confirmation that should be even higher than for consumers purchasing a TD. Receiving the desired, yet uncertain outcome in an MD will lead to stronger positive affective responses, such as positive surprise (Oliver et al. 1997) 2885 and relief (Ruan et al. 2018; Yang et al. 2016), compared to a TD. Based on these considerations, we hypothesize: H2. There will be a (dis)confirmation effect in the post‐ purchase phase such that MDs (dis)confirming consumers' preferences will decrease (vs. increase) loyalty through lower (vs. higher) levels of positive post‐purchase affective reactions compared to TDs. 2.4 | Pre‐Purchase Positive Affect and Positive Affective Spillover Effects An additional aspect, unique to MDs, that influences outcome evaluation is the positive affective response generated during the pre‐purchase phase, which can persist and continue to influence theconsumerexperiencewellintothepost‐purchase phase. Studies on positive uncertainty and curiosity in contexts such as teasing and advertising have shown that strong positive feelings, created in the pre‐purchase phase, ultimately lead to positive consumer responses (e.g., Bar‐Anan et al. 2009;Hüttl‐Maack et al. 2024;LeeandQiu2009). Although negative affective responses of curiosity are also possible, such as feelings of frustration and discomfort associated with the experience of not knowing (Berlyne 1954;LitmanandJimerson2004), research has demonstrated that using mystery deals in marketing overall elicits more positive than negative affect (Hill et al. 2016). The result is a net positive hedonic experience (Ruan et al. 2018). Based on previous evidence in other domains (e.g., Brown and Kirmani 1999; Geers and Lassiter 2002;Zillmann1971), we argue that these positive pre‐purchase affective reactions will spill over to the post‐purchase phase and positively influence the consumer's final response towards the deal. For instance, Brown and Kirmani (1999) and Zillmann (1971) propose that affective reactions towards a stimulus influence (1) affective reactions to a subsequent stimulus and (2) consumers' cognitive judgments of this second stimulus. Similarly, the affective priming paradigm suggests that the valence of an affective state influences the valence experienced by a subsequent stimulus in a consistent manner (Goldberg and Gorn 1987). In addition, the affective expectation model (e.g., Geers and Lassiter 2002) indicates that people with positive affective expectations of a situation are likely to actually experience that later situation more positively. Based on this evidence, we assume that MDs (vs. TDs) will create higher positive pre‐purchase affect, which spills over to the post‐ purchase phase and positively influences consumers' positive post‐purchase affective reactions and loyalty by buffering any potential negative effects. In fact, we posit that positive affective spillover effects will outweigh the negative effects of disconfirmation and reinforce the positive effects in case of a confirmation. We therefore propose: H3. There will be positive affective spillover effects such that MDs (vs. TDs) increase positive pre‐purchase affect that (a) increases customer loyalty directly and (b) indirectly through an increase in positive post‐purchase affect. The interplay of the hypothesized effects are summarized in Figure 2. It illustrates how the two distinct sources of affect, the pre‐purchase state involving curiosity and anticipation, and the post‐purchase situation in which uncertainty is resolved, interact to shape the overall consumer response. In the following, we present a total of six studies, including three online scenario‐based experiments, one laboratory experiment, one field study, and one supplementary study. For an overview of all studies, see Table B.1 in Appendix B. 3 | Studies 1a and 1b: Examining the Proposed Effects In the first two studies, we investigate our proposed effects in a product‐related (Study 1a) and service‐related context (Study 1b). 3.1 | Study 1a: Product‐Related Context (Chocolate) 3.1.1 | Design, Participants, and Procedure We conducted a single‐factor (MD vs. TD) between‐subjects online experiment with pre‐and post‐purchase phases. A total FIGURE 2 | Conceptual Model. 