The effects of easy and difficult business relationship evaluations on purchase intentions

Purpose – This research aims to investigate the different impacts of easy (fluency) and difficult (disfluency) business relationship evaluations on future purchase intentions. In addition, it seeks to investigate the mediating role of confidence in these evaluations and present a new form of fluency manipulation performed directly by satisfaction evaluations. Design/methodology/approach – This research includes a first experiment with a 3 (fluency: easy vs difficult vs control) x 2 (hypothetical scenario of satisfaction vs dissatisfaction) factorial design, using 180 undergraduate students, and a second single factor experiment (fluency: easy vs difficult vs control) with 326 consumers evaluating their financial services provider. Both used a between-subject design with the individuals being randomly distributed between the scenarios. Findings – Dissatisfied consumers who perceived ease in evaluating commercial relationships have increased confidence in these evaluations, negatively influencing their purchase intentions. Meanwhile, satisfied consumers are generally overconfident, being equally affected by perceived difficulty or ease due to a direct bias from the evaluation, which increases purchase intentions. Originality/value – The results demonstrate that fluency (perceived ease or difficulty) depends on the valence of the evaluations, directly affecting (positive valence) or indirectly affecting (negative valence) subsequent decisions. We also present a new, faster, and more practical way to manipulate fluency. Furthermore, we raise some ethical questions as these effects result in biased decisions.


Introduction
Individuals usually seek to simplify decisions using "shortcuts" or heuristics that minimize cognitive efforts and avoid stress and strain in their decisions (Kahneman, 2011). However, these "shortcuts" usually produce some biases that influence attitudes and decisions. In relationship marketing, one of these biases has its origin in satisfaction evaluations, leading to a pattern of future expectations associated with higher repurchase rates (Oliver, 2010). Similarly, the development of loyalty can also be seen as a cognitive bias that directs attitudes and behaviors (Bloemer, Ruyter, & Peeters, 1998). The explanation for these biased behaviors is based on the positive memories experienced (Lazarus, 1991), but it is also necessary to highlight the influence of fluency (Schwarz, 2004) on the creation and reinforcement of these biases, intensifying the impact of intuitions and opinions (Aydin, 2016;Simons & Nelson, 2006. At one end of the fluency continuum, individuals perceive ease in cognitive processing, creating comfort and pleasure (Labroo & Pocheptsova, 2016;Nunes, Ordani, & Valsevia, 2015) that would lead to biases such as: intensifying the effects of what is being interpreted (Claypool, Mackie, & Garcia-Marques, 2015;Landwehr, Golla, & Reber, 2017); evaluating and judging something more quickly (Schwarz, 2004); a greater perception of veracity, familiarity, learning, intelligence, and value (Alter & Openhaimer, 2006Kornell, Rhodes, Castel, & Tauber, 2011;Miele & Molden, 2010;Openhaimer, 2004;Yang, Huang, & Shanks, 2018). In addition, as evaluations get more intuitive, faster, and more confident (Kahneman, 2011), subsequent judgments and decisions show more biases (Aydin, 2016;Simons & Nelson, 2006. Perceived difficulty can be observed at the other end of the fluency continuum (Sanchez & Jaeger, 2015), linked to cognitive discomfort and thoughts about risk. However, if individuals could be stimulated to deliberate or to use their self-control (Kahneman, 2011), they would not develop overconfidence and would avoid using simple opinions and prior and/or erroneous beliefs -or even xenophobic thoughts -that could bias their decisions (Alter, 2013;Hernandez & Preston, 2013;Ryffel & Wirth, 2018).
Therefore, if consumers experienced easy and difficult assessments and judgments, would it be possible to lessen the impact of positive or even negative business experiences just by making the consumer doubt the evaluation? Is confidence in evaluations and subsequent decisions differently affected by positive and negative commercial experiences? In order to answer these questions, this study aims to verify the impact of fluency on commercial relationship evaluations and on subsequent future purchase intentions. In addition, the study also presents a new fluency manipulation (based on Oppenheimer, 2004), not requiring previous tasks to affect later evaluations, but directly using easy or difficult satisfaction evaluations, and therefore enabling it to be applied in practice.

