This study examines the association between behavioral loyalty and satisfaction scores for banks. Past work has generally viewed the link between satisfaction and loyalty to be one way – satisfaction causes or induces loyalty. This study suggests the relationship may not be just one-way, and that current loyal behavior towards banks (measured as using 1, 2 or 3 banks) may be related to satisfaction scores: the more banks used, the lower the satisfaction score.
The study employs large-scale survey data from the UK YouGov panel. It analyses satisfaction scores for 16 banks, from consumers who use either 1, 2 or 3 banks.
Banks receive lower satisfaction scores from their customers who use one other bank, compared to customers who do not use one other bank. Furthermore, users of two banks are less satisfied with either of them compared to users of one, and users of three banks are, on average, less satisfied with each of them compared to users of two.
The results will help managers and researchers better understand satisfaction scores. For example, part of the reason why a bank obtains low satisfaction scores could be that it has a large proportion of dual or multi-bank customers. Next, knowing that satisfaction scores differ according to the number of banks currently used may contribute to a more nuanced understanding of the link between satisfaction and future loyalty.
The study is highly original in proposing a novel hypothesis relating to bank usage and how it relates to satisfaction scores.
Introduction
A key concept in marketing is that firms should strive to satisfy their customers. Influential papers from many decades ago such as Drucker (1958) and Levitt (1960), to more recent work (Zhou et al., 2009) emphasize that careful attention to satisfying customer needs is essential for firms to prosper. The popular marketing concept is based on the idea that firms will be more successful if they understand customer needs and have a focus on satisfying their customers (Houston, 1986; Gamble et al., 2011). Indeed, it seems difficult to imagine that a firm can grow and prosper in a competitive market without being able to achieve satisfaction among its customers. In markets such as banking, factors such as new competitors (Buhler et al., 2023) and in some cases, government intervention (Competition and Markets Authority, 2023) have spurred banks to work harder to understand customer needs and satisfy them (Haines, 2022; Harrison, 2021).
Satisfaction is said to benefit firms, such as banks, in several ways, one being via enhanced customer loyalty (Cooil et al., 2007; Mainardes and Freitas, 2023). In turn, loyalty is claimed to pay off via increased sales (from current buyers), and potentially higher prices (Homburg et al., 2005) – although there are contrary findings on this issue (Reinartz and Kumar, 2000). Loyal customers are also said to generate positive word of mouth (Casaló et al., 2008; Szymanski and Henard, 2001). The focus of this study is the relationship between loyal behavior and satisfaction in the context of consumer banking. In service categories such as banking, prevalent behavioral loyalty metrics are the number of products a customer has with a firm (Nagar and Rajan, 2005), or the customer's share of category requirements allocated to a particular firm (Cooil et al., 2007). Another behavioral loyalty measure is tenure, i.e. staying with a provider over time (Becker et al., 2015; Dawes, 2009; East et al., 2001). Some studies examine the extent to which customers are solely loyal, that is, using only one brand instead of several; or use only a small subset of the brands available instead of many (Quoquab et al., 2014; Sharp et al., 2002; Trinh, 2014). The number of providers a customer uses is related to share of category requirements in that a user of only one brand allocates 100% of requirements to it; the more brands used, the lower the level of loyalty to each on average (Dawes, 2008). In this study we use the number of banks used by consumers as a measure of behavioral loyalty.
Many studies have endeavored to empirically verify that buyer satisfaction is associated with customer loyalty. The unit of observation or analysis in studies is often the individual customer, whereby an analysis will examine the extent to which satisfaction scores for individual customers are associated with various loyalty metrics for those same customers. These loyalty metrics could be actual purchase behavior, such as tenure (Dawes, 2009), repeat-buying (Mittal and Kamakura, 2001) and share of wallet (Mägi, 2003) or purchase intentions towards the brand (Dash et al., 2021; Kellar and Preis, 2011; Özkan et al., 2020). Some other studies use the firm or brand as the unit of analysis, and examine how firm's satisfaction scores are related to outcomes such as revenue (Babakus et al., 2004; Van Doorn et al., 2013), market share growth (Baehre et al., 2021) or profit (Bhattacharya et al., 2021).
There is compelling logic that customer satisfaction should affect loyalty. Many empirical studies have found a positive correlation between satisfaction and loyalty as cited above, as well as in meta analyses by Ladeira et al. (2016) and Buhler et al. (2023). That said, it is worth at least considering if the direction of causality between satisfaction and loyalty is strictly one-way (i.e. satisfaction causes loyalty).
Could it also be possible that loyalty, specifically past and present loyal behavior, is at least to some extent associated with higher satisfaction levels presently? Understanding if this could be the case could be illuminating for managers and scholarly researchers, because it would help contextualize satisfaction metrics. To understand why, take the example whereby a firm such as a bank surveys its customers, and obtains satisfaction scores as well as indicators of future behavioral loyalty. The firm finds a correlation between satisfaction and loyalty, and concludes that satisfaction drives future loyal behavior, and so aiming for even high satisfaction is warranted. However, if past behavior exerts at least a partial influence on satisfaction levels, and future behavior, or behavioral intentions, is correlated with past behavior (Shapiro et al., 2013; Smith et al., 2008), then the firm may obtain an inflated picture of the satisfaction-future loyalty relationship.
