Advertising agency retention – Views from South African advertisers

Commercial reality demands long and lasting relationships that are beneficial to both advertising agencies and their clients. This article presents a conceptual model to illustrate the salient choice criteria employed by advertisers when renewing advertising agency contracts. The model was empirically tested, and data obtained from 116 respondents were subjected to factor and correlation analysis to explore the dimensionality of the retention construct. Research results support that retention is multidimensional and highlights the limitations of research that measure single factors related to retention. The model postulates that retention is the result of a good working relationship, service satisfaction, respect and support, clear terms of engagement; and affective conflict resolution. Insight into the nuances of each of these antecedents is also offered. In management practice, this model will provide valuable insight to agencies with regard to client retention management. Findings of this study may also be extended to other high-value professional service industries.


Introduction
Banks around the world are nationalised, economies are in recession, and in South Africa, debt levels are rising along with inflation, while economic growth and consumer spending have slowed down (Furlonger, 2008). The impact of this economic downturn pose challenges to marketers due to budget cuts and marketing staff retrenchments. This is illustrated in a study of USA advertisers (Scanlon, 2009) where the majority (93%) were affected by cost cutting and almost 39% reported budget cuts greater than 20%. In South Africa, the conditions for marketers are also getting tougher. Although the advertising spend increased by 4.9% year-onyear during 2008 (The Mediashop, 2009), real growth, once inflation is factored in, actually showed a decline (Furlonger, 2008).
Even though experts are advising against advertising budget cuts during economic downturns (Field, 2008) some advertisers have no alternative. According to Furlonger (2008), advertising agencies confirm that clients (advertisers) are getting nervous and require agencies to stretch their budgets even further. If advertisers cut their advertising spend, resultant agency billing decreases. Since few advertisers represent a high concentration of revenue and small client numbers are indicative of this industry, tough times for advertisers lead to even tougher times for agencies. Consider, for example, the advertising spend of three major accounts served by The Jupiter Drawing Room Johannesburg, a South African advertising agency. During 2007, MTN spent R438 million, Absa spent R283.5 million, and Sasol R76.9 million (AdFocus, 2008). The Jupiter Drawing Room employs 226 people to oversee their 14 clients. As advertising agency compensation (billing) is mostly determined against a cost-based model that rewards agencies on the basis of hours spend per project (Williams 2008) billing linked to these advertisers are all significant to the turnover of this agency. Commercial reality therefore demands long and lasting relationships that are beneficial to both agencies and their clients.
In this article, retention of advertisers by agencies will be probed to gain an understanding of how advertisers' preferences and choices vary with their experience of a particular advertising agency. This article is structured as follows. First, retention constructs originating from service marketing and business-to-business marketing will be reviewed. Subsequently, the article reports on the results of a study that was conducted to empirically test variables to develop a conceptual model to illustrate the salient choice criteria employed by advertisers when renewing advertising agency contracts. The article concludes with a discussion of the theoretical and managerial implications and directions for future research.

Theoretical conceptualisation
Retention is seen as the advertiser's decision to continue the existing business relationship with an advertising agency. Theoretical positions relating to customer retention emerged from three main perspectives, namely service marketing, business-to-business (B2B) marketing and general management (Ahmad & Buttle, 2002). From the service marketing perspective, customer retention has been conceptualised as a consequence of customer-perceived service quality and customer satisfaction (Venetis & Ghauri, 2004;Ahmad & Buttle, 2002). From the B2B marketing perspective, the way to retain customers is through close and enduring relationships or creating barriers that could obstruct relationship termination (Ahmad & Buttle, 2002;Turnball & Wilson, 1989) and providing customer value through augmented products and services (Ganesan, 1994). General management argues the importance of customer retention to guide business strategies as opposed to service and B2B marketing that indicate the constructs that could lead to retention. Success in business is therefore dependent on clear thinking on the fundamental conceptual issues of retention (Aspinall, Nancarrow & Stone, 2001). From a conceptual and operational viewpoint, it may be that no single variable should be regarded as representing retention. The composition of the construct of retention therefore warrants further investigation.