Literature Review
Despite the need and importance, in several situations consumers do not rationally and structurally judge the attributes, benefits, and costs of each purchase. Instead, they use a series of simplified rules (i.e. heuristics) without making in-depth comparisons and evaluations (Baron, 2008). These heuristics generate systematic errors, called biases, arising from factors such as lack of self-control and fluency (Kahneman, 2011). The latter is the subject of this study.

Effects of fluency on evaluations and judgments
Fluency can be understood as a continuum between ease (fluency) and difficulty (commonly called disfluency) of processing some type of information. The intensity of fluency can originate from the individual's own characteristics or from the situation being evaluated and it can be classified as perceptual or conceptual fluency (Schwarz, 2004). Perceptual fluency results from physicalvisual characteristics, such as changes in brightness, contrast, or even in the print quality of texts. Conceptual fluency results from ease of processing a given stimulus, such as easy to pronounce names or even rhyming phrases. As an example, studies show that easy to pronounce stock names or symbols, therefore having greater fluency, create greater expectations of profitability for investors (Alter & Oppenheimer, 2006).
Stimuli perceived as easy have a greater tendency to be hedonically processed (Duke, Fiacconi, & Kohler, 2014). Thus, in general, people prefer music (rhythms and their lyrics), texts, and the like with simpler content, as they get cognitive comfort from the pleasure of being able to understand it without much effort (Bayliss, Constable, Tipper, & Kritikos, 2013;Maier & Dost, Juan José Camou Viacava / Paulo de Paula Baptista 2018;Nunes, et al., 2015). Ease of processing is linked to perceived pleasure, reducing uncertainties and eliciting more positive responses when evaluating products (Labroo & Pocheptsova, 2016). This cognitive comfort can even increase the acceptance of an idea, topic, or even a product (Labroo, Dhar, & Schwarz, 2008). However, pleasure is not felt when processing all stimuli types, only positive ones (Albrecht & Carbon, 2014). That is, there is no preference or pleasure in processing negative stimuli. In this sense, extreme ease of processing intensifies the effects of previously processed information, whether positive or negative, stimulating biases that influence future decisions (Claypool et al., 2015;Landwehr et al., 2017).
Heuristics and their biases are associated with System 1, where the process of seeking cognitive comfort leads to less and faster deliberation, producing overconfidence. On the other hand, in the case of System 2, there is slow thinking, more deliberation, and greater attention (Kahneman, 2011). Likewise, individuals who have previously had easy cognitive processing experiences overestimate their knowledge about these experiences, generating greater confidence (Ryffel & Wirth, 2018). Ease of processing can induce a greater perception of veracity (Silva, Garcia-Marques, & Reber, 2017) and overconfidence, because individuals do not have to deal with inconsistencies or new interpretations in deliberations (Bajšanski, Žauhar, & Valerjev, 2019). Similarly, studies testing the effects of technical versus common words (or even easy versus difficult to read typefaces) point out that perceived ease is related to a greater perception of information confidence, while difficulty leads to a greater risk perception (Park, Herr, & Kim, 2016).
Therefore, easily formed intuitions and opinions not only lead to results with greater perceived reliability, but also to greater confidence in the evaluation or judgment itself. Overconfident individuals do not engage in new information processing and are directly biased by these intuitions. However, if they have faced difficulties in forming these intuitions, this would signal to them that less intuitive alternatives should be considered, or even chosen (Simons & Nelson, 2006. As a result, individuals who easily formed prior choices of which product to buy would have greater confidence that these prior choices would be right, increasing subsequent purchase intentions (Aydin, 2016).
However, evidence indicates that this would not always happen because not only perceived ease or difficulty can affect confidence, but also mood states (Koch & Forgas, 2012) and the evaluated valence of memories (Kensinger & Schacter, 2008).