A second example of this need for contextualization is that a firm could survey its own customers, and ask them which other brands they use, and how satisfied they are with those brands. It concludes that its customers show higher levels of satisfaction than they do with its competitors. However, this result could be spurious. The reason is that some of the firm's own customers will only deal with it and no others, therefore are solely loyal. This means that the focal firm's average level of loyalty among its own customer base is higher than among any of its competitors in the same sample. Under this scenario, if past or present loyal behavior influences satisfaction, then the firm will obtain an upwardly biased view of satisfaction levels compared to competitors – unless it knows, from evidence, that current usage influences satisfaction.
These scenarios suggest further understanding of the loyalty behavior: satisfaction relationship would be important and useful. This study does not attempt to prove or disprove whether there is a causal link between satisfaction and loyalty, nor loyalty and satisfaction. Rather, the study goal is to test several hypotheses relating to whether satisfaction scores are (simply) related to current levels of brand loyalty (sole, dual or multi-brand use). We undertake the analysis in a way that allows us to infer that it is behavior that could be influencing loyalty. The results of the study may allow for a more informed interpretation of customer satisfaction and loyalty metrics, as per the types of scenarios in the two examples above. The study next examines past work related to this topic, then drawing on past work relating to consumer memory, self-perception theory and cognitive dissonance, poses several hypotheses relating to the association between current loyal behavior and satisfaction levels. An empirical study tests the hypotheses using large-scale consumer-panel survey data on banking in the UK. The study shows support for the idea that past and present behavior may indeed influence customer satisfaction scores, using banking as the study context. The results represent a novel contribution to the body of knowledge on satisfaction and loyalty.
Literature review and hypothesis development
The two key concepts examined in this study are brand loyalty and customer satisfaction.
Loyalty
Loyalty is an important topic for managers as well as among academic researchers (Casteran et al., 2019; Graham et al., 2017; Khamitov et al., 2019). Brand loyalty is said to provide numerous benefits to brands. Firstly, a brand's sales comprise how many customers it has, multiplied by how much they purchase of it (Uncles and Ellis, 1989) therefore brand loyalty is a source of ongoing revenue. Other stated benefits of brand loyalty include the ability to maintain price levels (Chaudhuri and Holbrook, 2001), resistance to competitor threats (Dekimpe et al., 1997), making brand or line extensions more viable (Martinelli et al., 2015; Grasby et al., 2022) and aiding cross-selling activities (Liu-Thompkins and Tam, 2013).
Brand loyalty is generally conceptualized in the academic literature as a commitment to consistently repurchase or support a product or service despite attractive alternatives (e.g. Dang et al., 2023). This two-factor view of loyalty reflects a long-held view in the literature that loyalty necessarily includes both favorable attitudes and behavior (Day, 1969; Jacoby, 1971). That said, it appears that only a small proportion of buyers meet the criteria of favorable attitudes and behavior such that they are committed to consistently repurchase a brand. This is because numerous studies find widespread polygamous or multi-brand buying, for goods (Banelis et al., 2013; Ehrenberg, 2000) as well as services, including banking (Bogomolova and Grudinina, 2011; Mundt et al., 2006; Sharp et al., 2023). Indeed, a rich stream of literature has examined behavioral brand loyalty, finding generalized patterns such as the double jeopardy effect whereby loyalty is systematically associated with the size (measured as penetration or market share) of a brand (e.g. Ehrenberg and Goodhardt, 2002; Heiens and Pleshko, 2014; Kucuk, 2008; Wilson and Winchester, 2019).
The topic of interest in this study is the association between behavioral loyalty and satisfaction scores. We examine consumer's use of one brand, i.e. sole-brand loyalty (Sharp et al., 2002), compared to dual brand or multi-brand loyalty (Quoquab et al., 2014) and how the number of brands used by consumers relates to the satisfaction scores they give to those brands. Using more brands in a category is interpreted as exhibiting lower behavioral loyalty to any one of them (Bandyopadhyay and Martell, 2007; Dawes, 2008). We acknowledge that number of banks used is a narrow measure of loyalty compared to combined measures incorporating behavior and attitudes. However, the key findings of this study, relating to how bank usage is related to customer satisfaction, are relevant to managers and researchers regardless of the definition or interpretation of loyalty.
Satisfaction
Our second key concept, satisfaction, is widely accepted to comprise an evaluative response to one's consumer experiences (such as buying, using, inquiring), often according to the extent that the experience fulfills expectations (Teeroovengadum, 2022; Torres and Kline, 2006). That is, consumers experience various interactions with service providers, make evaluative judgments about the provider's performance, and possibly whether that performance lived up to their expectations, and form or revise their perception of satisfaction with the provider. Satisfaction is said to be a strong antecedent to customer loyalty (Bloemer, 1995; Casaló et al., 2008), and a large number of empirical studies find a positive association between the two constructs (e.g. Mittal and Kamakura, 2001). That said, the association appears to be stronger for loyalty intentions than subsequent loyal behavior (Bodet, 2008; Mägi, 2003; Seiders et al., 2005). Customer satisfaction is a key marketing performance metric, and is widely measured and tracked by service organizations (e.g. Index, 2022).