Retention constructs originating from service marketing
Advertising agencies are service organisations that specialise in planning and executing advertising programmes for their clients (Kallmeyer & Abratt, 2001). In order to remain competitive, agencies are recommended to provide an appropriate range of services that are valued by clients, successfully manage their account teams and charge competitive rates (Palihawadana & Barnes, 2005). As these requirements are considered to be qualifying dimensions for agency appointment, this article will consider service quality and performance as determining criteria for future relationships consideration.
According to transaction-cost economics and the distributive justice theory of equity, advertisers should respond in proportion to the expected value from their future exchanges attributed to their agencies (Davies & Palihawadana, 2006). Expected value, in turn, is drawn from previous exchanges with service output and performance. Service quality, leading to perceived value, is therefore determined by service output and performance (De Ruyter, Moorman & Lemmink, 2001). When agencies deliver consistent service quality, clients often expect future value from the relationship that can help to reduce switching (Bolton, Lemon & Bramlett, 2004), thus leading to agency retention. Providing high levels of service quality is an essential strategy for agencies for success in today's competitive environment (Caceres & Paparoidamis, 2007). According to Davies and Palihawadana (2006), the most influential sources of account dissolution have been attributed to clients' perceptions of dissatisfying service quality based on either creativity or the quality of working relationships. Studies of client-advertising agency relationships further reveal the importance of performance dimensions associated with service quality. Performance indicates the quality level at which the primary characteristics of the product or service function (Kotler & Keller, 2006). In both the domains of production and in actually creating service performance, the client and the agency need to work closely together (Woonbong, Marshall & Son, 1999).
Service satisfaction is closely associated with service quality and performance. As the primary tool for managing customer retention, customer satisfaction has received unflagging attention in the marketing literature (Szymanski & Henard, 2001). Companies around the world have adopted customer satisfaction measures as a de facto standard for monitoring progress, motivated by the belief that customer retention and profitability will follow (Burnham, Frels & Mahajan, 2003). While the literature contains significant differences in the definition of satisfaction, all the definitions share some common elements as highlighted by Griese and Cote (2000) namely: 1) consumer satisfaction is a response (emotional or cognitive); 2) the response pertains to particular focus (i.e. expectations, product, and consumption experience); 3) the response occurs at a particular time (i.e. after consumption, after choice, based on accumulated experience).
Service quality, performance and satisfaction are therefore constructs to be considered by advertisers in their decision to retain advertising agencies.

Retention constructs originating from businessto-business marketing
Perspectives that originate from B2B marketing add to the existing literature and indicate that retention is the outcome of close and enduring relationships or the result of procurement realities such as dependency and/or switching barriers.