Effects of the valence of memories on evaluations and judgments
Positive mood states lead to judgments that are perceived as "more true," causing the maintenance of confidence, while negative mood states lead to less confidence in information processing (Koch & Forgas, 2012).
Confidence in what is being evaluated is negatively affected by negative recovered memories, involving more effort and need for control to avoid unwanted results (Caplan, Sommer, Madan, & Fujiwara, 2019). Negative emotions induce a more concrete adaptive processing state, leading individuals to process information with greater attention and for longer periods (Matovic, Kock, & Forgas, 2014). These effects relate to evidence that negative events are more accessible to memory, enhancing the recollection of details and central facts (Hostler, Wood, & Armitage, 2018;Kensinger & Schacter, 2008) that are essential and intrinsic to the events (Bookbinder & Brainerd, 2017;Kensinger, 2009) because they include more distinct memories (Brewin & Langley, 2019).
Negative emotions stimulate the ability to recapitulate episodes not only due to greater attention to coding and memories retrieval, but also due to the greater reactivation of these emotions (Bowen, Kark, & Kensinger, 2017). Thus, negative aspects, facts, and emotions result in greater attention and prevention in general (Prato & John, 1991), as well as a greater problem-solving focus (Orita & Hattori, 2018).
In contrast, positive memories lead to more general and peripheral attention (Talarico, Berntsen, & The effects of easy and difficult business relationship evaluations on purchase intentions Rubin, 2009). These memories need more associations to be remembered (Madan, Scott, & Kensinger, 2019) and they therefore have more inconsistencies (recall errors). The recovery of positive memories seems to be associated with greater confidence in their reports (Kensinger, 2009). Positive emotions lead individuals to focus on more abstract aspects, with less focus on and less attention paid to external information, seeking to maintain the pleasant feelings (Bless & Fiedler, 2006). This greater confidence also relates to the way people have been treated (Lazarus, 1991). People treated in a welcoming way, who had positive experiences, would feel safer and more "loose," reducing their worries and their need to not make a fool of themselves (p. 421). These positive mood states lead to the relaxation of inhibitory control, resulting in a fundamental shift in attention (Rowe, Hirsh, & Anderson, 2007).
In general, situations with negative affect lead to memories with greater precision and focus than positive affect (Kensinger, 2009), and similarly, positive memories lead to less details and accuracy compared to negative memories (Bohn & Berntsen, 2007;Brewin & Langley, 2018). With this lack of focus, in time individuals have a higher probability of remembering only the "essence" of the positive event -they would only "know" that something positive had happened, without any great precision or even forgetting details (Kensinger, 2009). Additionally, judgments focused on positive aspects are faster and more spontaneous than judgments focused on negative aspects (Herr & Page, 2004). Positive evaluations are relatively more automatic, leading to lower quality and more biased decisions (Herr, Page, Pfeiffer, & Davis, 2012). Also, Sorbeck and Clore (2005) indicate that positive aspects lead to a greater disconnection between confidence and the vividness of memories. That is, even if positive memories are not as distinct and vivid in certain situations, they still lead to greater confidence. Thus, while negative moods lead people to focus on details, positive moods make them deliberate more, based more on schematics or heuristics (Rowe et al., 2007).

Fluency and the effects of the valence of memories on commercial relationship evaluations
Among some of the "shortcuts" used for decision making, individuals commonly use prices (Luppe & Angelo, 2010) and previous satisfaction evaluations -their own or from others -as anchors in buying decisions (Oliver, 2010). Such behavior is related to individuals' tendency to reuse previous methods that have obtained satisfactory results (Baron, 2008). Consumers' past purchase satisfaction (Oliver, 2010) or even information from others' satisfactory experiences (Agnihotri, Dingus, Hu, & Krush, 2016) create positive expectations for the decision maker because they develops a belief that their needs can be fulfilled with similar performance in the future. This creates an expectation of satisfaction (Sirgy, 1984;Oliver, 2010) and the anticipation of positive emotions (Patrick, Chun, & Macinnis, 2009). In this sense, satisfaction evaluations made before purchasing decisions serve as an anchor for substitution heuristics and thus, as opinions or intuitions, they will mostly be used, replacing complex questions that would bias subsequent decisions (Kahneman, 2011).