The role of memory in eliciting satisfaction scores
In relation to individual's satisfaction scores, consider that consumers use many product categories and deal with many service providers (e.g. banks, general insurance companies, telecoms providers, health insurance, utilities and public transport). Given this complexity, it is possible that they do not carry all the information on their perceived satisfaction levels for all those providers in working memory. Indeed, consumers who agree to be surveyed are not given a chance to prepare for the survey, and so their responses are to some extent “in-the-moment”. Pedersen et al. (2011) find that remembered satisfaction scores are different to scores that are recorded while experiencing a service. This implies either that the valence of people's evaluations of interactions with products or firms dissipates over time, but also that consumer's memories of those interactions are imperfect recollections of how they felt at the time.
It is also the case that many aspects of using products or dealing with firms are routine or uneventful (Van Birgelen et al., 2006) and therefore unless subject to service failure, may not persist in memory (Morgan and Rao, 2003); despite some firms endeavoring to create memorable experiences (Williams et al., 2020). Therefore, it may be the case that consumer's internal knowledge of their current state of satisfaction with a provider, when asked, is less than perfect. In these circumstances, it is possible that, according to self-perception theory (Bem, 1967; Tybout and Scott, 1983; Valkenburg, 2017), consumers can form attitudes by reference to, or to be, consistent with their prior behavior. For example, a consumer may, in the absence of highly salient information on whether they should be satisfied or dissatisfied with a provider, infer they are satisfied because they are behaviorally loyal to only one provider in a category; but perhaps also if they use more than one bank, infer they are less satisfied with at least one of the banks they use because they use more than one.
Psychological theories relating to current bank usage/loyalty and satisfaction scores
Another prominent psychological theory, namely cognitive dissonance (e.g. Harmon-Jones and Mills, 2019) suggests that in some circumstances, consumers adjust their attitudes to be consistent with their prior behavior. Olson and Stone (2014) provide numerous examples of how this effect can occur. In the context of the present study, it is plausible to consider the following dissonance scenario: a consumer changed from being a sole-brand user (a bank brand, in the context of this study) to begin dealing with a second bank – for a benign reason such as reacting to an advertisement; but when later asked about their satisfaction with the bank they were formerly a sole-brand user of, they adjust their perceived satisfaction level to be consistent with their choice (I stopped being solely loyal, perhaps I was not satisfied). However, this theory would also suggest that to be cognitively consistent, the consumer might state a higher level of satisfaction with the second brand than the brand they were formerly solely-loyal to, to justify their action. Therefore, cognitive dissonance provides only a partial underpinning for the phenomena we are seeking to examine.
Another theory which may be used to understand whether behavior is related to satisfaction is expectation-disconfirmation. According to this theory, a consumer's satisfaction level is lower if one's perception of performance (such as customer service, amenities and value) received was below expectations, or disconfirmed (Ofir and Simonson, 2007; Serenko and Stach, 2009). Evidence suggests negative disconfirmation exerts a stronger influence on consumer satisfaction and loyalty than a positive disconfirmation (Yoon and Kim, 2000). Several studies suggest that there could be an association between past/present behavior and satisfaction that works via expectation-disconfirmation. That is, heavier use of a category results in higher expectations and subsequently lower satisfaction. Litvin (2023) found heavy users of a hotel chain gave slightly lower review scores (mean score of 4.2 versus 4.3 out of 5) than lighter users. In a similar vein, Curtis et al. (2012) found frequent flyers gave lower satisfaction scores for their airlines, although unfortunately the study did not report actual scores.
Applied to our study, we propose on the basis of expectation-disconfirmation theory that customers who are dual- or multi-brand bank customers are likely to be heavier users of the financial services category compared to sole-bank users. They therefore have more complex needs, and accordingly higher expectations, therefore lower satisfaction. We also suggest, on the basis of self-perception theory, that dual and multi-brand bank usage will be associated with somewhat lower customer satisfaction levels compared to sole bank usage.
Based on these arguments, we pose H1.
Consumers who are solely loyal to a bank report higher satisfaction with it than dual-brand consumers, i.e. customers of that bank who also use a second bank.
Next, if more banking usage is associated with lower satisfaction, as per H1, a logical extension is to test whether it holds not only for the comparison of sole-bank customers to dual-bank customers, but for the comparison of dual-bank customers to multi-brand (3 bank) customers. We therefore pose H2 as follows.
Customers who use a bank (the focal bank) and one other bank will report higher satisfaction with that focal bank, than consumers who use a bank and two other banks.
To illustrate this idea, we hypothesize that for example in the UK market, Barclays users who also use Natwest and no other bank will be more satisfied with Natwest, compared to Barclays users who also use Natwest and Santander, or who also use Natwest and Lloyds.
