Are satisfied long-term customers more profitable? Evidence from the telecommunication sector

Abstract

Companies strive to establish long-term customer relationships based on the premise that long-term customers with positive attitudinal dispositions (and therefore, loyal) are in general more profitable than others. Results indicate that such generalisations may be tenuous. While some effects of satisfaction and duration on behavioural intentions were found to support predominant beliefs in research, results were mixed. The combined effects of satisfaction and duration on behavioural intentions indicated that long-term customers might not necessarily be more profitable than new customers. These findings add an important cautionary note to firms who formulate their marketing strategies with the fundamental aim of establishing and maintaining long-term customer relationships, or indeed, believe that long-term customers with a positive attitudinal disposition are essentially profitable customers.

INTRODUCTION

Traditionally, there has been a dominant view, especially among practitioners that customer satisfaction, in the long term, leads to customer loyalty and that loyal customers are profitable customers. This assumption was based on extensively cited research, which offered empirical evidence indicating that long-term customers are satisfied, loyal customers and that (1) they generate more profits because they get accustomed to the service and use the service more; (2) they are less price sensitive and thus, companies can charge more; (3) they bring extra business through referrals and (4) they are more profitable because acquiring new customers is more costly than retaining them.¹ The significant investments in customer relationship management programmes over the last decade were at least partly influenced by these findings. Others, however, argued that ‘the contention that loyal customers are always more profitable is a gross oversimplification’.² The original work by Reichheld and Sasser¹ was also challenged by others,^{3, 4} who demonstrated the presence of high and low profitability customers among both short- and long-term customers. However, a possible criticism of some of these latter studies is that not all long-term customers are likely to be truly loyal customers, that is, have a positive attitudinal disposition towards their service provider,⁵ and that it is long-term, satisfied — and therefore loyal — customers who are profitable. The current study accounts for this limitation in the previous work by incorporating a measure of the consumer's positive attitudinal disposition towards the service provider as a proxy of true loyalty. This approach enables the examination of long-term behaviour among customers who hold different levels of positive attitudinal dispositions towards their service provider. This approach is also consistent with recent research, which examined the moderating influences on the relationship between satisfaction and behavioural outcomes, based on the same premise that the relationship between long-term satisfied customers and profitability is not simple. For instance, convenience, competitive intensity, customer involvement and household income have all been found to have moderating influences on the relationship between customer satisfaction and some behavioural dimensions that lead to profitability, such as repurchase.⁶ Furthermore, given the significant investments in relationship marketing being made by industry, a further examination of some of the intricacies of long-term customer behaviour is likely to be of practical value.

LITERATURE REVIEW

Relationship marketing emphasises the need for maintaining long-term customer relationships, and is based on the premise that it is in general beneficial to the service provider to do so,¹ especially in contractual settings.³ Analysis of customer behaviour over time has been conducted in both contractual^{7, 8} and noncontractual settings.^{9, 6} The current study focuses on a contractual setting, but distinct from previous work, examines the effects on a number of behavioural dimensions. Customer loyalty has been found to be a multi-dimensional construct consisting of purchase/repurchase behaviour, word-of-mouth (WOM) communication, complaining behaviour and price sensitivity.¹⁰ Thus, these four dimensions, which have been increasingly examined as unique constructs in the recent literature, form the dependent/outcome variables in this study.

Impact of satisfaction on outcome variables

A solid body of literature supports the positive relationship between customer satisfaction and repurchase intentions,^{11, 10} as well as actual repurchase behaviour.⁷ Indeed, research has also found significant nonlinear relationships between satisfaction and customer retention.¹² The focus of the current paper is, however, on the moderating impact of duration on the relationship between satisfaction and its outcomes. Customer satisfaction is known to be a major driver of positive WOM.^{13, 14, 15} The impact that lack of satisfaction has on negative WOM or complaining behaviour is also well recognised.^{16, 10} Empirical examinations of the impact of satisfaction on price sensitivity are less common. Yet, perceived service quality, often used as a proxy for satisfaction in service settings has been found to be a driver of a service provider's ability to charge price premiums from those having positive perceptual dispositions towards the service provider.¹⁷ This illustrates the possible negative association between satisfaction and price sensitivity. Based on this discussion, it follows that:

H₁::

Higher the level of satisfaction (a) higher the level of spend (b) higher the level of positive WOM (c) lower the level of negative WOM and (d) lower the level of price sensitivity.