2886 Psychology & Marketing, 2025 of 148 participants (50% women; M age = 46.32 years, SD age = 14.63 years) from a German professional panel, all interested in chocolate, took part. 3.1.1.1 | Pre‐Purchase Phase. Participants viewed a German fictitious chocolate advertisement featuring either a TD or an MD and were randomly assigned to one condition. The TD offered a10‐pack of one specific flavor for €9(discountedfrom€10), while the MD offered a 10‐pack of one of four possible flavors for €7, without revealing which. Flavors—milk chocolate, hazelnut, strawberry, and caramel—were chosen for their popularity (Nier 2019; Splendid Research 2017;seeAppendixC). Reflecting managerial practice and prior studies (e.g., Fay and Xie 2008; Granados et al. 2008), the MD was further discounted (€7vs.€10). Compared to real‐world cases (Table A.1 Appendix A), the MD discount was moderate for conservative testing. Participants viewed the advertisement for at least 15 s, rated their positive affective response using a 5‐point smiley scale by Shampanier et al. (2007), and indicated their expected flavor. They also rated their purchase intention on a 7‐point single‐ item scale (e.g., Laran and Tsiros 2013; see Appendix D, for constructs and items). Cronbach's alpha values for multi‐item constructs are also reported in Appendix D. Other constructs, mostly single‐item, were adapted from established literature to ensure reliability and theoretical fit. 3.1.1.2 | Post‐Purchase PhaseParticipants imagined receiving a 10‐pack of chocolate after accepting the deal. TD participants got the advertised flavor; MD participants were assigned either hazelnut (most preferred) or strawberry (least preferred) to confirm or disconfirm preferences. After viewing the flavor for 15 s, brand loyalty intentions were measured using a 4‐item scale (Chaudhuri and Holbrook 2001) and repurchase intentions with a single‐item scale (Habel et al. 2016;Oliver1980). Participants also rated post‐ purchase affect (Shampanier et al. 2007)andrankedflavorsto assess preference confirmation. All constructs and items are detailed in Appendix D. 1 3.1.2 | Results of Study 1a 3.1.2.1 | Inflated Expectations and Pre‐Purchase Affect. We first tested whether consumers had overly optimistic expectations in the MD condition. While a rational expectation would be 25% (with all four options equally likely), 86% of MD participants expected their preferred flavor. A chi‐ square test confirmed a significant difference (i.e., 25%; χ 2 (1, n= 75) = 149.15, p< 0.001), supporting H 1 that MDs lead to inflated expectations. 3.1.2.2 | (Dis)Confirmation Effects and Positive Affective Spillover Effects. We used Model 4 of the PROCESS macro (Hayes 2022) with 5000 bootstrapped samples and 95% bias‐corrected confidence intervals to test our mediation hypotheses. The MD sample was split into confirmed (MD con , n= 20) and disconfirmed (MD dis ,n= 55) groups based on whether participants received their preferred flavor. Then, two serial mediation models were estimated: one tested (dis)confirmation effects by contrasting MD con and MD dis with the TD; the other examined positive affective spillover by comparing MD groups with TD. We controlled for MD con versus MD dis differences in all post‐purchase measures to align path coefficients across models and included age and gender as covariates. In the first model, we found significant disconfirmation (a×b=−0.16, 95% CI [−0.2830, −0.0485]) and confirmation effects (axb= 0.18, 95% CI [0.0479, 0.3439]) on brand loyalty intentions. Compared to a TD, MD dis lowers and MD con raises brand loyalty intentions via post‐purchase affect. Similar effects were found for repurchase intentions—disconfirmation (a×b=−0.17, 95% CI [−0.3361, −0.0492]) and confirmation (a xb= 0.20, 95% CI [0.0648, 0.3663]). These results support H 2 . The second model showed positive affective spillover: MDs (vs. TDs) increased brand loyalty and repurchase intentions through higher pre‐purchase affect (MD vs. TD →pre‐purchase affect → brand loyalty, axb=0.12, 95% CI [0.0177, 0.2552]; MD vs. TD → pre‐purchase affect →repurchase intentions, axb=0.14, 95% CI [0.0339, 0.2847]) and through both pre‐and post‐purchase affect (MD vs. TD →pre‐purchase affect →post‐purchase affect →brand loyalty intentions, axb=0.13,95% CI [0.0361, 0.2371]; MD vs. TD →pre‐purchase affect →post‐purchase affect →repurchase intentions, axb= 0.13, 95% CI [0.0409, 0.2639]). These results support H 3a and H 3b . Key results are shown in Figure 3. 