It can be argued the reason that multi-brand bank customers use a second (or third) brand is because they became dissatisfied with their original bank, therefore they commence dealing with a second bank. However, if we observe that consumers who use more than one bank are less satisfied with either of the banks they use compared to consumers who only use one bank (and the same for consumers who use three compared to two banks), this would be consistent with the idea that current bank loyalty (sole, dual or multi-bank usage) is influencing satisfaction judgments. We therefore pose H3.
Customers who use more banks will report lower satisfaction scores for each of the banks they use.
To empirically test our hypotheses we employ data for the UK retail banking market. This market is suitable for the study, with reports that many consumers simultaneously bank with more than one institution (Moon et al., 2015).
Method considerations
Several issues pertaining to method need to be considered before examining the link between behavioral loyalty and satisfaction. These issues relate to obtaining measure of the constructs of interest (loyalty and satisfaction) from the same source such as a survey. That is, if one asks consumers how satisfied they are with a provider (often involving multiple questions), arguably these questions can influence the respondent's answers to subsequent questions on behavioral loyalty. Consistent with this concern, De Jong et al. (2012) reported there is considerable state dependence in surveys, that is, responses to a given question are quite highly correlated with the responses to the previous question. This means the correlations obtained between two constructs in the same survey could be inflated. Relatedly, literature identifies that survey questions can affect subsequent responses via priming (Schiff et al., 2022): psychologically disposing respondents to answer questions in a certain way due to the content or wording of previous questions. An example is Mieczkowski et al. (2020) that finds evidence of strong priming effects of asking questions about social media addiction on subsequent questions relating to depression. Minton et al. (2017) reviews priming effects in marketing and notes that survey responses are open to priming effects. Auh et al. (2003) find that asking satisfaction attributes first in a survey, then an overall satisfaction score results in higher scores for the latter (average scores of 7.6 vs. 7.2 out of 10), which is consistent with a priming effect.
One possible way to possibly reduce the problem of state-dependence or priming is to match actual behavioral data from purchase records (such as bank records of accounts) to survey responses, but a typical firm can only access data relating to its customers behavior with the firm itself, not for competitors. Another approach is to not specifically ask about both satisfaction and loyalty in the one survey, but rather simply ask firstly about bank usage, then ask about satisfaction with the banks used (which arguably results in less response bias than asking questions about satisfaction, then about loyal behavior or attitudes). This is the method we adopt here. We employ survey data, obtained from the UK YouGov BrandIndex. YouGov runs a demographically weighted survey panel in the UK with over a million respondents (YouGov, 2023). The YouGov contains data on brand usage as well as satisfaction, allowing us to derive the measure of loyalty based on the number of banks used. The survey sample is UK residents. Detail on the sample composition is provided as Table 6. The sample covers age, gender, income and region groups well and closely reflects the UK population (CountryMeters, 2024; Statista, 2023b, c).
Sample composition
| Total sample size, 12 months 2020 | 1,361,371 |
|---|---|
| % of sample | |
| Age | |
| 18–34 yrs | 21 |
| 35–49 yrs | 26 |
| 50+ yrs | 58 |
| 100 | |
| Gender | |
| Male | 49 |
| Female | 51 |
| 100 | |
| Income | |
| Income under £25k | 32 |
| Income £25k–under £50k | 31 |
| Income over £50k | 37 |
| 100 | |
| Region | |
| North East | 4 |
| North West | 11 |
| Yorkshire–Humber | 9 |
| East Midlands | 8 |
| West Midlands | 8 |
| East of England | 10 |
| London | 11 |
| South East | 13 |
| South West | 9 |
| Wales | 5 |
| Scotland | 10 |
| Northern Ireland | 2 |
| 100 | |
Source(s): The author's own creation
The survey questions we use are worded as follows. First, respondents are asked “Do you currently have a financial product with any of the following banks/building societies?” with a full list of banks shown. Then, respondents are shown the subset of banks they use, and are asked “Of which of the following banks would you say you are a SATISFIED CUSTOMER [response categories Satisfied, Neutral, Dissatisfied]”.
The YouGov data are supplied via a dashboard-type interface that allows users to select brands, metrics and filters to create specific consumer profiles. We selected the top 16 banks in the UK, for the calendar year 2020. The usage levels of these banks range from 22% for the largest, Nationwide, to 3% for the smallest of the 16, Starling. We then filtered the results from each bank's self-stated customer group to identify solely-loyal customers, defined as using a particular bank but not any of the other 16 banks [1]. We then created additional customer groups based on using one other bank – for example, Santander users who also bank with Halifax but no other bank, or Santander users who also bank with HSBC but no other bank, and so on. We repeated this process for each of the banks. The process was then repeated to identify consumers who used a particular bank and two other banks. In total we created 183 bank-usage groups with every bank having a user group that was solely loyal, as well as other user groups who used that bank and one of a set of other banks, and lastly user groups that used that bank and two other banks. We then calculated the satisfaction scores for each group. As mentioned, YouGov measures satisfaction using a three-point scale of satisfied, neutral or dissatisfied. While finer scales offer more discrimination for respondents, other published studies use similarly coarse scales from YouGov (e.g. Du et al., 2019), or use “top 2 box” satisfaction scores (Van Doorn et al., 2013), therefore there is precedent for using a three-point measure. Also, while finer scales are likely more desirable for research that seeks to analyze at the level of the individual respondent, the present study is conducted at the aggregated group level. Lastly, a recent study by Malshe et al. (2020) found high convergent validity between the YouGov satisfaction metric and other widely used satisfaction scores, suggesting it is a valid measure for the present purpose. For each brand/customer group, we calculated the average satisfaction score using 1 for satisfied, zero for neutral and −1 for dissatisfied with a theoretical range for each bank of −100 to 100.