Client relationships
One of the most critical elements in B2B markets, and particularly in a service market such as the advertising industry, is the development of client relationships. Following an annual American-based survey on the relationships between advertisers and agencies, a senior agency executive concluded "There are no bad advertising agencies, only bad agency-client relationships" (Elliott, 2006). Although a plethora of work exists regarding the initiation of agency-client relationships, significantly less focuses on the nuances associated with nurturing and developing ongoing relationships (Palihawadana & Barnes, 2005). According to Davies and Palihawadana (2006), indicators of relationship strength comprise of commitment, trust, collaboration, cooperation and satisfaction. As satisfaction was already addressed in the previous section, the discussion that follows will focus on the first four indicators.
Relationship commitment exists where a partner believes the relationship is important enough to warrant maximum efforts at maintaining that relationship in the long term (Caceres & Paparoidamis, 2007). Several different motivations can underlie this intention, leading to two different types of commitment: affective and calculative commitment (Han, Kwortnik & Wang, 2008). Both types reflect relatively stable attitudes and beliefs about the relationship but stem from different motivations for maintaining a relationship. This twofold character of commitment is best described in the words of Johnson (cited in Söllner 1999:222), "People stay in relationships for two reasons: because they want to and because they have to." The motive underlying affective commitment is a generalised sense of positive regard for and attachment to the other party. On the other hand, calculative commitment stems from an anticipation of high termination costs associated with ending the relationship (Gounaris, 2005).
The second indicator of a relationship's strength is trust. The degree of trust that develops between companies has been described as a "fundamental relationship building block" and a "critical element of economic exchanges" (Gounaris, 2005:127). Gounaris (2005:128) conceptualised trust as "the confidence of the exchange actors in the goodwill of each other". Trust encompasses two essential elements -trust in the partner's honesty or credibility and trust in the partner's benevolence. Honesty refers to the belief that one's partner stands by his/her word, fulfils promised role obligations, and is sincere. Benevolence reflects the belief that one's partner is interested in the firm's welfare and will not take unexpected actions with negative consequences (Geyskens, Steenkamp, Scheer & Kumar, 1996).
The third and fourth factors that influence the relationship strength are collaboration and coordination. Collaboration and coordination between companies can facilitate both strategic and operational focus, increasing the potential for cross-enterprise gains (Daugherty, Richey, Roach, Chen, Arndt & Genchev, 2006). Inter-organisational collaboration focuses on sharing of information, joint development of strategic plans and synchronising operations (Daugherty et al., 2006). Cooperation refers to similar or complementary coordinated actions taken by companies in interdependent relationships to achieve mutual outcomes or singular outcomes with expected reciprocation over time (Anderson & Narus, 1990). Cooperation and collaboration therefore do not differ in terms of whether or not the task is distributed, but by virtue of the way in which it is divided. This article will consider these constructs simultaneously by investigating account support, communication, and conflict harmonisation.
Account support should be an important element in relationship management as advertisers consider the "quality of people assigned to the account" as a critical attribute in the overall agency evaluation/selection process (Cagley & Roberts, 1984). Communication refers to the formal as well as informal sharing of meaningful and timely information between companies (Anderson & Narus, 1990). According to De Ruyter et al. (2001), communication is considered when relations are evaluated by customers. Communication furthermore provides important input to customer commitment. Lastly, conflict may occur in relationships as a result of disagreement or perceived impediments of the attainment of mutual goals and objectives. Although conflict can have a negative effect on relationships, solving conflicts constructively may actually strengthen inter-organisational relationships and can thus lead to greater trust and affective commitment (De Ruyter et al., 2001). Enduring relationships are however not always the result of positive cognitive choices and may also be the result of procurement realities as discussed in the next section.

Procurement realities
Inter-organisational exchange relationship theory suggests that exchange partners can become attached or committed to each other, either from the lack of pressures for change or from the blocking of such pressures. Consistent with the latter view, attachment has been defined as a binding force between exchange partners that can lead to the maintenance of an existing relationship to the exclusion of alternatives (Nielson, 1996). Companies today are increasingly focusing their strategic efforts on creating such attachments or actions that will have an impact on a customer's decision to remain with a service provider.
One such strategy is the introduction of switching barriers. Switching barriers represent any factor, which makes it more difficult or costly for advertisers to change providers (Jones, Mothersbaugh & Beatty, 2000). As switching barriers make customer defection difficult or costly it could foster greater retention. The cost associated with switching can either be financial or psychological in nature (Jones et al., 2000). In many cases, a failed relationship means that the client pays in the end not only in cost but through anguish over the break, followed by the delays involved in the new agency selection process, time-consuming induction briefings, and the gradual, tedious development of rapport and trust with the new agency (Michell, 1987). Burnham et al. (2003) developed a useful switching-cost typology considering procedural, financial and relationship switching costs useful to this study.
Relationships may also be maintained as a result of dependency. Dependence of an advertiser on an agency refers to an advertiser's need to maintain the relationship in order to achieve the desired goals (Ganesan, 1994). Companies with greater relative dependence have, by definition, relative greater interest in sustaining the relationship (Hunter, Bunn & Perreault, 2006). In comparison, buyer power is the advertiser's ability to command favourable outcomes in the supply market, including terms of sale, seller concessions and transfer of risk to the agency (Hunter et al., 2006). Dependence is thus determined by the extensiveness of the choice set, the purchase importance and existing buyer power.
In conclusion, although the initiation of agency-client relationships is well researched, there is significantly less focus on the nuances associated with nurturing and developing ongoing relationships (Palihawadana & Barnes, 2005). Despite considerable research that has focused on the attributes associated with evaluating and selecting advertising agencies, and an emerging body of literature emphasising customer loyalty and retention, advertising agency-client investigations have often analysed these attributes in isolation. This created the need for models that represent the interrelated effects that engender retention, especially for services, whereby evaluative as well as relational factors can influence buyer responses. Such a model has the potential to provide a holistic and integrative perspective to client retention. It could also offer a framework to advertising agency executives to manage retention and optimally allocate resources for maximum customer equity.