Impact of duration on outcome variables

Reichheld and Sasser¹ were the first to suggest the idea that customers who stay with the same service provider are loyal customers, and that therefore, they are more profitable than others. This indicates a positive relationship between duration of stay and customer loyalty to the service provider. Research has also found that the effect of prior cumulative satisfaction on the duration of the provider–customer relationship is larger for customers who have more experience with the service provider.⁷ Results indicate that customers, who have already stayed for long, are more likely to stay longer (and thus, be even more loyal). These results also indicate a positive impact of duration on loyalty outcomes. More recent work,³ however, based on data from a noncontractual setting, found that long-term customers are not necessarily profitable customers. In relation to the specific outcome variables, they found among other things that long-term customers do not necessarily pay higher prices, indicating the absence of a significant, positive relationship with price sensitivity. Although there does not seem to be real agreement on the impact of duration, based on the greater weight of extant literature, especially in contractual settings, it is hypothesised that:

H₂::

Higher the length of service patronage (duration) (a) higher the level of spend (b) higher the level of positive WOM (c) lower the level of negative WOM and (d) lower the level of price sensitivity.

Moderating effects of duration

The research by Bolton and Lemon¹⁸ is one of the early studies that touched upon the concept that duration of stay would combine with other attitudes to determine future behaviour. Specifically, they found that customers use price and usage over time to update their evaluations of the fairness of the exchange, which in turn determine future courses of action. Recent research has also proposed relationship age as a moderator of the relationship between satisfaction and purchase intentions.⁶ There is also ample anecdotal evidence regarding the behaviour of long-term customers that implies a significant moderating effect. For instance the premise behind relationship marketing assumes that satisfied customers, in the long term, start displaying increasing loyalty behaviours.¹⁹ This indicates that, for instance, for a given level of satisfaction, long-term customers would display more loyalty behaviour than others. If so, it is hypothesised that:

H₃::

Duration of stay will have a positive reinforcing effect on the relationship between satisfaction and spend. At high levels of satisfaction, those who have stayed longer with the same service provider are likely to spend more.

H₄::

Duration of stay will have a positive attenuating effect on the relationship between satisfaction and price sensitivity. At high levels of satisfaction, those who have stayed longer with the same service provider are more likely to be less price sensitive.

H₅::

Duration of stay will have a positive reinforcing effect on the relationship between satisfaction and positive WOM. At high levels of satisfaction, those who have stayed longer with the same service provider are more likely to give positive WOM.

H₆::

Duration of stay will have a negative reinforcing effect on the relationship between satisfaction and negative WOM. At low levels of satisfaction, those who have stayed less time with the same service provider are more likely to give negative WOM.

METHODOLOGY

The data for this study were obtained through a large-scale survey of 3,000 fixed line telephone customers in south-eastern England. The survey achieved a valid response rate of 15.2 per cent. There were no significant (0.05 level) nonrespondent biases. While the UK fixed line telecommunication sector is still dominated by a former monopoly in terms of market share, the industry is also characterised by a level of competition that offers customers a clear choice, at least between two major players for both local and long-distance services. The tariff structure is characterised by per second/minute charges for both local and long-distance calls. The key measures in the study were conceptualised and operationalised based on previously validated scales, and were found to have high levels of reliability and validity.

Core construct — Customer satisfaction: Customer satisfaction has been described as ‘an evaluation of an emotion’.²⁰ It has also been suggested that ‘it reflects the degree to which a consumer believes that the possession and/or use of a service evokes positive feelings’.²¹ Others argued that satisfaction with a service provider is perceived as being both evaluative and emotion based.²² Satisfaction has also been conceptualised both in relation to specific transactions and as a cumulative outcome or overall satisfaction. The latter definition fits the current context of a contractual setting better. Cronin et al.²³ used a tool that reflected both these ‘emotion based’ measures and ‘evaluative’ measures, and was an overall measure of satisfaction. Our three-item, seven-point, Likert-type scale was based on this latter work.

Dependent construct: The overall dependent construct in this study was what is often referred to in the literature as customer loyalty. Customer loyalty is known to be multi-dimensional,¹⁰ and in the current study each of the four dimensions, identified in previous research,¹⁰ was studied as a unique dependent construct. What follows is a discussion of the extant literature on the operationalisation of each of these constructs.