Analysis and results
We now examine the evidence for H1. We begin with a model-free analysis. In Figure 1 we show the mean satisfaction score for each bank among its solely-loyal customers, compared to its customers who use one other bank (i.e. are dual-brand customers). Figure 1 shows that in 14 of 16 cases the satisfaction score is lower among dual-brand loyals (scores are lower on average by seven points). This provides initial evidence that supports H1.
We employ a one-way ANOVA to more formally test H1. Note that in relation to this analysis we have one score for each bank from its sole loyals, but multiple scores for each bank among its dual-loyals, because there are many potential banks they could also be a customer of. For instance, we have one aggregated score for Santander from its sole-loyal customers, but we have multiple scores for Santander among its dual-brand customers: those who also use Barclays, those who also use HSBC and so on. This is seen in the number of observations for dual-brand and multi-brand loyals are larger than for solely-loyals in Table 2. In some cases, the mean score for a particular bank among dual-brand or multi-brand customers was based on a very small sample. This is because the count of customers of a small-share bank who also use another small-share bank becomes very small, even with a large panel sample. To avoid the results being biased by scores from very small samples such as these we restricted the analysis to samples of at least five responses or more. In other words, each observation in the analysis was based on at least five responses. The results of the ANOVA are shown in Table 1. We see that the mean satisfaction score is higher among sole-loyals than dual-brand loyals.
Satisfaction score comparisons: formerly used one other bank, vs. currently use one other bank
| Bank 1 customer group | Status | Bank 2 | Satisfaction score for bank 1 | Bank 1 | Status | Bank 2 customer group | Satisfaction score for bank 1 | Difference (current – former) |
|---|---|---|---|---|---|---|---|---|
| Barclays | Former users of | Halifax | 60 | Barclays | Current users of | Halifax | 64 | 4 |
| Barclays | Former users of | Nationwide | 63 | Barclays | Current users of | Nationwide | 34 | −29 |
| Barclays | Former users of | Santander | 58 | Barclays | Current users of | Santander | 52 | −6 |
| Halifax | Former users of | Barclays | 70 | Halifax | Current users of | Barclays | 58 | −13 |
| Halifax | Former users of | Nationwide | 81 | Halifax | Current users of | Nationwide | 60 | −21 |
| Halifax | Former users of | Santander | 78 | Halifax | Current users of | Santander | 55 | −23 |
| Nationwide | Former users of | Barclays | 83 | Nationwide | Current users of | Barclays | 75 | −7 |
| Nationwide | Former users of | Halifax | 85 | Nationwide | Current users of | Halifax | 78 | −7 |
| Nationwide | Former users of | Santander | 81 | Nationwide | Current users of | Santander | 80 | −2 |
| Santander | Former users of | Barclays | 63 | Santander | Current users of | Barclays | 68 | 5 |
| Santander | Former users of | Halifax | 69 | Santander | Current users of | Halifax | 66 | −3 |
| Santander | Former users of | Nationwide | 64 | Santander | Current users of | Nationwide | 60 | −4 |
| Average | 71 | 62 | −9 |
Source(s): The author's own creation. YouGov BrandIndex UK 2022 © All rights reserved
Analysis of variance
| Descriptives | ||||
|---|---|---|---|---|
| Number of observations | Mean satisfaction score | Std. dev | Std. error | |
| Solely loyal | 18 | 73.3 | 12.7 | 3.0 |
| Dual-brand loyal | 104 | 64.0 | 15.4 | 1.5 |
| Multi-brand loyal | 40 | 56.9 | 23.0 | 3.6 |
| Total | 162 | 63.3 | 17.8 | 1.4 |
| ANOVA | |||||
|---|---|---|---|---|---|
| Sum of squares | df | Mean square | F | Sig | |
| Between groups | 3526.3 | 2 | 1763.2 | 5.9 | 0.003 |
| Within groups | 47,748 | 159 | 300.3 | ||
| Total | 51,274 | 161 | |||
| Post-Hoc tests | Mean difference (I − J) | Std error | Sig. (2-tailed) | |
|---|---|---|---|---|
| I | J | |||
| Sole loyal | Dual-Loyal | 9.3 | 4.4 | 0.10 |
| Multi-brand loyal | 16.4 | 4.9 | 0.003 | |
| Dual loyal | Multi-brand loyal | 7.1 | 3.2 | 0.08 |
Source(s): The author's own creation
The ANOVA supports H1: the mean score among solely-loyal bank customers is higher than it is for dual-brand customers, (73.3 vs 64.0). The two-tailed significance test (Bonferroni adjusted for multiple comparisons) is p = 0.10. Given we hypothesized that it would be higher, not just different to that of dual-bank customers, we can use the one-tailed probability, which is statistically significant at the p = 0.05 level. The result is not only statistically significant (i.e. not attributable to random sampling variation) it is managerially significant, with a nine-point difference in satisfaction scores between sole and dual-brand loyals and a seven-point difference between dual-brand and multi-brand loyals.