Objective of the research
The objective of the research reported on in this study was to provide a holistic and integrative perspective to client retention. The research question to be answered was: Which salient choice criteria are employed by South African advertisers when renewing advertising agency contracts?
Following a positivistic paradigm, quantitative research methods were employed. Causality was established by measuring South African advertisers' perspectives using a survey approach.

Research methodology
A quantitative research methodology was used and data was collected with a survey approach. This approach is appropriate as the reasons for retention are classified as latent variables that cannot be directly observed and which should therefore be inferred from a group of indicators (Page & Meyer, 2000). In the case of this study, respondents were uniquely qualified to provide the desired information by virtue of their past experience, and respondent selections followed clear respondent profile compliance. Data was collected in this study by means of a structured questionnaire administered via a web-based survey.

Sample
The population of the study on which this article is based, comprised of South African advertisers who employ advertising agencies for advertising services, including below-the-line activity, media planning and buying. This sector provided an accessible example of agency-advertiser relationships, where evaluation occurs relatively frequently. Due to a variety of characteristics displayed by this population, a minimum transactional value of R500 000 was set as a population parameter to identify an appropriate sample frame. Contracts with a high transactional value assume that agencies would mostly strive to build close and long-lasting relationships and that customer retention of these accounts would be of particular importance to agencies. Due to the high transactional cost, it was further to be expected that advertisers would employ accountable appointment processes and would therefore undertake formal and elaborate pre-purchase studies.
The sample frame was provided by List Perfect who provided, by industry standard, the best and most up-to-date database of corporate companies whose advertising budgets exceed R500 000 annually. The database contained information about 743 companies that was set as the target population. The size of this relevant target population suggested that a census was feasible. The size of the target population was decreased after 57 (8%) of the companies indicated that they did not employ an advertising agency but produced their own advertising material and other related services in-house; and 12 (2%) of the companies indicated that their international head offices were responsible for the appointment and relationship with their advertising agencies. This could imply that the sample frame overestimated the sample units suitable for the study. The target population was therefore reduced to 674 companies, albeit the possibility that this could still be an overestimation. Hundred and twenty (120) respondents submitted their surveys online of which 116 were considered suitable because they have completed at least 90 per cent of the questionnaire items. The response rate for this survey was therefore 17.8% and deemed to be representative of the population.

Measurement instrument
Following a literature review, constructs or dimensions that could explain the advertiser's decision to retain an advertising agency were identified. Constructs with a previous proven positive relationship to retention in a service context were considered. Table I provides a summary of the constructs and driving forces considered. Measurement items associated with the constructs listed in Table I were obtained from the listed references. Items were selected by considering construct reliability and contextual alignment and scales that showed internal consistency were used.
To insure nomological, discriminant, convergent and face validity of the measurement instrument, the measurement items were refined and developed through the research process, and pilot-tested in the specific context of this research. Pilot testing obtained the opinions of a group of industry experts (four) on the representativeness and suitability of the questions. The questionnaire was also reviewed by a qualified statistician from a data perspective, marketing colleagues (four) from a respondent perspective and lastly members (three) of the information technology department from a technical perspective.
Rating questions were mostly used to collect opinion data. A four-point Likert rating scales was used to enable respondents to make a definite choice rather than choose neutral or intermediate positions on a scale (Garland 1991).