2 3.1.2.3 | Overall Effects on Customer Loyalty. The indirect effects show that positive affective spillover (1) fully offsets the negative impact of disconfirmation and (2) amplifies the positive impact of confirmation. As a result, total effects reveal no significant decrease in brand loyalty (b= 0.16, p= 0.37) or repurchase intentions (b= 0.20, p= 0.25) for MD dis versus TD. In contrast, MD con (vs. TD) led to significantly higher brand loyalty (b= 0.51, p< 0.05) and repurchase intentions (b= 0.59, p< 0.05). 3.1.2.4 | Supplementary Study. In a supplementary study (n= 213), we tested whether the effects held when MDs and TDs were priced equally at €7, despite past research suggesting MDs are typically cheaper (e.g., Granados et al. 2008). We found similar results and patterns in the case of the same prices for brand loyalty intentions and repurchase intentions toward the deal (see Appendix E). In addition, we applied an alternative 3‐item scale for overoptimistic expectations toward the MD outcome adapted from Daume and Hüttl‐Maack (2020) to measure inflated expectations. At the low price (€7), MDs triggered significantly higher expectations than TDs (M MD, low price (€7) = 3.53 vs. M TD, low price (€7) = 2.54; F(1, 222) = 14.69, p< 0.01). At the high price (€9), no difference was found (M MD, high price (€9) = 3.08 vs. M TD, high price (€9) = 2.81; F(1, 222) = 1.07, p= 0.30). Thus, these findings additionally support the rationale that not the price per se, but the mystery deal (at a specific discount rate) drives these effects. 3.2 | Study 1b: Service‐Related Context (Pasta Restaurant) 3.2.1 | Design, Participants, and Procedure Study 1b used a similar design and procedure to Study 1a. We recruited 248 participants from a German professional online 2887 panel (64% women; M age = 31.21 years, SD age = 13.56 years) for an online scenario experiment. In contrast to Study 1a, the deal featured a pasta meal in a restaurant, a typical service setting (see Appendix Cfor stimuli). 3.2.1.1 | Pre‐Purchase Phase. Participants were asked to imagine encountering a deal at a pasta restaurant. In the TD condition, a specific pasta meal was offered for €9 (discounted from €10). In the MD condition, a “surprise pasta”was offered for €7. Four possible meals were used: meat sauce (most preferred), fish (least preferred), carbonara, and gorgonzola, selected based on German consumer taste preferences (YouGov 2019). 3.2.1.2 | Post‐Purchase Phase. Participants were then asked to imagine accepting the deal. Those in the TD condition received the advertised pasta meal, while those in the MD condition were assigned either the most preferred meal (meat sauce) or the least preferred (fish), to either confirm or disconfirm their preferences. The same scales and measures from Study 1a were used in both phases. Additionally, participants were asked about their perceived disconfirmation to validate whether their outcomes were confirmed or disconfirmed (Oliver 1980). All measures are depicted in Appendix D. 3.2.2 | Results of Study 1b 3.2.2.1 | (Dis)Confirmation Effects and Positive Affective Spillover Effects. As in Study 1a, we split MD participants into confirmed (MD con ,n= 32) and disconfirmed (MD dis , n= 97) groups. A one‐sample t‐test showed that perceived disconfirmation was below the scale midpoint for MD con (M MDcon = 3.13;t(31) = −3.22, p< 0.01) and above the midpoint for MD dis (M MDdis = 4.96;t(96) = 6.39, p< 0.001), confirming that MD con participants perceived the outcome as better and MD dis participants as worse than expected. We followed the same procedure as in Study 1a, estimating two serial mediation models using PROCESS Model 4 (5000 bootstrapping samples, 95% bias‐corrected confidence interval; Hayes 2022). The