We performed an additional analysis to gain more insight into the association between current loyalty and satisfaction. If brand loyalty (as we measure it here, being the number of banks used) is associated with satisfaction scores, then we should also see a difference in the scores among people who currently use two banks A and B, compared to people who currently use bank A and formerly used B, or currently use B and formerly used A. The YouGov data asks not only about current brand usage, but identifies which brands the respondent formerly used. We selected the four largest brands, and for each we compare the scores for those currently using any two of them (and no others) against the scores for consumers currently using one of them, and formerly used one of the others. The results are shown in Table 2.
To illustrate the results in Table 2, we take the example of Santander users who formerly banked with Nationwide: their satisfaction score for Santander is 64, whereas Santander users who currently also bank with Nationwide give Santander a lower score of 60. The difference in scores is higher among bank customers who formerly used one bank, compared to bank customers who currently use another bank in 10 of the 12 comparisons in Table 2. We ran a t-test to confirm if the results were statistically significant, they were at under the 0.10 level (t = −1.8, df 22, p = 0.07). There is an unusually large difference in the Barclays-Nationwide comparison, whereby the score for Nationwide among current users of both brands is only 34% compared to 63% for Nationwide customers who formerly banked with Barclays. This could be an outlier that skewed the results. We checked the scores for a second year (2021) and the results were similar (an average difference of six points, t = −1.67, df 22, p = 0.11).
Overall, we see these results confirming there is certainly an association between current brand loyalty (measured as number of brands used), and satisfaction scores.
We now turn to H2, which relates to the difference in satisfaction scores among dual-brand loyals compared to multi-brand loyals. We refer again to the ANOVA in Table 1 and compare the satisfaction scores for these two groups. The mean score for dual-brand customers is 64.0, and for multi-brand customers it is 56.9. The two-tailed significance test (Bonferroni adjusted) is p = 0.08. Again, since we hypothesized that the satisfaction score would be lower among multi-brand customers, the appropriate test is one-tailed, therefore the result is more highly significant, at p = 0.04. Therefore, H2 is supported.
We now examine the evidence relating to H3. This hypothesis posits that consumers who use more banks will give lower satisfaction scores for each or all of the banks they use.
We firstly present some descriptive evidence that also illustrates the analysis. We start by examining dual-bank customers compared to sole-bank customers. In Table 3 we show the satisfaction levels of solely loyal users of banks such as Barclays and Lloyds, then show the satisfaction levels of dual bank users of both Barclays and Lloyds, for both the brands they use. We see, for instance that Barclays and Lloyds satisfaction scores are both lower among customers who use both, than among customers who use only one. The findings are similar for the comparison of Natwest and Santander customers.
Example analysis comparing satisfaction for sole-loyal customers vs. satisfaction with both brands used by dual-loyals
| Customer type | Satisfaction with Barclays | Satisfaction with Lloyds |
|---|---|---|
| Solely loyal to Barclays | 62 | – |
| Solely loyal to Lloyds | – | 59 |
| Bank with both Barclays and Lloyds | 60 | 50 |
| Customer type | Satisfaction with Natwest | Satisfaction with Santander |
|---|---|---|
| Solely loyal to Natwest | 64 | – |
| Solely loyal to Santander | – | 65 |
| Bank with both Natwest and Santander | 60 | 54 |
Source(s): The author's own creation. YouGov BrandIndex UK 2022 © All rights reserved
We now illustrate this analysis approach further to compare scores from dual-bank customers with scores from customers who bank with three banks. For example, in Table 4 we show the satisfaction scores from customers who bank with Halifax and Lloyds, or Halifax and Barclays, or Lloyds and Barclays, then compare those scores with scores from customers who bank with all three of those banks. We see in two of three cases the average satisfaction score is higher among those who bank with only two of these banks than among those who bank with all three.
Example analysis comparing satisfaction for dual-loyal customers vs. satisfaction with all brands used by multi-brand-loyals
| Halifax, Lloyds and Barclays | Satisfaction with Halifax | Satisfaction with Lloyds | Satisfaction with Barclays |
|---|---|---|---|
| Bank with Halifax and Lloyds | 60 | 70 | |
| Bank with Halifax and Barclays | 57 | 67 | |
| Bank with Lloyds and Barclays | 64 | 50 | |
| Bank with Halifax, Lloyds and Barclays | 67 | 54 | 52 |
Source(s): The author's own creation. YouGov BrandIndex UK 2022 © All rights reserved
To formally test H3 we first compared scores from sole-loyalty customers to scores from dual-loyalty customers from the full sample. We assembled all the scores for each bank from their solely-loyal customers, along with the scores for the other banks among their dual-brand customers. For instance, we take the score for Barclays among its sole-loyals, and compare it to the score for Barclays among its dual-brand loyals, then for a series of other banks that Barclays dual-brand customers use – Bank of Scotland, Halifax, HSBC and so on. We then compared the scores from dual-bank customers to those from sole-bank customers. In total we created 109 comparison satisfaction scores from dual-brand bank customers. In 90% of cases (n = 90), the scores from dual-bank customers were lower than the score from sole-bank customers. If H3 was incorrect we would expect that dual-bank scores would be the same as sole-bank scores, therefore would be higher in only 50% of cases due to random variation. The 95% confidence interval for 81% from a sample of 94 is 71–88%, which is certainly above 50%. Therefore, in the majority of cases, satisfaction scores from dual-loyalty customers are lower for both the brands they use, compared to scores from sole-loyalty customers – supporting H3. We now test H3 further by comparing satisfaction scores from dual-loyalty to multi-brand loyalty customers.
