This article reports on research conducted using the Critical Incident Technique to explore the nature of service failure and associated service recovery strategies used in hotels. Using a questionnaire-based approach, data were collected in hotels in both the UK and China; in total some 79 separate incidents were collected. Using a classificatory schema derived both from previous research and the incidents collected in this study, it has been possible to explore the relationships between service failure and service recovery. The findings show that hotels need to clarify their policies and procedures with regard to service recovery if they are going to be successful in turning the dissatisfaction deriving from a service failure into the possibility of customer retention. Even with clear recovery strategies, the research suggests that only half of customers experiencing a service failure of any type will return to the same hotel.

**********

With increasing competition at all levels of the hospitality industry, managers are increasingly aware of the importance of keeping existing customers as well as winning new ones. Despite efforts to deliver consistent service to their customers, due to the nature of service operations it would appear that breakdowns in the delivery of service or service failures (Spreng, Harrell, & Mackoy, 1995) are inevitable (Hart, Heskett, & Sasser, 1990).

While service failures may be inevitable, losing customers following these failures is not. Service recovery actions can be taken that may repair all or some of the damage done. The customer's final perception of the quality of the service provider will be affected by the number and seriousness of the problems they encounter and how these problems are handled by the operation (Colgate & Norris, 2001). Indeed there are some claims that customers may rate the encounter more favourably after a failure has been corrected than if the transaction had been correctly performed the first time (Kelley & Davis, 1994).

On the other hand, service failures that are not properly recovered can result in a decline in customer confidence, lost customers, negative word-of-mouth, potential negative publicity, employee dissatisfaction and the direct costs or reperforming the service (Berry & Parasuraman, 1992; Lewis & Clacher, 2001).

To satisfy customers, hotel companies must both provide good products and deliver excellent service. Although the standard of the physical facilities and equipment that hotels provide has been seen to improve over the last few years, there is a widely-held belief that it is the level of service provided that distinguishes successful hotels from unsuccessful or mediocre ones (Morrison, 1996).

Purpose and Objectives

The purpose of the research reported here was first to identify and classify service failures occurring within the hotel context and to assess customers' perceptions of the seriousness of these failures and their effect on customer satisfaction.

The second objective was to identify and classify the recovery strategies that had been used to redress the service failures encountered and to assess the effect of these strategies on customers' subsequent satisfaction and intended purchase behaviours.

The relationships between these key variables were also to be investigated and tested.

Data Collection

The research methodology chosen to achieve these objectives was the Critical Incident Technique (CIT), first developed by Flanagan (1954) and since widely used and supported as suitable, effective and reliable for this type of research (Lockwood & Gilbert, 1999).

For the purposes of this study, a questionnaire-based approach was adopted. This approach has advantages in terms of the level of detail that can be collected along with simplifying and standardising the data collection process. However, the response rate for such questionnaires delivered by mail is known to be low (Lockwood & Gilbert, 1999) and so the questionnaires were delivered by hand. With the permission of hotel management, guests were approached on a random sampling basis to identify if they had had any experience of service failure within the last 12 months and, if so, whether they would be interested in participating in the research.

The Critical Incident Questionnaire used in the research was developed from that first used in retail contexts by Kelley, Hoffman, and Davis (1993) but with changes to the content and layout to adapt it to the specific context and purpose of this research. The questionnaire was designed in four parts:

* Part 1--to collect information about failure incidents. Respondents were asked to tell of a personal experience of service failure including where and when the hotel stay was and its purpose. Two 5-point scales were also included to evaluate the significance and the respondent's level of dissatisfaction with the service failure.

* Part 2--to collect information about the recovery strategy used by the hotel to correct the mistakes and what the hotel should have done, in the guest's opinion. The efforts of the hotel staff to correct the situation were also evaluated with a 5-point scale.

* Part 3--to collect information about customers' perceptions of the recovery strategy employed by the hotel. The customers' satisfaction after recovery was evaluated using a 5-point scale. Their subsequent intentions to recommend the hotel and return were also tested on a yes/no basis.

* Part 4--to collect demographic data on frequency of stay and frequency of experience of failure in addition to nationality, gender, age and occupation background.