Data analysis
The online questionnaire automatically entered and saved the data to a computer file which was exported into SPSS in order to perform statistical analysis. Most data obtained were classified as ordinal qualitative data. Multivariate statistical techniques afforded the researcher the opportunity to gain greater insight into relationships and patterns of the marketing issues under study (Cooper & Schindler, 2001). In order to measure the association between two variables, the analysis considered Kendall's Tau B, Phi and Cramer's V (Saunders, Lewis & Thornhill, 2007). Next, factor and correlation analysis were used to identify representative variables to present a consideration set appropriate for agency retention. Correlation analyses considered Pearson's correlation coefficients to assess the strengths of relationships between variables and to calculate the level of significance. Only variables with significant strong positive relationships were considered (r ≥ .5 and p < .05) in the development of new factors. The purpose was to retain the nature and character of the original variables, but to reduce their number.
Finally, factor analysis was used to define the underlying structure in the data matrix and to summarise the data as well as for data reduction (Hair, Black, Babin, Anderson, & Tatham 2005:90). To confirm the suitability of the variables contained in the correlation matrix and the significance of all correlations, the Kaiser-Meyer-Olkin (KMO) measure of sampling adequacy and Bartlett's test of sphericity were calculated.
In the case of both correlation and factor analysis the reliability of the new factors was measured and scales/factors with a Cronbach's alpha higher than 0.5 were accepted. While a Cronbach alpha of 0.7 is usually considered a cut-off point this requirement may be reduced in exploratory research within social sciences and humanities. In this regard, Kent (2001) suggests that the minimally acceptable reliability should be in the range of 0,5 to 0,6. One-way analysis of variance (one-way ANOVA) was used to analyse the variance within and between groups of data by comparing means (Saunders et al., 2007).

Results
The results will report on the respondent profile, data quality, and the salient choice criteria that are employed by South African advertisers when renewing advertising agency contracts.

Respondent profile
Analysis of the data revealed that the majority of respondents were appointed in a senior marketing position for more than four years. On average the advertising agencies have been appointed for a period of five years and 70% of the respondents indicated that they would continue with the current relationship if given the choice. Most of the respondents (62%) were involved in the appointment of the current advertising agency.

Salient choice criteria
The next section reports on the results obtained after constructs were empirically tested to determine the effects that constructs have on one another and their correlation to, and association with retention. This was done by testing variables with a proven positive relationship to retention within the research context (South African advertising industry) through factor and correlation analysis. In doing so representative variables was identified from a much larger set of variables and data was reduced to a smaller number of variables for inclusion in the final consideration set. The purpose was to retain the nature and character of the original variables, but to reduce their number.

Constructs removed from the retention consideration set
Dependency was removed from the consideration set as correlations to and associations with retention are almost non-existent. Switching barriers were considered next. The literature review suggested that switching barriers based on procedural, financial and relationship-switching costs would be appropriate to this study. The data matrix of these measurement items showed sufficient correlations to justify the application of factor analysis. This was done to establish whether the measurement items display similar factors within the research context of this study. The results indicated that three factors should be retained to explain switching barriers. These factors included switching uncertainty ( = 0,835), switching effort ( = 0,766) and switching time ( = 0,767). The next step was to determine whether these factors were correlated to and associated with the dependent variable (retention) of this study. In all cases, the association between the factors and the propensity to reappoint were close to total independence and Pearson correlations indicated that the associations were insignificant. Analysis of variance (ANOVA) furthermore indicated that there were no significant differences for any of the factors between groups displaying different levels of retention propensity. Therefore, even though most respondents agreed that switching barriers are present when changing agencies, the data analysis confirmed that these barriers do not lead to customer retention. These barriers were subsequently removed from the consideration set.

Client satisfaction
Client satisfaction was described by 16 measurement items. Initial analysis on these items indicated that partial correlations between all of these items were significant and high, indicating that there was only one underlying factor. Factor analysis was therefore deemed inappropriate (Hair et al., 2005). All the variables were thus considered important to describe a general client satisfaction measure. Measurement items were combined to create a new scale to measure client satisfaction and exhibited internal consistency reliability. Considering item total statistics (Table II), it is evident that the factor client satisfaction is reliable and composite of 16 measurement items. The next step was to determine whether client satisfaction is correlated to and associated with the dependent variable (retention) of this study. The Pearson correlation value of 0.546 showed that the association between client satisfaction, and the propensity to re-appoint was strong. The significant level of less than 0.0001 indicates that the association is significant. Next ANOVA was used to assess the likelihood of any significant difference between different (three) groups displaying varying levels of retention. Results indicate that the F ratio value is 23.272 with 2 and 108 degrees of freedom (df) and has a probability of occurrence by chance alone of less than 0.001 between the three groups. There is thus a significant difference between groups displaying different levels of retention propensity. The relationship between client satisfaction and client retention is indicative of a strong positive association between customer satisfaction and the contract renewal. This variable should therefore remain part of the consideration set.