Positive WOM: Positive WOM has been studied as a unique construct for some time.¹³ More recently, Gremler and Gwinner²⁴ conceptualised WOM communication as consisting of both voluntary and involuntary recommendation. This same two-dimensional typology was used to develop the two-item measure of positive WOM used in the current study. It captured both voluntary recommendation, as well as recommendation to those seeking advice.

Negative WOM: Negative WOM too has been studied to some degree in the extant literature. One of the early and established conceptu-alisations of the construct was by Singh¹⁶ who studied unfavourable behavioural intentions. He identified three dimensions of negative WOM, consisting of voice responses (seeking redress from seller), private responses (to friends and family) and third-party responses (eg legal action). The measure used in the current study consisted of two items, and captured both voice responses and private responses, but not the third-party action, since such action is not typical of the industry studied. This approach was also consistent with the procedure adopted by other well-cited studies.¹⁰

Price sensitivity: Measures of price sensitivity or, indeed, price tolerance originated in the economics stream and since then have been used in various management disciplines. Price tolerance has traditionally been defined as the price differential needed to switch, and has been measured using what is referred to as the ‘$ Differential score’.²⁵ Similar operationalisations have been used extensively since then, including in the Swedish customer satisfaction barometer study.²⁶ The measure of ‘sensitivity’ in the current study was the reverse of ‘tolerance’. It measured propensity to switch in the event of 5, 10 and 20 per cent price increases. Weighted summate (4, 2 and 1 ×, respectively) of the score was used to reflect the overall construct.

Level of Spend: Average spending level of customers was captured by a single-item measure split into five equal spending categories.

Moderating variable-duration

Moderating variable-duration: Self-reported duration of service patronage was measured in years. Where duration was less than one year, it was measured in months and converted to years.

All the reflective-type scales used in the study had high levels of internal consistency demonstrated by Cronbach's alpha values much greater than the minimum values suggested by Nunnally.

RESULTS

Table 1 illustrates the Pearson's correlation coefficients among the key constructs. The results suggest a significant positive effect of satisfaction on positive WOM (r = 0.57), and a negative effect on both negative WOM (r=−0.5) and price sensitivity (r=−0.41), confirming hypotheses H_1b, H_1c and H_1d. The effect of satisfaction on spend is, however, opposite to what was hypothesised. The results also suggest a small but significant positive effect of duration of stay on spend (r = 0.13), and a negative effect on negative WOM (r=−0.11) confirming hypotheses H_2a and H_2c. The effect of duration of stay on positive WOM is, however, opposite to what was hypothesised. Furthermore, contrary to expectations, duration of stay does not seem to have a significant effect on level of price sensitivity.

Table 1 Pearson correlation coefficients (2-tailed)

To examine the interaction effects, the following graphical and statistical analysis was conducted as per the recommended practice.²⁸ First, the satisfaction and duration scores were split into quartiles with 0.00 reflecting the lowest quartile and 3.00 reflecting the highest. Then, these scores were plotted against the mean levels of spend, price sensitivity, positive WOM and negative WOM, respectively. Subsequently, tests were undertaken to assess whether there were statistically significant differences between the lowest and highest quartiles. The results of this analysis are as follows. Among the highly satisfied respondents, a comparison of new and long-term customers showed no significant differences (α=0.05 level) in their levels of spend and price sensitivity (Figures 1 and 2).

Figure 1

Observed means of Ł spend

Figure 2

Observed means of price sensitivity

Therefore, there is inadequate support for hypotheses H₃ and H₄. At high levels of satisfaction, new customers were, however, significantly more likely (α=0.05 level) to give positive WOM than long-term customers (Figure 3). Thus, while there is evidence to indicate a significant moderating effect, the direction of the relationship is opposite to what was hypothesised in H₅. At low levels of satisfaction, new customers were significantly more likely (α=0.05 level) to give negative WOM than long-term customers (Figure 4), thus, supporting the presence of a significant moderating effect as hypothesised in H₆.