Satisfaction itself is an overall evaluation of expectations, benefits, value, and quality compared to the perceived costs of business relationships. It is an evaluation of the performance of these factors compared to expectations. Positive disconfirmation of expectations would lead the consumer to be satisfied and feel positive emotions and affect. Negative disconfirmation leads the consumer to be unsatisfied, feeling negative emotions and affect (Oliver, 2010). Commercial interactions with positive evaluations develop trust, commitment, and true loyalty (Grönroos, 2009;Larivière et al., 2016) -or even a bias that can influence attitudes and behaviors (Bloemer et al., 1998) -resulting in better financial results (Francisco-Mafezzolli & Prado, 2013;Fornell, Morgeson, & Hult, 2016).
As dissatisfied consumers have negative memories, emotions, and affections associated with their suppliers, they have more vivid and distinct memories (Brewin & Langley, 2019). Thus, decisions regarding these suppliers would be more focused on central aspects (Kensinger, 2009), that is, with attention paid to essential facts and intrinsic details (Bookbinder & Brainerd, 2017;Kensinger, 2009) for problem solving (Orita & Hattori, 2018). Remembering their negative episodes, these consumers focus on their negative memories. As they perceive higher fluency (perceived ease) in their satisfaction evaluations, they develop greater confidence in their own evaluations (Simons & Nelson;Aydin, 2016), intensifying the negative aspects (Claypool, et al., 2015;Landwehr, et al., 2017) of the commercial relationship, thus decreasing the purchase intentions for that supplier. However, even if they have less confidence because of perceived difficulty, this will only make them more vigilant and assume a general state of prevention, focusing on the problem (purchase decision) and beginning to doubt their intuitions (in that they should not buy from the supplier). Therefore, we propose that: H1: Fluency (perceived ease) will increase a dissatisfied consumer's confidence in their relationship evaluations, subsequently reducing their purchase intentions.
Satisfied consumers have positive memories, emotions, and affections associated with their commercial products/services suppliers and, therefore, they make more abstract decisions, more focused on more general aspects (Talarico et al. 2009) and less so on external information (Bless & Fiedler, 2006). In this case, they feel safer and more "relaxed," with fewer worries (Lazarus, 1991), less attention paid, and less self-control (Rowe et al. 2007). Satisfied consumers make more automatic judgments and thus more biased ones. Even if they cannot recall as many details, they have greater confidence in their evaluations (Kensinger, 2009;Sorbeck & Clore, 2005) and are more influenced by heuristics (Rowe et al. 2007). Therefore, even in disfluency situations (perceived difficulty) satisfied consumers do not experience a significant drop in confidence in their ability to evaluate their satisfaction. Therefore, if they have remembered positive facts/memories before a purchase decision, their interpretations and evaluation results will still be intensified (Claypool et al. 2015;Landwehr et al. 2017), despite the difficulty perceived. Therefore, we propose that: H2: Commercial relationship evaluations (fluency or disfluency) will directly bias satisfied consumers to increase their subsequent purchase intentions.

Methodological Procedures
We used an experimental methodology to test the hypotheses (Figure 1), manipulating the independent variables and thus increasing the probability of an effect occurring (Shadish, Cook, & Campbell, 2002), measuring the manipulation (checking) and its effects on the dependent variables while at the same time controlling the other external variables (Malhotra, 2010). We used non-probabilistic sampling for convenience in both experiments. The first experiment used a sample of undergraduate students while the second one used a general sample of several bank customers contacted through social networks and via e-mail.
In both experiments, a new form of fluency manipulation was used based on Oppenheimer's (2004) manipulation, which manipulated fluency with texts using difficult words.