Service output
Service output was initially described by 17 measurement items. Once data screening was completed, the data matrix had sufficient correlations to justify the application of factor analysis (KMO=0,845, Bartlett's Test of Sphericity with p < 0,001). Three factors were obtained based on Eigenvalues (> 1) as indicated in Table III. Factor one was labelled 'respect and support' as the seven variables displaying high loadings all relate to elements associated with respect and support. Factor two was labelled 'clear terms of engagement' as the two variables displaying high loadings indicated the verbalisation of the terms of the relationship, and lastly, factor three was labelled 'conflict resolution' as the three variables that display high loading relate to the manner in which conflict is dealt with.

Respect and support
Respect and support is the result of seven variables and the factor indicates internal consistency reliability (Cronbach's  = 0.866). The next step was to determine whether respect and support is correlated to and associated with retention. The Pearson correlation value of 0.475 showed that the association between respect and support and the propensity to re-appoint is moderate. The significant level of less than 0.0001 indicates that the association is significant. ANOVA results show that there is a significant (F = 15.719 with 2 and 108df, p < 0.05) difference between groups displaying different levels of retention propensity. Respect and support is therefore an underlying factor of service output and based on the data analysis, this factor should be part of the consideration set model.

Clear terms of engagement
Clear terms of engagement is the result of two variables and the factor indicates internal consistency reliability (Cronbach's  = 0,859). The Pearson correlation value of 0,322 indicated that the association between 'clear terms of engagement' and the propensity to re-appoint was moderate. The significant level of less than 0,0001 indicates that the association is significant. The ANOVA analysis shows that there is a significant difference (F = 15,047 with 2 and 108df, p < 0,05) between groups displaying different levels of retention propensity. Clear terms of engagement is therefore an underlying factor of service output and, based on the data analysis, this factor should be part of the consideration set model.

Conflict resolution
Conflict resolution comprises of three variables and the factor indicates internal consistency reliability (Cronbach's  = 0,547). The Pearson correlation value 0,302 showed that the association between conflict resolution and the propensity to re-appoint was moderate. The significant level of 0,001 indicates that the association is significant. ANOVA shows that there is a significant difference (F = 5,445 with 2 and 108df, p < 0,05) between groups displaying different levels of retention propensity. Conflict resolution is therefore an underlying factor of service output and based on the data analysis, this factor should be part of the consideration set model.

Client relationship
Due to an overlap of measurement items associated with commitment, trust, and communication, these constructs were merged into one concept. This allowed for the analysis of patterns of a multidimensional relationship within the research context. Initially, all the measurement items associated with the constructs listed above were included as variables for factor analysis, regardless of initial variable groupings. As a result of data screening to avoid multicollinearity and singularity, six items of the initial 15 measurable items were removed. After this, the data matrix had sufficient correlations to justify the application of factor analysis (KMO = 0,819, Bartlett's Test of Sphericity with p < 0,001). Only one factor had obtained an Eigenvalue greater than 1 as indicated in Table V.
The factor identified was labelled good working relationship, and was the result of eight variables that displayed high loadings as indicated in Table VI.   Extraction method: Principal axis factoring. This factor indicated internal consistency reliability (Cronbach's  = 0.879). The next step was to determine if good working relationship is correlated to and associated with the retention. The Pearson correlation value of 0.596 shows that the association between good working relationship and the propensity to re-appoint was strong. The significant level of less than 0.001 indicates that the association is significant. The ANOVA analysis shows that there is a significant difference (F = 30.419 with 2 and108df, p < 0.05) between groups displaying different levels of retention propensity. Good working relationships should be part of the consideration set model.

Agency retention consideration set
In the study on which this article was based, the subject of advertiser retention was probed to gain an understanding of how advertisers' preferences and choices vary with their experience of a particular advertising agency. In response, a conceptual model to illustrate the salient choice criteria employed by advertisers when renewing advertising agency contracts was developed. This model, illustrated in Figure 1 below, provides a holistic and integrative perspective on client retention and also a framework to advertising agency executives to manage retention and optimally allocate resources for maximum customer equity. What started as a very complex set of relationships has been simplified into a parsimonious model. This model postulates that retention is the result of a good working relationship between an advertiser and agency, service satisfaction, respect and support, clear terms of engagement, and effective conflict resolution.