The questionnaire came to three pages in length but with a large amount of white space for completion. It took around 20 minutes to complete fully. Following successful pilot testing of the instrument, over a 2-week period, customers in two hotels, one in the United Kingdom (UK) and one in China who bad recent experience of a service failure within a hotel, were asked on a random sampling basis to complete a critical incident questionnaire. Overall, a total of 79 separate incidents were collected and analysed.

Results

Demographics

The demographics of the sample were as follows.

Some 56% of the respondents stay in hotels at least once or twice a month. In this section, respondents were also asked to indicate how frequently they encountered problems when staying in hotels. Twenty-five percent of respondents reported that they very rarely encountered problems and 32% rarely encountered them. A further 34% thought it was common for them to encounter problems. Only 9% reported frequent problems and no respondents reported very frequent problems. This would seem to suggest that service failures do occur in hotels, but for the majority this is a fairly uncommon experience.

Service Failure

The categorisation of service failures at the top level followed that developed originally by Bitner, Booms, and Tetreault (1990). These top-level categories are: first, core service failure (where the customer does not receive the basic service promised by the provider) which accounted for 72.5% of the failures reported; second, responses to implicit and explicit customer requests which accounted for 10% of the failures reported; third, employee error (incidents caused by inappropriate action or behaviour on the part of hotel employees) which accounted for 17.5% of the incidents reported. These top-level categories were further subdivided according to the nature of the incidents reported (see Table 2).

From Table 2 it can be seen that the most common problems were caused by equipment failure, such as a lack of hot water or key card failures, accounting for over a quarter of all reported incidents. This was followed, at nearly 19%, by problems with hotel reservations, such as overbooking or misallocated rooms, and then, at 15%, by employee errors of action, such as carelessness or mischarging. The least commonly reported incidents were concerned with inappropriate employee attitudes, customers' own mistakes and responses to customer requests.

The reported level of significance of the different failure categories demonstrates that failure to respond to a customer request were rated the highest at 5, closely followed by cleanliness, customer error and employee attitude at 4.5. The incidents with the lowest level of significance were reported in the categories of employee mistakes, disruption caused by other guests and slow or unavailable service.

The highest levels of customer dissatisfaction prior to attempts at service recovery included a lack of response to customer requests followed by slow or unavailable service, customer mistakes and employee attitude. Least dissatisfaction was caused by other disruptive customers, and employee carelessness. Considering the customers' level of satisfaction after attempts at service recovery, customers were still clearly dissatisfied with responses to their own errors and to slow or unavailable service. However, higher levels of satisfaction were reported after recovery for dealing with other disruptive guests, responding to customer requests, and employee attitude. The highest improvements in satisfaction were recorded in customer requests, employee attitudes, and dealing with reservation problems. The lowest improvements in satisfaction were in the categories of dealing with customers' own errors and a lack of correct information about the hotel's policies.

Table 2 also records the percentage of customers who, having suffered a particular service failure, intend returning to the hotel. The 50% retention rate indicates that only half of customers who suffer a service failure, even if attempts at service recovery have been made, would return to the same hotel again. This contradicts previous research findings that have reported increases in customer loyalty following recovered failures. It would appear to be relatively easy to retain customers in situations involving disruptive others (where the hotel cannot be held directly responsible) and employee carelessness, where the hotel may be seen as less culpable. However, it is much more difficult to recover from issues concerning employee attitude, clarity of information on hotel policies and slow or unavailable service. Two thirds of customers would be unlikely to return following an issue with hotel cleanliness.

Service Recovery

The incidents were then analysed to identify patterns in recovery strategies and their effects on customer perceptions. The classification of recovery strategies was developed directly from the incidents collected, drawing particularly on the work of Levesque and McDougall (2000) and other research studies.

As anticipated, the most common response to a service failure was prompt correction of the error, accounting for 35% of the total incidents collected, followed (perhaps more surprisingly) by nearly 14% doing absolutely nothing. One anecdotally popular strategy--giving free meals or drinks--was actually the least popular recovery strategy accounting for only 5% of the incidents.

The significance of the failures reported by recovery strategy is more difficult to interpret as this is a measure of significance to the customer and not to the hotel, although clearly the two should be linked. Table 3 shows that giving free food was related to some of the more significant incidents and that prompt correction showed low significance. The relationships here, however, are not clear.

