IMPACT OF DIGITAL MARKETING CAPABILITIES ON MARKET PERFORMANCE OF SMALL TO MEDIUM ENTERPRISE AGRO-PROCESSORS IN HARARE, ZIMBABWE

The importance of marketing capabilities continues to grow yet research remains concentrated in developed markets. Although several researchers provide evidence of the influence of marketing capabilities on market performance, very little of similar evidence exists in the digital marketing domain. Empirical evidence of the impact of digital marketing capabilities on market performance of small to medium enterprise (SME) agro-processors particularly from developing countries remains scarce. The purpose of this paper was to investigate the impact of digital marketing capabilities on SME agro-processors’ both intermediate and final market performance outcomes. To achieve this, we conducted a survey of 298 SME agro-processors’ managers and owners in Harare, Zimbabwe. A mixed sampling approach consisting of quota and stratified sampling approaches was adopted. We distributed a closed-ended questionnaire through the drop-off & pick-up and interviewer-based methods. The data was analysed statistically using STATA version 15. Multiple logistic regression was conducted to determine impact of digital marketing capabilities on market performance. Our findings indicate that digital strategy development & execution, digital market innovation, e-market sensing and leadership capabilities positively influence intermediate market outcomes of customer awareness, customer attitudes, availability, and brand associations. However only digital strategy development & execution capability was positively associated with final market performance outcomes of sales growth, market share and profitability. These results imply that agro-processors must develop digital marketing capabilities that enable them to move beyond intermediate market outcomes to attain the primary business objectives of profitability, sales growth and market share. Attaining intermediate market outcomes only is not enough for business sustainability. The study contributes to literature by extending the marketing capabilities discussion to the digital marketing environment in a developing country context. This was important because marketing knowledge is contextual, as such cannot easily be transferred from one market to the other.


Introduction and background
This paper extends prior studies on marketing capabilities and market performance by empirically interrogating the impact of digital marketing capabilities on SME agro-processors' market performance (both intermediate and final market performance outcomes). Existing knowledge and practices have been challenged (Foltean, 2019;de Ruyter et al., 2018;Wymbs, 2011) due to the strong and fast paced digital developments that keep changing markets. As changes in marketing environments occur, consumer needs change and marketers must find appropriate communication needs to satisfy their customers (Davidaviciene et al., 2019). (Trainor et al., 2013). Digital marketing environments are highly dynamic, fast-paced and unstable therefore require dynamic and adaptive marketing capabilities. Digital marketing capabilities are thus a class of marketing capabilities required in the digital marketing environment which are not the same as those required in traditional marketing environments. The variation originates from the application of technology required to execute the same traditional marketing capabilities. For example, market sensing is required in all marketing environments, but in digital marketing environments there is more use of digital marketing tools/technology than in physical environments to execute market sensing. That "extra" need for technology calls for a different set of capabilities which we are calling digital marketing capabilities. Researchers are increasingly realising the need for specific marketing resources and capabilities for the digital marketing environment. For example Habibi et al. (2015) suggested a move from traditional market orientation of Kholi and Jaworski (1990) and Narver and Slater (1990) to electronic market orientation that is developed to meet needs of digital marketing. Marketing capabilities such as intelligence gathering and information dissemination have been emphasised on before (Kholi & Jaworski, 1990), however the missing part is on how firms execute the same in digital marketing environments. Thus, the transformation from traditional markets to digital marketing requires new marketing capabilities (Gregory et al., 2019) and new knowledge on the nexus between the digital marketing capabilities and market performance.
Marketing researchers have always linked marketing capabilities to market performance (Qureshi et al., 2017;Mathews et al., 2016;Teece, 2016;Morgan et al., 2009;Hooley et al., 2005). The prominence of the capabilities approach among marketing researchers resulted in several studies substantiating the influence of marketing capabilities on market performance. However, the marketing capability approach has evolved in response to changes in technology and the general marketing environment. These changes brings to question existing evidence on the linkage between "traditional" marketing capabilities and market performance. In addition, disparate evidence was found on the linkage between different marketing capabilities and market performance. For example, researchers found dynamic capabilities to differently influence market performance under different conditions. One stream of research believes dynamic capabilities are more important in dynamic environments, another stream believes dynamic capabilities are relevant in moderately stable environments than in stable and dynamic environments whilst the third stream believe they are relevant in both dynamic and stable environments (Zhou et al., 2019). There is therefore a general lack of empirical research on the link between marketing capabilities and market performance (Zhou et al., 2019) neither is there a widely accepted classification of marketing capabilities (Kaleka & Morgan, 2019). Little is known on the influence of technology-based capabilities on market performance (Gregory et al., 2019) consequently posing a question on the identification of appropriate digital marketing capabilities to develop and deploy (Guo et al., 2018). Knowledge of appropriate digital marketing capabilities is important because not all capabilities contribute to market performance in the same way. In addition, marketing capabilities of large firms are widely known, whilst little is known in small firms (Khan, 2017) particularly in the digital marketing context. This is despite the recognition that, SMEs require different capabilities to those of larger entities (Khan, 2017).