For the "easy" task (perceived ease), the individuals answered four overall satisfaction questions (Prado, 2004) before answering a purchase intention question. The four satisfaction questions to evaluate commercial relationships were: three questions adapted from Bettencourt (1997) and Fornell, Johnson, Anderson, Cha, & Bryant (1994) (overall satisfaction, expectations distance, and global disconfirmation) and one affective evaluation question from Garbarino & Jonhson (1999). For the "difficult" task (perceived difficulty), instead of directly asking these four questions, the concept of each item (taken from Prado, 2004) was written in a question format and these four conceptual questions were asked before the individuals indicated their purchase intention. As an example, the first question from Prado (2004), "How would you rate your satisfaction regarding your relationship with ALPHA, the provider of residential telephony, internet, and cable TV services?" was adapted in the "difficult" task scenario to "Overall, how would you evaluate your commercial relationship with ALPHA, the provider of residential telephony, internet, and cable TV servicesfocusing on the services offered regarding the act or effect of satisfaction?" For the "control" group, the individuals indicated their purchase intentions and later completed the four-item scale from Prado (2004).
In both experiments, all groups (set of "easy" vs "difficult" commercial relationship evaluations before indicating the purchase intention, vs Control -without commercial relationship evaluations before indicating the purchase intention) answered the four standard satisfaction questions (Prado, 2004) at the end of the questionnaire to check manipulation (experiment 1) and unbiased evaluations of business relationships (experiment 2). Moreover, to verify that these manipulations not only resulted in greater difficulty, but were also convergent with the results of other studies, the results of the satisfaction evaluations were compared with Haddock (2002) and other authors. Haddock (2002) demonstrated that, in a task involving remembering one politician's positive aspects, individuals in easy tasks (remembering few facts) developed more favorable opinions about the politician than those in difficult tasks (remembering more facts). Likewise, in tasks involving remembering one politician's negative aspects, individuals in easy tasks (remembering few The effects of easy and difficult business relationship evaluations on purchase intentions facts) developed less favorable opinions about the politician than those in difficult tasks (remembering more facts). Similar results were found when investigating personal use of bicycles (Aarts & Dijksterhuis, 1999), aspects of childhood (Winkielman & Schwarz, 2001), and satisfaction evaluations (Viacava, Mantovani, Korelo, & Prado, 2015). Thus, we also expect unsatisfied consumers with perceived difficulty in commercial relationship evaluations to have higher satisfaction scores than those with evaluations they perceived as easy. Similarly, we expect satisfied consumers with perceived difficulty in commercial relationship evaluations to have lower satisfaction scores than those with evaluations they perceived as easy.

Experiment 1
Procedures: the experiment presented a 3 (fluency: "easy" vs "difficult" vs control) x 2 (positive scenario: "satisfaction" vs negative: "dissatisfaction") factorial design. A total of 189 undergraduate students started the test but only 180 completed the survey and were therefore considered as valid (52.20% men, M age = 22.14 years, SD = 4.36).
In a classroom at the university, the students randomly assigned to one of six conditions filled in a survey (Appendix A -first experiment questionnaires) about commercial relationships (cable TV, home telephone, and internet providers). The students were given one of two scenarios ("satisfaction" vs "dissatisfaction") consisting of fictional stories about a consumer named Eduardo. In the "positive" scenario, the customer had a threeyear relationship that resulted in positive expectancy disconfirmation (general and attendance service quality, deadlines fulfilled, perceived cordiality, etc.) and the development of positive overall satisfaction. In the "negative" scenario, "Eduardo" also had a three-year relationship but the same situations resulted in negative expectancy disconfirmation and the development of negative overall satisfaction ("dissatisfaction"). After reading the scenarios, they answered two other unrelated surveys ("fillers") averaging five minutes, so that the scenarios were not so evident and close to the evaluations phase (Su, Li, Zheng, Hu, Fan, & Luo, 2018).