Good working relationship
A good working relationship is an essential element of client retention. The study on which this article reports, offered insight into nuances associated with nurturing and developing ongoing relationships and found that in order to foster a good working relationship with agencies, advertisers expect agencies to:  be committed to the relationship  enjoy working together  be sincere  put maximum effort into the relationship  verbalise the terms of the relationship  be concerned about their welfare  have matching philosophies  verbalise the terms for sharing information Advertisers therefore expect that agencies will display real commitment to relationships from a corporate and staff level.

Service satisfaction
The research further found client satisfaction is indeed indicative of a strong positive association with contract renewal and thus client retention. Satisfaction dimensions (service output and performance) that are deemed important to retain clients are: Service satisfaction is closely associated with service quality and performance and should consider both dimensions.

Respect and support
Factor analysis was applied to measurement items measuring service output. A new factor labelled respect and support was identified and further analysis confirmed a correlation and association with client retention. Respect and support are present when the advertiser:  can count on the agency to be sincere and  is satisfied with the quality of the agency personnel.
In addition, the advertiser expects the agency to:  verbalisation and discussion of the basic terms of the relationship and  explicit verbalisation and discussion of the basic terms for sharing information between the client and advertising agency.

Conflict resolution
The last factor that was developed as a result of factor analysis applied to measurement items measuring service output was labelled conflict resolution. Further analysis confirmed a correlation and association with client retention. Conflict may occur in relationships as a result of disagreement or perceived impediment of the attainment of mutual goals and objectives. Although conflict can have a negative effect on relationships, solving conflicts constructively may actually strengthen inter-organisational relationships and lead to greater trust and effective commitment (De Ruyter et al., 2001). Conflict resolution occurs when:  conflict is seen as productive discussion  disagreement improves productivity and  agencies give advertisers the benefit of doubt.
Although conflict resolution is not a new factor, this study confirmed that conflict resolution is indeed a retention consideration. Insight is further provided to agencies on clients' expectations regarding the approach to conflict resolution.

Conclusion
Marketing academics and practitioners generally agree that customer retention is vital to business success. There is less agreement on factors that determine customer retention, particularly in service contexts. The result of the current research suggests that a multi-dimensional model of retention as illustrated in Figure 1 is more appropriate than considering single-dimensional factors. The model was tested within the South African advertising industry, which provided an accessible example of agency-advertiser relationships. The research contributes to the literature by providing a more complete, integrated view of customer retention and its determinants in service contexts.
From a practitioner's perspective it is important to note that agencies need to manage challenges present in the industry. The South African advertising industry is considered to be a buyers' market and clients expect nothing but the best. Despite the fact that advertisers agree that switching between agencies creates uncertainty and requires time as well as effort, these barriers will not lock them into a relationship or lead to client retention. Client retention is not only an outcome based on agency effort and performance but rather the result of interactional quality dimensions. It can therefore be concluded that mutual effort is required to achieve client retention. Furthermore, it should be highlighted that a good working relationship is a prerequisite for service satisfaction. The model presented in Figure I can therefore be used to measure and manage business relationships between advertisers and agencies. This will improve the business relationship for both parties and can be used to strengthen and build long-term relationships.
From an academic perspective the research offers insights into the structure of client retention for services. This research has developed a comprehensive theory-driven retention model. A contribution emanating from the study is the use of multi-item scales to measure and test retention. This research supports that retention is multi-dimensional and highlights the limitations of research that measures single factors related to retention. Furthermore, the study produced two new factors to consider when retention is measured, namely support and respect as well as clear terms of engagement. The research therefore contributes to the literature by providing a more complete, integrated view of customer retention and its determinants in service contexts.
As this model was developed for the South African advertising industry, a potential first limitation may be presented when applied to other parts of the world or other service industries. Secondly, the sample size prohibited the utilisation of multi-varied modern techniques such as structural equation modelling. Lastly, the cross-sectional design of the study presented a limitation as retention is dynamic and the relative influence of antecedent factors evolves as relationships mature. Future research would be useful to extend the model to other service industries both locally and globally as the generalisation of this model would be enhanced by replication in other settings.