In the Zimbabwean SME agro-processors context, research on the linkage between digital marketing capabilities and market performance is scarce. Previous research has dominantly focussed on agro-processors challenges (Mhazo et al., 2012), SME policy issues (Chigwenya & Mudzengerere, 2013), informal sector (Matsongoni & Mutambara, 2018), and government policies on SMEs (Bomani, 2016). Linking SME agro-processors' digital marketing capabilities to performance is important in the Zimbabwean context and other developing countries with similar conditions. Zimbabwe is an agro-based economy therefore agro-processors significantly contribute to job creation, poverty alleviation and livelihoods (AfDB, 2018;ITAC, 2016;Ampadu-Ameyaw & Omari, 2015). The closure of large enterprises also left a huge gap that SMEs filled in, as such they are important to the economy. However, the SMEs digital marketing capabilities and market performance remains vague. Existing research rarely consider traditional sectors such as SME agro-processors of developing countries (Zhou et al., 2019;Wendra et al., 2019). Elsewhere, although researchers are recognising the need for new knowledge, and marketing capabilities in the fast-paced digital marketing environment, research in this area is still sparse. Extant literature is dominantly traditional marketing oriented except a few studies on internet marketing capabilities (Mathews et al., 2016;Jean & Kim, 2019), digital business models (Verhoef & Bijmolt, 2019), and e-commerce marketing capabilities (Gregory et al., 2019).
This study primarily interrogated the impact of digital marketing capabilities on market performance of SME agro-processors in Harare, Zimbabwe. To achieve this objective, the study adopted a cross-sectional survey that used a closed questionnaire for data collection. The data was collected from SME agro-processors' marketing managers or owners. To determine digital marketing capabilities impact on market performance, a multiple logistic regression was conducted using STATA version 15.
The study contributes to the marketing capabilitymarket performance discussion by interrogating the impact of selected digital marketing capabilities on market performance of SME agro-processors in Harare, Zimbabwe. In-depth knowledge on agro-processors is central to the economic development of Zimbabwe because of the country's strong economic linkages with agriculture. Evidence from a developing country context contributes to the applicability and transfer of market knowledge since marketing is contextual (Sheth, 2011) thus the study provides an opportunity to validate theory developed from the West and generalise findings (Zhou et al., 2019). Knowledge on digital marketing capabilities of agro-processors in Zimbabwe also contributes to the body of knowledge on SMEs, agro-processors and market performance. The study offers a new integrative position on the linkage between digital marketing capabilities and market performance of a relatively under researched industry (agro-processing). This is important because capabilities of larger enterprises are always different to those of SMEs (Khan, 2017). To the best of our knowledge, this study is the first to link digital marketing capabilities and market performance of agro-processors in Zimbabwe. The rest of this paper is structured as follows, section 1, theoretical framework, section 2, conceptual and hypothesis development, section 3, methodology, section 4, results, discussion and conclusions, limitations and further research.

Theoretical framework
This study is anchored on the capabilities approach which is an extension of the resource based view (RBV). The RBV attributes performance differences in firms to differences in resources possessed by the firm. According to the RBV, resources that are valuable, rare, inimitable, and organisable in a unique way (VRIO) leads to superior performance (Barney & Hesterly, 2015). However, researchers criticised the RBV for being static, and internally oriented thus failing to explain performance differences in dynamic environments. Researchers further pointed that possession of resources alone is not enough, capabilities are required to profitably convert resources into value (Trainor et al., 2013). Marketing capabilities are complex bundles of skills and concentrated knowledge that effectively deploy and update existing resources (Day, 1994;Grant, 1991). Marketing capabilities are influenced by internal and external marketing conditions (Qureshi et al., 2017). Therefore the marketing capabilities approach view performance differences in firms as a result of differences in ability to deploy and utilise marketing resources. Extant literature, (Hunt & Madhavaram, 2019;Bitencourt et al., 2019;Zhou et al., 2019;Teece, 2016;Day, 2014;Teece et al., 1997) explain different types of capabilities ranging from static, dynamic to adaptive that influence market performance. Static capabilities are found in the RBV, thus are internally oriented and fail to recognise the need for agility in fast paced digital marketing environments. Contemporary marketing research has shifted focus to dynamic capabilities (Bitencourt et al., 2019;Gregory et al., 2019;Zhou et al., 2019) and adaptive capabilities (Day, 2014). Dynamic capabilities emphasise the "key role of management in appropriately adapting, integrating, and reconfiguring internal and external experience, resources, and functional competences within a changing environment" (Gregory et al., 2019). Dynamic capabilities take the view that an organisation can swiftly adapt and alter its internal resource configuration to support marketing processes to demand after receiving clear market signals (Morgan, 2012). Dynamic capabilities are those capabilities that "enable the firm to implement strategies using new and different combinations and transformation of resources" to match the changing market conditions (Teece, et al., 1997). Performance differences under dynamic capabilities are a result of dynamic markets and differences in capabilities that firms acquire and utilise to deploy resources instead of heterogeneous resources (Wang & Kim, 2017). Although dynamic capabilities take cognisance of environmental dynamism, they take an inside-out approach (Day, 1994) thereby risk missing some weak market signals. According to Day (1994), fast-paced unstable environmental conditions require adaptive capabilities that take an outside-in approach. By so doing, adaptive capabilities allows marketers to view the market with an open mind thereby increasing chances of picking weak market signals. Adaptive capabilities starts with the external environment, and takes a proactive approach (Guo et al., 2018) enabling a deep focus into customer problems and issues before looking inside the organisation for solutions. Adaptive capability require "extensible ability to proactively sense and act on market signals, continuously learn from market experiments, integrate and coordinate social network resources to adapt to market changes and predict industry trends" (Guo et al., 2018). A study by Guo et al. (2018) testing the influence of the three capabilities found adaptive marketing capabilities to have the largest influence on market performance.