Next, fluency ("easy" vs "difficult" vs "control") was manipulated by the complexity of the satisfaction evaluations. For the "easy" task (perceived ease), the individuals answered four overall satisfaction questions (Prado, 2004) before indicating their purchase intentions. For the "difficult" task (perceived difficulty) they answered four adapted questions made for each concept item (taken from Prado, 2004) before indicating their purchase intention. The control group only indicated their purchase intention and, later, manipulation of the scenarios was checked (Figure 2 -experiment 1 flow).
As a dependent variable, all answered the purchase intention question (regarding a cable TV offer made by the same company described in the scenario, based on 1 -certainly NOT to 10 -definitely YES). Then

Figure 1. Conceptual Framework for the Hypotheses
Juan José Camou Viacava / Paulo de Paula Baptista they answered the fluency manipulation check, which questioned the general perceived difficulty (Graf, Mayer, & Landwehr, 2018) to evaluate satisfaction (1 -very difficult to 10 -very easy). The students also reported their actual satisfaction with their own internet, cable, and phone provider services and how much they paid for them, followed by socio-demographic questions (these variables reveal non-significant relationships with the results). Before ending the experiment, all answered a four-item overall satisfaction scale (Prado, 2004) as a scenario manipulation check.
Discussion: Although the two hypotheses were initially corroborated, some limitations arose. First, there was a small, though significant difference in perceived The effects of easy and difficult business relationship evaluations on purchase intentions difficulty between the fluency groups ("easy" vs "difficult"). Next, confidence in the evaluations was not measured (Aydin, 2016;Simmons & Nelson 2006, so we can only deduce -not state -that individuals reinforce (or abandon) their intuitions/anchors before deciding on their purchase intentions. Last, and yet more important, business relationships can have more events (positive and/ or negative ones) in their lifetimes (Oliver, 2010) and so the scenario manipulation may not fully incorporate this temporal view that results in overall satisfaction (Grönroos, 2009;Oliver, 2010).

Experiment 2
The objectives of experiment 2 were to check for congruence with the results of experiment 1, to resolve its limitations and to test the mediation effect of confidence. In the second experiment, the respondents' actual business relationships were assessed through an online survey (Appendix B -second experiment questionnaire, disfluency/perceived difficulty). In addition, the questions in the "difficult" manipulation were printed in bold letters to increase the perceived difficulty (Diemand-Yauman et al., 2011).
Procedures: The experiment had a 3 (fluency: "easy" vs "difficult" vs control) factor design. We opted not to manipulate positive or negative commercial relationships but instead to use satisfaction evaluations of the respondents' present financial service provider. A total of 361 undergraduate students started the test but only 326 completed the survey and were therefore considered valid (43.86% men, M age = 29.32 years, SD = 8.72).
The individuals were invited by email and social media to participate in a Qualtrics survey about financial services companies. They started by answering sociodemographic questions and then were randomly assigned to one of the three conditions. Similarly to in experiment 1, in the "easy" task (perceived as easy), the individuals answered four overall satisfaction questions (Prado, 2004) before indicating their purchase intentions. For the "difficult" task (perceived as difficult) they answered four adapted questions made for each concept item (taken from Prado, 2004) -this time printed in bold letters -before indicating their purchase intentions. The control group first indicated their purchase intentions and then evaluated their commercial relationships. The purchase intention question was similar to the one used in experiment 1, but regarding an automobile insurance offer from their present financial services provider. Next, the respondents answered the fluency manipulation check (general perceived difficulty) and reported their confidence in their commercial relationship evaluations, followed by financial services questions regarding their use and buying frequency (these variables reveal non-significant relationships with the results). Following the same procedure Juan José Camou Viacava / Paulo de Paula Baptista from experiment 1, before ending the experiment, all the respondents answered a four-item overall satisfaction scale (Prado, 2004) regarding their present financial services provider (Figure 4 -experiment 2 flow).