Figure 3

Observed means of positive WOM

Figure 4

Observed means of negative WOM

DISCUSSION

The results are consistent with existing research in a number of ways. They also, however, challenge extant research fundamentally. Effect of satisfaction on loyalty intentions and behaviours support predominant beliefs in research that in general, the more satisfied the customers, the more likely they are to display loyalty towards service provider, and indeed less likely to display negative behaviours.^{1, 3, 6, 7, 12} Effect of satisfaction on spend was, however, opposite to what was expected, indicating that the more satisfied customers are likely to spend less money on service. This, challenges the original well-cited research by Reichheld and Sasser.¹ It is plausible that those who spend less derive greater value from the service, and thus, are more satisfied, giving rise to the possibility that the direction of the causal relationship between satisfaction and spend may be opposite to what is commonly assumed, or indeed that the relationship is nonrecursive.

Effect of duration of service patronage on loyalty intentions and behaviours also illustrates mixed results. They indicate support for the belief that in the long run, customers spend more on service, and that they are less likely to give negative WOM.¹ Results also, however, challenge these previous findings,¹ in that the relationship between duration and price sensitivity is nonsignificant. Given that long-term customers spend more, yet are as price sensitive as new customers could mean that they spend more in the long term not necessarily out of any form of affective loyalty to the service provider, but rather due to pragmatic reasons, such as larger families or higher incomes. Furthermore, effect of duration on positive WOM is significant, yet the direction of the relationship is opposite to what was expected. This indicates that it is in fact new customers who give more positive WOM than the long-term customers. This may be due to possible inertia that may overtake the effect of satisfaction in the long term. Thus, while previous research,⁷ in a contractual setting, found that the longer the customers stay, they are even more likely to stay longer, which may reflect a certain level of commitment to the relationship, and this commitment does not necessarily extend to all loyalty behaviours. It is likely that numerous emotional responses as well as pragmatic reasoning take precedence depending on the type of loyalty behaviour in question.

The tests for moderating effects offer further evidence to support such an argument. Controlling for the effect of satisfaction enabled this study to account for those who stay in the long term due to reasons other than satisfaction, such as switching barriers. These results indicate that while there is a small, but significant increase in level of spend over time, a comparison of highly satisfied new and long-term customers show no significant differences (0.05 level) in their levels of spend. Results are similar as far as level of sensitivity to price increases is concerned. Furthermore, duration of stay appears to have a significant moderating effect on the relationship between satisfaction and WOM, suggesting that, for a given level of satisfaction, new customers are more likely to make both positive and negative WOM, than long-term customers. These findings further reinforce the idea that when it comes to WOM, whether positive or negative, and for a given level of satisfaction, it is the new customers who are most critical to a firm. This effect is also likely to be affected by the low-contact nature of a telephone service. Other than during the initial installation period, over the long run, contact may become almost nonexistent unless due to some critical incident. As such, it is plausible that even satisfied customers tend to forget about the service all together. Therefore, it is argued that, among other things, the greater propensity for critical incidents and the ‘recency’ effect during the initial relationship period, explain the higher level of WOM among new customers. Such an explanation could clearly account for the observed phenomenon. If new customers give more WOM than the long-term customers, and in terms of level of spend, and price sensitivity, satisfied long-term customers are only as good as satisfied, new customers, the greater profitability of the so-called ‘loyal’ customers appears to be overstated. Indeed, out of a sample of 432 customers, not a single customer who fit Reichheld and Sasser's¹ definition of a ‘loyal’ customer could be found.

CONCLUSIONS

Overall, this paper illustrated that the results, although confirmatory in some ways, are also intriguing and counter-intuitive in other ways. The results have clear implications on both theory and practice. The premise that having long-term satisfied customers is the best predictor of having profitable customers was shown to be too much of a generalisation. While previous research has suggested results similar to current results in noncontractual settings, it was widely believed that in subscription-type settings, long-term satisfied customers are likely to be more profitable than new customers. The indicators of profitability captured by the so-called ‘loyalty behaviours’, however, illustrate that such a view could be tenuous. The results therefore add an important cautionary note to firms that make significant investments in building long-term customer relationships based on the premise that such programmes are a panacea to profitability problems.

Indeed, it needs to be acknowledged that this research did not measure profitability per se, and that some of the constructs measured were behavioural intentions, rather than actual behaviours. The indicators of profitability, however, included some objective measures such as level of spending. Therefore, while the results do not necessarily suggest that having long-term, satisfied customers will lead to lower profitability, especially given the significant costs of acquiring new customers, they do suggest that claims about the higher profit potential of ‘loyal’ customers made in the literature are likely overstated.