For this next test of H3 we assembled the scores for each bank from its dual-bank loyals, then compared them to the scores from the bank's multi-brand loyals. For example, we take the scores for Barclays given by Barclays customers who are dual-brand users, and compare them to the scores for Barclays given by Barclays customers who are multi-brand users, as well as the scores those customers gave the other banks they were customers of. The number of observations is smaller than for the sole-to-dual analysis because there are fewer consumers who bank with three brands. We created 63 comparison groups (it was not possible to cross-compare every combination due to sample sizes), and in 46 of them (72%), the satisfaction scores were lower for each of the brands used by multi-brand loyals compared to both of the brands used by dual-brand loyals. The 95% confidence interval is 62–83%, above the 50% chance level. This result supports H3 for the dual-to-multi brand loyals case. We summarize the results in Table 5.
Summary of results
| Hypothesis | Findings |
|---|---|
| H1. Sole-brand bank customers will report higher satisfaction levels than dual-brand (use two banks) customers | Supported |
| H2. Customers who use a focal bank and one other bank will report higher satisfaction with the focal bank compared to those customers who use a focal bank and two other banks | Supported |
| H3. Customers who use more banks will report lower satisfaction scores for each of the banks they use | Supported for both the sole-loyal to dual-brand loyal case, and the dual-brand loyal to multi-brand loyal case |
Source(s): The author's own creation
While the present findings point to an association between bank usage levels and satisfaction, the question arises whether there is an alternative explanation, pertaining to customer characteristics. It is plausible that consumers who use more banks tend to have higher incomes. This is because with more wealth, they use more financial products and services. Next, it could be the case that income is negatively associated with satisfaction levels. People on higher incomes may become accustomed to higher-quality goods and services, therefore based on expectation-disconfirmation theory, their expectations increase (Ofir and Simonson, 2007) thereby lowering their satisfaction levels. Therefore, it is possible that the observed relationship between bank usage and satisfaction level could be partly driven by income. We checked this possibility by analyzing the behavioral loyalty–satisfaction relationship by income level. We found the pattern of lower satisfaction scores among those who used more banks, within each of the low, medium and high-income groups. The results therefore do not appear to be due to income effects.
Discussion and implications
The study finds support for the broad hypothesis, primarily based on self-perception (Bem, 1972; Tybout and Scott, 1983; Valkenburg, 2017), and expectation-disconfirmation theories (Ofir and Simonson, 2007; Van Ryzin et al., 2004), that customer satisfaction levels will be related to current brand usage. In short, customers who use more banks will report lower satisfaction scores. The obvious question arises, is it the usage that is influencing the satisfaction score, or is it satisfaction that has influenced the usage? It is extremely plausible that a customer who is less satisfied with a bank decides to deal with a second bank, thereby down-weighting their reliance on the first bank. However, this scenario does not fully explain the pattern of results here, which is lower satisfaction for both banks when two are used, than when one is used. We acknowledge it is not possible to infer causality in relation to loyalty and satisfaction in this study – that is, to answer the question of whether brand loyalty causes lower satisfaction scores – this was not the intent of the study. Moreover, we are employing cross-sectional data, which is generally ill-suited to evaluating causality (Nogueira et al., 2022).
That said, we can also consider the result that not only is the average satisfaction score lower among customers who use two banks compared to one, and is lower among customers who use three banks instead of two – but also that customers who use two banks are less satisfied with each of them, compared to customers who use those same banks but use only one of them. Similarly, customers who use three banks are on average less satisfied with each of them than customers who use only two of those three banks. If this pattern was solely driven by low levels of satisfaction with a bank causing its buyers to commence dealing with another bank (albeit not ceasing to deal with the original bank) then arguably, customers would be pleased with the second or third bank they commence dealing with, and the satisfaction levels among dual-bank customers would be approximately similar to those among sole-brand customers. It is possible that some consumers, who happen to be sensitive to customer service, also wish to spread their financial requirement across multiple banks. However, we still see lower scores among customers who bank with a particular bank A and also bank with B, compared to those who ceased dealing with B and now only bank with A.