The recovery strategies related to the lowest levels of pre-recovery satisfaction were assistance (not actually correcting the error but offering some help and support), and managerial intervention, discounts and free food. There is some rationale here for very dissatisfied customers receiving direct attention from the manager or receiving money off the bill or a free drink or meal but why the least satisfied customers should receive next to nothing is hard to understand. This may simply be a post hoc rationalisation on the customer's part.

Levels of post-recovery satisfaction are easier to interpret. Customers were happiest following a free room upgrade and intervention from the manager, and reasonably happy with prompt correction. They were least satisfied with no action being taken, only assistance being offered; slow correction; and (surprisingly) by being offered a discount. Similarly the largest improvements in satisfaction came from free room upgrades and managerial intervention, with the lowest coming from discounts, slow correction and assistance and with doing nothing resulting in a negative improvement or decrease in satisfaction. Fail to act on a customer grievance at your peril!

Overall the level of customer retention for all recovery strategies was 50%, thus even after recovery attempts half of customers expressed an intention never to return. However, some recovery strategies are associated with higher customer retention levels than others. The highest level of customer retention came from offering a free room upgrade, which represents a very low cost to the hotel, assuming that rooms are available. The second was prompt corrective action, which would seem to support the notion of front-line empowerment to take immediate action, although wheeling out the manager in a problem situation still seems to impact significantly on the customer. The lowest levels of customer retention came from the assistance strategy and doing nothing which, surprisingly, resulted in a 27% level of retention.

Correlations

Tests of the correlation effects between variables revealed some interesting results. Not surprisingly, there was a statistically significant correlation between the perceived significance of the failure and the level of satisfaction prior to recovery. The results show a correlation of -.649 with significance higher than the 99% level. It is important therefore that hotels understand what failures their customers perceive as being significant, as correcting these will have the greatest effect on customer satisfaction scores.

Using the Kruskal-Wallis test, a nonparametric technique, tests of the relationship between the significance of the failure and the type of failure as indicated by the identified subcategory were not found to be statistically significant.

There was a very strong (Pearson .759, significance .00) correlation between the perceived level of effort of the staff involved in the recovery efforts and the level of post-recovery satisfaction. The more attentive the efforts of the employees and managers were, the higher the level of postrecovery satisfaction that customers enjoyed.

Using the Kruskal-Wallis test, the relationship between recovery strategy and satisfaction after recovery was significant at the 99% level. As shown in Table 3, the highest levels of post-recovery satisfaction were linked to free upgrades and managerial intervention, closely followed by prompt correction, while the lowest levels of satisfaction were linked to offering assistance only and doing nothing. Surprisingly, offering a discount came towards the bottom of this relationship and came lower than simply offering an apology.

The relationships between customer retention and other variables were also tested. The relationship between customer retention and the significance of the failure was strongly supported, as was that between satisfaction after recovery and retention. Those customers who perceived the significance of their service failure as lower were much more likely to return to the hotel. Those customers with high postrecovery satisfaction were also much more likely to return.

Implications

Overall, this exploratory study of service failure and recovery in a hotel context has produced some interesting and potentially important results and provides clear avenues for further research work. However, it must be recognised that this was an exploratory study with a relatively small sample size and cannot be claimed to be representative of all hotel customers, although the results could be indicative of some fruitful areas for consideration.

From a managerial perspective, to answer the question posed in the title of the article, the response must be yes, but only so far. Overall, even after the service recovery action taken by the hotel, only half of those customers experiencing a service failure would return to the same hotel--supporting the belief that dissatisfied service customers simply don't come back and contradicting the suggestion that customers who have experienced a recovered failure are likely to be more loyal than those who received good service.

There is some suggestion that certain types of failure should be treated more seriously by hotels than others. Those failures with a high significance to customers are more likely to stop them returning. In this study, these failures included not responding to a customer's request and problems with hotel reservations, closely followed by errors due to the customers' own mistakes, inappropriate employee attitude and cleanliness. Dealing with problems arising from a customer's own mistakes does seem to be a particular issue. There may be some reluctance on the part of the hotel to respond to this sort of problem but it does seriously annoy the customer if it is not handled appropriately. A denial of involvement from the hotel's perspective would seem to be a recipe for disaster.