first model revealed significant disconfirmation (axb=−0.45, 95% CI [−0.6055, −0.3009]) and confirmation effects (axb= 0.38, 95% CI [0.2372, 0.5486]) for brand loyalty, as well as for repurchase intentions FIGURE 3 | Results of Mediation Analyses of Study 1a. (a) Dependent variable: Brand loyalty intentions, (b) Dependent variable: Repurchase intentions of the deal. Note: *p< 0.05, **p< 0.01. AFF (t1) = positive pre‐purchase affective reactions, AFF (t2) = positive post‐purchase affective reactions, BLI = brand loyalty intentions, RI = repurchase intentions, X1 = MD con versus TD; X2 = MD dis versus TD. 2888 Psychology & Marketing, 2025 (disconfirmation effect: axb=−0.40, 95% CI [−0.5659, −0.2610]); confirmation effect: axb= 0.35, 95% CI [0.2149, 0.5002]). These results support H 2 . The second mediation model showed significant positive affective spillover effects. MDs (vs. TDs) positively influenced brand loyalty and repurchase intentions through higher pre‐ purchase affect (MD vs. TD →pre‐purchase affect →brand loyalty, axb= 0.09, 95% CI [0.0245, 0.1694]; MD vs. TD →pre‐ purchase affect →repurchase intentions, axb= 0.15, 95% CI [0.0693, 0.2550]) or through both pre‐and post‐purchase affect (MD vs. TD →pre‐purchase affect →post‐purchase affect → brand loyalty, axb= 0.12, 95% CI [0.0618, 0.1978]; MD vs. TD → pre‐purchase affect →post‐purchase affect →repurchase intentions, axb= 0.11, 95% CI [0.0532, 0.1808]). 4 Results are shown in Appendix F. 3.2.2.2 | Overall Effect on Customer Loyalty. Total effects showed that MD dis (vs. TD) did not significantly reduce brand loyalty intentions (b=−0.07, p= 0.59) or repurchase intentions (b=−0.12, p= 0.36), even though the disconfirmation effect was larger than the two affective spillover effects. This indicates that the spillover effects were strong enough to counterbalance the negative impact of disconfirmation. Moreover, brand loyalty and repurchase intentions were higher for MD con participants than for TD participants (brand loyalty: b= 0.62, p< 0.01; repurchase intentions: b= 0.50, p< 0.05). 3.3 | Discussion of Studies 1a and 1b The results of Studies 1a and 1b revealed two effects that influence the evaluation of MDs after the purchase. First, a (dis) confirmation effect in which customer loyalty depends on whether the outcome of the MD matches the customers' preferences (confirmation) or not (disconfirmation). In the case of (dis)confirmation, MDs lead to more (less) customer loyalty because of higher (lower) levels of positive post‐purchase affect. Second, positive affective spillover effects in which consumers' positive affective reactions to MDs in the pre‐purchase phase carry over to the post‐purchase phase and positively impact customer loyalty. Moreover, these spillover effects counteract negative effects of the disconfirmation and complement positive effects of the confirmation. Consequently, customer loyalty is at equal levels for disconfirmed MD and TD customers—which is important given the high likelihood of having disconfirmed customers through MDs due to inflated expectations. The effects hold in a service context where potential negative effects of a disconfirmation are particularly strong due to the need for immediate consumption. Importantly, these results remain consistent even when the prices of MDs and TDs are identical, as demonstrated in the supplementary study. 4 | Study 2: Elaborating on Affective Reactions with Facial Expression Data In Studies 1a and 1b, we demonstrated that consumers' affective reactions in both the pre‐and the post‐purchase phases play an important role in understanding customer loyalty in MDs. However, we focused on self‐reported affective reactions and did not capture consumers' spontaneous and unconscious responses. In Study 2, we used AI‐based facial recognition software to directly track consumers' affective responses to MDs in the pre‐purchase phase before and after the deal and in the post‐purchase phase immediately before and after resolving the mystery. 