Next, we tested whether the confidence in the commercial relationship evaluations could be mediating the effects on the consumers' purchase intentions. First, we classified the individuals into groups according to the satisfaction averages (Cronbach's alpha = 0.790) at the end of the questionnaires. Using Hart and Johnson's (1999) results, the individuals were grouped into: dissatisfied customers (average from 1 to 5.5, n = 89) and satisfied customers (6.5 to 10, n = 93) c excluding neutral-satisfied customers from the analyses (5.51 to 6.49, n = 39).
In contrast, the mediation analyses for dissatisfied customers revealed that their purchase intentions were only affected by evaluation confidence (coef =-0.3161, t = -2.7233, p = 0.0074, LLCI = -0.5459, ULCI = -0.0863). Dissatisfied customers with easy questions before the dependent variable had their confidence increased and then Discussion: the results from experiment 2, with respondents evaluating their business relationships with financial services companies, converged with the results of experiment 1, with undergraduate students evaluating scenarios (a home telephone, internet, and cable TV company). Both hypotheses were corroborated and it was further demonstrated that dissatisfied customers, despite experiencing a significant decrease in their evaluation confidence in difficult tasks, did not increase their purchase intentions compared to the control group. This is an important result and can only be verified using a control group, which is not pointed out in other studies, as they usually only make the comparison between fluency (perceived ease) and disfluency (perceived difficulty) tasks (Aydin, 2016;Kahneman, 2011;Schwarz, 2004;Simmons & Nelson, 2006. Furthermore, although other studies point to higher confidence in tasks perceived as easy, and less confidence in difficult tasks (Aydin, 2016;Epley & Norwick, 2006, Simmons & Nelson, 2006Kahneman, 2011), this experiment demonstrated that satisfied consumers maintain their overconfidence in their commercial relationship evaluations even in more difficult judgment tasks.

Theoretical implications
This study demonstrated the implications of the ease and difficulty of satisfaction evaluations on subsequent purchase intentions. It also demonstrated that purchase intention results were influenced by two different mechanisms. Dissatisfied customers with a perception of ease in commercial relationship evaluations developed higher overconfidence, lowering their purchase intentions and thus corroborating the indirect effect hypothesis. Otherwise, satisfied customers with either easy or difficult evaluations showed overconfidence, therefore being directly affected by their evaluations with positive results.
Moreover, the study demonstrated congruent results with other fluency manipulations, but it corroborated the hypotheses by not only comparing difficult versus easy tasks (as in Aydin, 2016;Simons & Nelson 2006;among others), but also in comparisons with a control group. Fluency was directly manipulated conceptually by overall satisfaction evaluations, showing the same pattern of results as other studies. However, these other studies' manipulations usually took longer or needed more effort to be concluded -such as in a fact-remembering task (easy: 6 facts vs difficult: 12 facts) (Aarts & Dijksterhuis, 1999;Haddock, 2002;Viacava et al., 2015). Furthermore, other fluency manipulations, even more subtle ones, can be tested and can lead to similar results. It is possible to verify these effects by manipulating the perceived difficulty with very small letters (Su et al., 2018), using green, yellow, or light blue fonts on a white background (Reber & Schwarz, 1999), with Brusch or Mistral typefaces (versus Arial) (Song & Schwarz, 2008), with italics (Diemand-Yauman et al., 2011), or by inserting more technical or low-knowledge keywords (Oppenheimer, 2004).
Some attention must be paid when trying to use these manipulations, altogether or with the intention of intensifying the effects. When the difficulty is set too high, individuals will stop thinking that "they" have difficulty completing the task, and will start thinking that "the task itself " is too difficult, which can lessen the effects (Schwarz, 2004). In addition, making tasks too difficult and asking individuals to keep doing them could lead to ego depletion effects. Ego depletion is the momentary loss of cognitive capacity (depletion of cognitive resources) due to great effort and/or longer tasks. In ego depletion manipulations individuals are required to suppress their thoughts, inhibit their emotions, act against their principles, avoid reading words in a video, or try to solve great effort anagram tasks (or even unsolvable ones), which can also lead to greater purchase intentions (Baumeister & Tierney, 2011;Usta & Häubl, 2011).