It would appear to be relatively easy to retain customers following the actions of other customers, where the hotel cannot be held directly responsible, and following employee carelessness, again where the hotel may be seen as less culpable. However, issues concerning employee attitude, clarity of information on hotel policies and slow or unavailable service are much more difficult to recover from. Following an issue with hotel cleanliness, two thirds of customers would be unlikely to return.

In terms of recovery strategies, the most effective approach from this research would be a free room upgrade closely followed by managerial intervention and prompt corrective action. It costs very little for a hotel to upgrade a room as the marginal costs are negligible and so this seems to be a particularly clear winner, but it does depend on the hotel having free rooms to upgrade to. Whether the offer of a free upgrade on a subsequent visit would be as effective is hard to say. The idea of empowering front-line staff to take prompt corrective action without reference to a manager would also seem to be strongly supported by this research. However, having a manager who can intervene in problem situations is also highly effective.

On the down side, it would seem that doing nothing is simply not an option and yet almost 15% of the reported incidents fell into this category. Not being able to correct the problem and simply offering assistance--"I know the restaurant is closed but there is one round the corner"--is also a poor approach which will lose customers. Offering correction of a problem but then taking a long time to action it will also result in dissatisfaction and low customer retention.

The clear message here is that hotels need to review their policies and procedures with regard to service recovery and to ensure that they are not falling into the traps identified in this research.

From a research perspective, there are a number of interesting issues raised. For example, at what point should a service failure be dealt with by an employee directly and when should it be referred to a manager for their involvement? There is no doubt that an employee may be able to solve the problem straight away and that the result will be positive but there are occasions where it would seem that involving a manager may be even more effective. This clearly requires further study to arrive at an appropriate procedure.

As this was an exploratory study the most obvious need is perhaps for replications that would result in a larger sample size and may be able to test other relationships. For example, although data on gender and age were collected it was not possible to test these reliably as intervening variables. Similarly, incidents were collected in both the UK and in China with the intention of being able to identify and test differences but once again the sample was too small to allow this to be done in a reliable way.

This research has employed the technique of critical incident using a questionnaire-based approach and has proved to be successful in obtaining a high quality response that respondents have found relatively easy to complete. It has also linked the description of the incident to the respondent's perceptions and feelings about the incident and their subsequent behavioural intentions. This has allowed CIT to be used to test statistical relationships, which it would otherwise not have been able to do.

Table 1
Sample Demographics

N = 79                            Number  Percentage

Gender                  Male        48        61
                       Female       31        39
Age                     20-29       25        32
                        30-39       18        23
                        40-49       19        24
                        50-59       12        15
                         60+        5         6
National background    Western      47        60
                       Chinese      24        30
                     Other Asian    8         10

Table 2
Comparison of TCA and Organisational Capability Perspectives

                               The TCA perspective

Unit of analysis               Transaction

Primary area of focus          Transaction characteristics

Key assumption                 Opportunism

Source of competitiveness      Efficient management of transactions

Primary orientation in the     Cost minimisation
management of know-how

Key consideration to choice    TC minimization; fit between transaction
of ownership form              characteristics and form of governance

Temporal orientation           Essentially static and
                               equilibrium-oriented

                               The organisational capability
                               perspective

Unit of analysis               Firm

Primary area of focus          Firm capabilities

Key assumption                 Bounded rationality

Source of competitiveness      Development and exploitation of
                               capabilities

Primary orientation in the     Management of value
management of know-how

Key consideration to choice    Contributions towards and demands
of ownership form              placed on firm's capabilities

Temporal orientation           Essentially dynamic; learning and
                               capability building as
                               developmental processes

Source: Madhok. A., (1997). Cost, value and foreign market entry mode.
The transaction and the firm. Strategic Management Journal, 18, p. 41.

Table 3
Comparison of Different Frameworks for Studying Foreign Entry Mode

                   Transaction Cost           Dunning's Eclectic
                   Approach                   Framework

Basic theory       Transaction cost theory    Eclectic theory

Explanatory        Asset specificity,         Ownership, locational
variables          uncertainty                and internalisational
                                              advantages

Decision criteria  Transaction cost           Trade-offs between
                   minimisation               return risk, control
                                              and resources

Modes of entry     Several classifications,   Several classifications,
                   e.g., contractual          e.g., independent
                   transfer, joint venture,   mode, cooperative
                   wholly owned               mode, integrated
                   subsidiary                 mode