4.1 | Design, Participants, and Procedure We recruited 285 students from a German university for a laboratory experiment conducted in a specialized lab equipped with facial emotion recognition software. Participants received monetary compensation. A student sample was chosen due to the on‐campus nature and technical complexity of the lab setup, which required standardized, in‐person procedures. While student samples have known limitations, they are commonly used in theory‐driven experiments investigating psychological processes such as affective responses. The design mirrored Study 1a (Appendix C), with video recording added to capture facial expressions. After informed consent, we obtained usable data from 264 participants M age = 21.64 years, SD age = 2.50 years), analyzed via the TAWNY Emotions Analytics Platform (TAWNY n.d). This software is based on the established facial action coding system (FACS) by Ekman and Friesen (1978) used to identify the seven basic emotions. It reports second‐by‐ second index values ranging from zero to 100 for each emotion, indicating the probability that a person is expressing that emotion. The procedure followed Study 1a with three adjustments: (1) random assignment of participants to 15 workstations, (2) participants spent 60 s showing a neutral expression before starting, and (3) in the post‐purchase phase, participants had a chance to receive any flavor, with hazelnut or strawberry assigned 40% of the time and milk chocolate or caramel 10%. 4.2 | Measures We focused on the emotion of happiness, as it best represents positive affect in this study area (Ruan et al. 2018). We analyzed participants' happiness index in four phases: (1) the calibration phase, capturing natural expressions of happiness; (2) the creation phase, where participants encountered the TD or MD; (3) the pre‐resolution phase, measuring happiness just before opening the virtual parcel; and (4) the resolution phase, when the specific TD or MD was revealed. For the calibration phase, we used the full 60‐second timeframe. For the creation and resolution phases, we analyzed the first 10 s to capture quick, unconscious emotional reactions (Ekman 1992), avoiding cognitive influences. Our data supported this decision, with mean times to minimum and maximum happiness values within these phases being 2.22 (SD = 3.80) and 6.35 (SD = 7.69), respectively. Similarly, the mean time (seconds) to reach a minimum value in the resolution phase was 2.16 (SD = 3.41) and to reach a maximum value was 6.42 (SD = 6.33). For the pre‐resolution phase, we used a 33‐second timeframe, as it was 2889 the median time from when participants accepted the deal to when the product was revealed. We calculated mean happiness scores for each phase and deal type. In both the pre‐resolution and resolution phases, we split the MD group into MD con (confirmed) and MD dis (disconfirmed) subgroups, based on whether participants received their preferred alternative. 5 Covariates included workstation, session, and gender. 4.3 | Results of Study 2 Affective Reactions in the Pre‐Purchase Phase.Weused planned contrasts for our analysis and compared the happiness scores for each deal (n TD =132;n MD =132)inthecreationphase to the scores in the calibration phase to investigate whether the deal significantly increased participants' happiness. The MD significantly increased happiness scores in the creation phase compared to the calibration phase (M MD, calibration =1.04 vs. M MD, creation = 3.49; F(1, 262) = 8.07, p< 0.01). In contrast, the happiness scores were not significantly affected by the TD (M TD, calibration =1.93 vs.M TD, creation =3.27;F(1, 262) = 2.42, p=0.12). Affective Reactions in the Post‐Purchase Phase. We compared happiness scores across TD, MD con (n= 28) and MD dis (n= 104) groups before and after mystery resolution. MD con participants reported significantly higher post‐resolution happiness (M MDcon, pre‐resolution = 3.97 vs. M MDcon, resolution = 9.06; F(1, 261) = 4.72, p< 0.05). In contrast, MD dis scores were unaffected by the unfavorable outcome of the MD, (M MDdis, pre‐ resolution = 2.93 vs. M MDdis, resolution = 4.04; F(1, 261) = 0.83, p= 0.36). TD participants' scores remained stable across phases (M TD, pre‐resolution = 3.78 vs. M TD, resolution = 5.39; F(1, 261) = 2.22, p= 0.14). Figure 4summarizes our findings. 