The study is informative to both managers and academics who conduct research on satisfaction and its links to outcomes such as loyalty. While there is a primary focus in literature and practice on how satisfaction may be linked to future-oriented outcomes as indicated via intentions to stay or switch from a brand (Dash et al., 2021; Kellar and Preis, 2011) or cross-buy from a brand (Verhoef et al., 2001), the study here identifies that there is a systematic association between current usage and satisfaction levels. That is, respondent's satisfaction scores for a bank – while at an individual level are influenced by recent interactions, service quality and perceptions of value (Caruana et al., 2000) and different expectations among different customers, e.g. (Christopher et al., 1991; Hu et al., 2019) – do appear to be systematically lower among customer who use more banks. The findings have implications for practitioners and researchers.
Consider the case of a study either by a marketing firm or academic researcher, that gathers satisfaction scores from customers of various service providers (such as banks) as well as obtaining purchase intentions. The study finds a correlation between satisfaction and purchase intention, consistent with numerous past studies in the literature. The conclusion is that satisfaction drives future loyal behavior. However, based on the results of this study, dual or multi-brand bank customers tend to give lower satisfaction scores. It is also logically likely that they would give lower scores for purchase intention than sole-loyalty customers, because the dual or multi-brand buyers already deal with several banks. For example, if a solely loyal customer is asked how likely they will purchase additional products from their current bank their answer will likely include perceptions of switching or set-up costs (Lam et al., 2004; Yang and Peterson, 2004) to commence dealing with a new bank, consequently their purchase intention to buy from their current bank will be high. By contrast, a dual or multi-bank customer already deals with several banks, so when asked how likely they will purchase additional products from one of the banks they deal with their answer will likely factor in the fact that it would be very easy for them to purchase from that bank or another bank they deal with – and therefore their purchase intention will be lower. Consequently, at least part of the satisfaction-loyalty correlation is attributable to current usage, and so without factoring in past and present usage the estimated strength of the satisfaction–loyalty relationship will be biased upward. Managers and academics need to take this phenomenon into account, in order to make correctly informed decisions about investments in service improvement. Managers therefore need to understand it may not be possible to achieve the same levels of customer satisfaction scores among dual and multi-brand loyal clients compared to what the firm achieves among sole-brand clients. The findings may therefore aid managers in setting realistic objectives for different customer groups.
Next, the results of this study are potentially informative to help explain variation in customer satisfaction scores for firms such as banks. If a bank has a significant number of dual and multi-bank customers, based on these results, this may help explain why it has lower scores than certain competitors with different profiles in terms of other-bank usage. Consider that some services providers such as banks may not offer a full suite of products. These institutions are less likely to attract customers who will be solely-loyal because it is not possible to do all one's banking needs with that provider. In turn, it could be that a limited-services bank will have a higher proportion of dual-and multi-bank customers (and very few solely-loyal), thereby resulting in lower satisfaction scores. Managers of limited-services banks, in particular, should use this contextual information when examining their customer satisfaction scores, knowing that differences in their customer base may be at least partly driving the results compared to competitors.
Limitations and directions for future research
The scope of this study was to pose hypotheses and examine the link between current behavior (specifically, the number of banks used) and satisfaction. A desirable next step is to start to understand why. If, as posited, the reason relates to self-perception theory then further studies can endeavor to isolate this as the prime explanation. It is certainly reported that many satisfied customers do leave service providers (Kumar et al., 2013; Naumann et al., 2010).
It may therefore also be the case that many satisfied customers decide to deal with new or additional service providers. Self-perception theory predicts their satisfaction levels with the original provider would decline after commencing with a new/additional provider. Longitudinal data may help understand if this is the explanation. The procedure would be to observe customers of various service providers who are solely loyal, gauge their satisfaction levels, then, at a later date, identify which customers commenced dealing with an additional provider and ascertain if their satisfaction levels with their original provider are now lower. If they are, this would implicate self-perception theory.
The study also suggested expectation-disconfirmation theory (Serenko and Stach, 2009) as a basis for the study hypotheses. Using this theory, the explanation is that heavier category users have higher expectations, and also use more bank brands. The question arises how to test if this is the superior explanation to self-perception theory. It may be possible to analyze the satisfaction scores of sole, dual and multi-brand bank customers controlling for their level of category usage, and identify if the results are consistent with expectation-disconfirmation theory.
A research question arises from the present work in relation to cross-bank satisfaction. For example, take the example of a manager of a large bank such as Santander – its customer base comprises customers of other banks (Sharp et al., 2023) and a research question concerns whether the satisfaction scores for Santander, for example, vary systematically depending on the other bank that its customer use. For instance, there are new “challenger” brands in banking such as Starling, Revolut and Monzo (FinTech Global, 2020) that are growing, but which also enjoy very high satisfaction scores (Statista, 2023a). Do these customers give Santander lower scores than its other customers, perhaps on the basis that their expectations are now being shaped by new, high-satisfaction providers? If so, the result could be that a bank's satisfaction scores decline as the composition of its customer base in terms of other-bank users increases. Given that dual and multi-brand loyalty are quite prevalent in services industries (Adiani et al., 2024; Mundt et al., 2006; Sharp et al., 2023), understanding this cross-bank satisfaction is a potentially important research direction for practitioners and academic scholars (Table 6).
Note
There are numerous other small banks outside the top 16 but they account for less than 5% of usage.