Past research      Williamson, (1975);        Dunning, (1980);
                   Anderson & Gatignon,       Dunning & Kundu,
                   (1986); Erramilli & Rao,   (1995); Agarwal &
                   (1993)                     Ramaswami, (1992)

                   Organisational             Unified Theory
                   Capability                 (Eclectic Theory *)
                   Perspective

Basic theory       Resource-based theory      Transaction cost theory,
                                              eclectic theory, strategy

Explanatory        Firm's capabilities        Strategic, environmental
variables          (in particular,            and transaction specific
                   know-how)                  variables

Decision criteria  Trade-offs between         Corporation of level of
                   value and cost             control, resource
                                              commitment and risk
                                              of dissemination

Modes of entry     Internalisation vs         Several classifications,
                   collaboration              (e.g., WOS, JV,
                                              licensing)

Past research      Aulakh & Kotabe,           Hill et al., (1990);
                   (1997), Madhok,            HT & Kim, (1988);
                   (1997; 1998)               Kim & Hwang, (1992)

Note: * Hill et al., (1990) named this eclectic theory in their article
but in order to avoid the confusion with Dunning a eclectic paradigm,
it will be referred to as unified theory in this article.

Table 4
Proportion of Entry Modes in Sample, Percentage

Hotel Chains                WOS   Leased  JV     MC    Franchised

Accor                      45.38    --    --    50.00     4.62
Marriott                    0.95    --    --    52.38    46.67
Hilton International       11.83  12.90   --    64.52     1.08
Inter-Continental           9.09   1.82   --    41.82    20.00
Hyatt International         --      --    --   100.0       --
Renaissance & New World     --      --    --    75.56    24.44
Four Seasons                9.09    --    --    90.91      --
Shangri-La                  --      --    --   100.0       --
Ritz-Carlton                --      --    --   100.0       --
Swissotel                  35.71    --    --    64.29      --
Tokyu                       --      --    --   100.0      --
Nikko                       8.33    --    --    91.67      --
Total                      14.38   2.20   --    66.16    13.20

References

Berry, L., & Parasuraman, A. (1992). Prescriptions for a service quality revolution in America. Organizational Dynamics, 20(4), 5-15.

Bitner, M.J., Booms, B.H., & Tetreault, M.S. (1990). The service encounter: Diagnosing favourable and unfavourable incidents. Journal of Marketing, 54 (January), 71-84.

Colgate, M., & Norris, M. (2001). Developing a comprehensive picture of service failure. International Journal of Service Industry Management, 12(3), 215-233.

Flanagan, J.C. (1954). The critical incident technique. Psychological Bulletin, 51(4), 327-358

Hart, C.W.L., Heskett, J.L., & Sasser, W.E. (1990). The profitable art of service recovery. Harvard Business Review, 68, 148-156.

Kelley, S.W., & Davis, M.A. (1994). Antecedents to customer expectation for service recovery. Journal of the Academy of Marketing Science, 22, 52-61.

Kelley, S.W., Hoffman, K.D., & Davis, M.A. (1993). A typology of retail failure and recoveries. Journal of Retailing, 69 (Winter), 425-452.

Levesque, T.J., & McDougall, G.H.G. (2000). Service problems and recovery strategies: An experiment. Canadian Journal of Administrative Sciences, 17, 20-37.

Lewis, B.R., & Clacher, E. (2001). Service failure and recovery in UK theme parks: The employees. International Journal of Contemporary Hospitality Management, 13(4), 166-175.

Lockwood, A., & Gilbert, D. (1999). Critical incident technique. In B. Brotherton (Ed.), The handbook of contemporary hospitality management research (pp. 207-232). Chichester: Wiley.

Morrison, A.M. (1996). Hospitality and travel marketing (2nd ed.). Albany, NY: Delmar.

Spreng, R.A., Harrell, G.D., & Mackoy, R.D. (1995). Service recovery: Impact on satisfaction and intentions. Journal of Service Marketing, 9(1), 15-23.

Correspondence

Andrew Lockwood, Professor of Hospitality Management and Deputy Head of School, School of Management, University of Surrey, Guildford, Surrey, UK GU2 7XH. Email: a.lockwood@surrey.ac.uk

Andrew Lockwood and Ni Deng

University of Surrey, United Kingdom