4.4 | Discussion of Study 2 The findings of Study 2 elaborated on consumers' positive affective responses to MDs in the creation and resolution phases. Using AI‐based emotion tracking, we observed how consumers naturally respond to MDs, confirming that positive affect increases during both phases. When an MD's outcome matches customer preferences, customers experience even higher positive affect post‐resolution. Moreover, the positive affect of those disconfirmed by the MD remains similar to (1) that of TD consumers and (2) to their pre‐resolution levels, suggesting that pre‐purchase happiness counteracts the negative effects of disconfirmation on post‐purchase happiness. 5 | Study 3: Robustness Checks Study 3 compared the MD condition with a different TD baseline version (TD choice ) to rule out alternative explanations for the higher positive affect in the MD pre‐purchase phase. It controlled for (a) the greater number of options in the MD condition and (b) the lack of choice in the original TD condition. In TD choice , participants saw all four flavors and chose one freely. The study also reduced demand effects by removing the MD prompt asking participants to state their expected flavor. 5.1 | Design, Participants, and Procedure We recruited 299 participants from a professional online panel representing the German population (50% women; M age = 44.71 years, SD age = 13.77 years) for an online scenario experiment. We used a single‐factor (MD vs. TD choice vs. TD) between‐subjects design. Our primary focus was the comparison between the MD and TD choice conditions, with the original TD condition included to test robustness. Pre‐Purchase Phase. Participants were randomly assigned to one of three conditions: MD, TD choice ,orTD non‐choice . TD conditions mirrored those from Study 1a. In the TD choice condition, participants saw all four chocolate flavors (like the MD condition) and were informed they could choose one flavor (see Appendix C). The same measures from Study 1a were used, but participants were not asked which flavor they expected to receive. Additionally, TD choice participants also selected one of the available flavors. Post‐Purchase Phase. As in Study 1a, participants were asked to imagine purchasing the deal and receiving a parcel with a 10‐pack of chocolate. In the MD condition, they were randomly given one of four flavors. In the TD conditions, participants received either the advertised flavor (TD non‐choice condition) or the flavor they had chosen (TD choice condition). The procedure and measures were identical to those in Study 1a (see Appendix D). FIGURE 4 | Results for the Happiness Index in the Different Phases of the Lab Experiment. 2890 Psychology & Marketing, 2025 TABLE B.1 | Study overview. Nr. Sample Study purpose Design Methodological approach 1a 148 Investigation of consumers’post‐purchase reactions toward the resolution of a mystery deal in a product context. Single‐factor between subjects Online‐scenario experiment 1a supplementary 213 Supplementary investigation of consumers’post‐purchase reactions when MD and TD are priced equally. Single‐factor between subjects Online‐scenario experiment 1b 248 Investigation of consumers’post‐purchase reactions toward the resolution of a mystery deal in a service context. Single‐factor between subjects Online‐scenario experiment 2 264 Investigation of consumers’natural and spontaneous reactions to mystery deals. Single‐factor between subjects using AI‐based facial recognition software Laboratory experiment 3 299 Elimination of the presence of a higher number of presented option as a cause for higher positive affective reactions to mystery deals in the pre‐purchase phase. Single‐factor between subjects Online‐scenario experiment 4 235 Addition of external validity to previous findings. Single‐factor between subjects Field study with online survey TABLE C.1 | Stimuli in the Pre‐Purchase phase in studies 1a, 1b, 2, and 3. Scenarios of Studies 1a and 2 (Product Context) TD MD Scenarios of Study 1b (Service Context) TD MD Scenarios of Study 3 TD TD choice MD Appendix B Appendix C (See Table C1) 2897 TABLE D.1 | Measurement items and validity assessment of constructs. Construct source αMeasurement items Pre‐Purchase Phase Purchase Intention e.g., Goldsmith and Amir (2010); Laran and Tsiros (2013) single item (used in Studies 1a, 1b and 3) How likely would you buy the deal you have just seen? Positive Pre‐Purchase Affective Reactions Shampanier et al. (2007) single item (used in Study 1a, 1b and 3) How do you feel with regard to [the ad]? Positive Expectations Adapted from Daume and Hüttl‐ Maack (2020) Study 1a:not used Study 1b: α= 0.79 Study 3: not used The ad I just saw… …made me expect something interesting. …triggers high expectations about what is advertised here. …triggers my hope to receive my mostly preferred flavor from [the promotion]. Uncertainty e.g., Lee and Qiu (2009) single item (used in Studies 1a, 1b, 2 and 3) I feel uncertain about which [product] I would receive within the previously seen ad, when I participate in the deal and buy the [product]. Post‐Purchase Phase Brand Loyalty Intentions Chaudhuri and Holbrook (2001) Study 1a: α= 0.88 Study 1b: α= 0.87 Study 3: α= 0.85 I will buy [this brand] the next time I buy [product name]. I intend to keep purchasing [this brand]. I am committed to [this brand]. I would be willing to pay a higher price for [this brand] over other brands. Repurchase Intentions of the Promotion Adapted from Oliver (1980); Habel et al. (2016) single item (used in Studies 1a, 1b and 3) How likely would you purchase [such a promotion] of [the brand] again? Positive Post‐Purchase Affective Reactions Shampanier et al. (2007) single item (used in Study 1a, 1b and 3) How do you feel with regard to [the received product/ service]? Satisfaction e.g., Habel et al. (2016) single item (used in Studies 1a, 1b and 3) Generally, I am satisfied with [the product] I received. Perceived Disconfirmation Oliver (1980) single item (used in Studies 1b and 2) Generally, [this product] is…(“better than expected”– “as expected”–“worse than expected”) Note: Positive affective reactions were measured with a 5‐point scale; all other constructs were measured with 7‐point scales. Appendix D (See Table D1) 2898 Psychology & Marketing, 2025 Appendix E (See Table E1) TABLE E.1 | Results of Supplementary Study with TD and MD at the Same Price (€7). MD (€7) versus TD (€7) Results Effects on Brand Loyalty Intentions Disconfirmation Effect MD dis versus TD →positive post‐purchase affect →brand loyalty intentions axb=−0.11, 95% CI [−0.2670, −0.0004] Confirmation Effect MD con versus TD →positive post‐purchase affect →brand loyalty intentions axb= 0.26, 95% CI [0.0496, 0.4985] Affective Spillover Effects MD versus TD →positive pre‐purchase affect →brand loyalty intentions axb= 0.09, 90% CI [0.0036, 0.1857] MD versus TD →positive pre‐purchase affect →positive post‐purchase affect →brand loyalty intentions axb= 0.02, 90% CI [0.0001, 0.0625] Total Effects MD dis versus TD: b= 0.17, p= 0.38 MD con versus TD: b= 0.74, p< 0.01 Effects on Repurchase Intentions of Promotion Disconfirmation Effect MD dis versus TD →positive post‐purchase affect →repurchase intentions axb=−0.21 95% CI [−0.4393, −0.0139] Confirmation Effect MD con versus TD →positive post‐purchase affect →repurchase intentions axb= 0.52, 95% CI [0.2971, 0.7838] Affective Spillover Effects MD versus TD →positive pre‐purchase affect →repurchase intentions axb= 0.08, 90% CI [0.0049, 0.1919] MD versus TD →positive pre‐purchase affect →positive post‐purchase affect →repurchase intentions axb= 0.05, 90% CI [0.0029, 0.1104] Total Effects MD dis versus TD: b=−0.01, p= 0.95 MD con versus TD: b= 0.81, p< 0.01 Note: We followed the same procedure and conducted the same analyses (PROCESS models; Hayes 2022) as in Study 1a; n= 110. 2899 FIGURE F.1 | Integrated Results from Mediation Analyses of Study 1b and Study 3. Note: † p< 0.10, *p<0.05, **p< 0.01, AFF (t1) = positive pre‐ purchase affective reactions; AFF (t2) = positive post‐purchase affective reactions, BLI = brand loyalty intentions, RI = repurchase intentions, X1 = MD con versus TD/TD choice ,X 2 =MD dis versus TD/TD choice . Appendix F (See Figure F1) 2900 Psychology & Marketing, 2025 FIGURE F.1 | (Continued) 2901