Managerial implications
Despite the experiments' results involving a small sample, we were able to verify that the perceived difficulty in satisfaction evaluations led to greater subsequent purchase intentions. If difficult tasks were used in practice (such as here or in other manipulations), satisfied consumers would not have their purchasing decisions negatively affected, while dissatisfied consumers would at least start to deliberate, and thus consider a less intuitive option (Epley & Norwick, 2006;Simons & Nelson 2006. However, here some ethical dilemmas begin. Although The effects of easy and difficult business relationship evaluations on purchase intentions these tasks could not deplete all the consumers' cognitive resources (Baumeister & Tierney, 2011) they do affect their ability to make decisions. These tasks do affect consumers' confidence and in doing so they affect their opinions, intuitions, evaluations, and judgments. It is also important to emphasize that although individuals believe they are not being affected by biases (Santos & Barros, 2011) these effects are present in the daily lives of consumers and decision makers (Kahneman, 2011). Hence, they can make decisions against their intuitions and even against their initial beliefs, with non-beneficial results for them. Nevertheless, we found that the main effect on purchase intentions was due to their commercial relationship evaluations, denoting the need to focus on better quality services that lead to satisfaction (Grönroos, 2009;Oliver, 2010).

Limitations and future studies
Firstly, the main limitation is the use of experiments with small samples and mainly in commercial relationship situations. Thus, further studies using different approaches or assessing different contexts are needed to verify this phenomenon. Here, a new fluency manipulation was demonstrated, but it is still possible to verify some of the fluency effects occurring in our daily lives (Alter & Oppenheimer, 2009), or to verify the influences of perceived ease or difficulty and overconfidence related to the individuals' ability to process information (such as having difficulties in mathematics) as well as other characteristics. It is possible to verify the impact of factors such as age, the level of education (Mendes-da- Silva & Yu, 2009), greater intrinsic experience that affects selfconfidence (Bortoli & Soares, 2019), and the acceptance of anchors that could influence subsequent decisions (Tronco, Löbler, dos Santos, & Nishi, 2019). Verifying whether consumers' purchase intentions are affected by their difficulty to calculate monthly interest and installments, or even differences in younger and older people's overconfidence (Mendes-da- Silva & Yu, 2009), would help store owners to think about how to proceed in such situations.
Moreover, the different techniques to manipulate visual fluency (size or color typefaces, etc.) may produce results without managers' awareness, for example, in eWOM (electronic word of mouth) results. Much has been studied about the motivations for making, receiving, or sharing eWOMs and the impact of perceived credibility (Chen, Yang, & Wang, 2016;Hennig-Thurau, Gwinner, Walsh, & Gremler, 2004). Studies indicate that higher argument quality and writing clarity (general aspects) can positively influence credibility and eWOM acceptance (Cheung, Luo, Sia, & Chen 2009;Moran, Muzellec, & Nolan, 2014). Therefore, visual fluency, or even conceptual fluency, may affect the credibility and thus the acceptance of eWOMs.
In addition, future studies could verify whether these effects are related to perceived pleasure or cognitive comfort when processing information (Duke et al., 2014;Labroo & Pocheptsova, 2016). However, despite the fact that dissatisfied customers may feel greater ease in their commercial evaluations (Claypool et al., 2015;Landwehr et al., 2017), they might not experience pleasure or comfort remembering negative commercial events (Maier & Dost, 2018).
Finally, individuals may expect difficulties (or ease) in their decisions. If these expectations fit with the difficulty/ease encountered, this congruence would lead to better product evaluations (Jiang & Hong, 2014). These situations, involving evaluations or decisions, could be more complex and relevant in real-life environments due to low expertise (Payne et al., 1992) -mainly due to low general knowledge about the same or other products.
Thus, it is possible to find different results when consumers are making decisions about the same or other products/services, depending on their involvement.