Even though there has been a shift to dynamic and adaptive capabilities, little or no specific attention has been given specifically to digital marketing capabilities. This has led to lack of evidence on how SME agro-processors who are generally known to be resource constrained deploy and capture digital marketing capabilities' influence on market performance. It is therefore necessary to develop new evidence for the linkage from a developing country context. The development of marketing capabilities contributes to competitive advantage (Morgan et al., 2009) and firms develop capabilities through repetition, constant application of skills and efforts (Qureshi et al., 2017) but how does this happen in fast paced dynamic digital marketing environments? SME agro-processors require marketing capabilities to recognise, enhance, and evaluate opportunities and in turn to develop products that fulfil customer needs in selected markets (Day, 1994;Zhou et al., 2019). This entails appropriate pricing, real time communication that includes all aspects of interest to the customers and product availability in customers' channels of choice.
Although evidence of how marketing capabilities have evolved exist, no similar evidence exist on the linkage of digital marketing capabilities and market performance. While researchers seem to converge on the influence of marketing capabilities to market performance (Zhou et al., 2019;Qureshi et al., 2017;Morgan et al., 2009;Bitencourt et al., 2019;Davick & Sharma, 2016;Cacciolatti & Lee, 2016;Day, 2014), different perspectives with little empirical evidence exist on the contribution of different marketing capabilities. Further, extant literature on market performance is characterised by conflicting definitions and views on concepts and drivers of market performance. The extent to which variables such as marketing capabilities influence market performance differ from one market to the other, and from one study to another. There is no agreement on the inclusion or exclusion of financial measures in market performance discussions. For example Frosen et al. (2016) consider market performance measures of customers, competitors, and financials whilst Milfelner et al. (2008) consider market share, sales and loyalty. Whilst Frosen et al. (2016) included financials in market performance, Milfelner et al. (2008) considered profit levels, margin and return on investment (ROI) as part of financial performance. In another definition Homburg et al. (2007) clearly mixing financial measures and market measures considered market or marketing performance as a measure of an organisation's abilities to attain revenue, market share and growth goals through its activities and right use of resources in a cost-effective way. On the other hand Clark (2007) market performance measurement framework splits market performance outcomes into two that is intermediate market outcomes and final market outcomes. Although the framework allows combining financial and non-financial measures, it does not provide specific marketing activities or capabilities that can be linked to market performance. Therefore, no single approach can succinctly capture market performance, researchers always need to carefully adopt a basket of measures (Sergie et al., 2007).
Building on this theoretical background, the study views digital marketing environment as highly unstable, dynamic and unpredictable therefore requires a good balance of dynamic and adaptive marketing capabilities. This mix of outside-in and inside-out approaches equips marketers to create superior market performance that competitors cannot easily match. We take the view that dynamic capabilities enables a firm to develop capabilities to identify and respond to market opportunities (Gregory et al., 2019) which complements the RBV perspective of resource and capability exploitation based on market opportunities. Further, adaptive capabilities encourages active sensing, experimentation and open learning which strengthens an organisation's capability base. Because of its power to explain performance differences in dynamic, volatile and unpredictable digital markets, the marketing capability approach forms the theoretical foundation for our arguments. The study extends the marketing capability approach to dynamic digital marketing environments and links it to market performance. Considering the disparities in market performance measurement, the study considers market performance as a measure of intermediate market outcomes and final market outcomes (Clark, 2007). Final market performance outcomes blended financial and market related measures of sales growth, revenue and profitability whilst intermediate market performance measures considered consumer attitudes, awareness, availability, brand associations and customer satisfaction.

Conceptual and hypothesis development
The first set of digital marketing capabilities examined in this paper are e-market sensing capabilities. Firms need market sensing capabilities to identify opportunities, and seizing capabilities to exploit the opportunities (Teece, 2014). Market sensing are outside-in (adaptive) capabilities (Day, 1994) that promote "free" exploration of the external environment. According to Day (1994), marketsensing capabilities entail active gathering, interpretation, and dissemination of market information. The digital world gives customers a "voice", more choices, and access thus empowering them (Chaffey & Ellis-Chadwick, 2016). In view of the empowered customers, it is imperative that marketers constantly monitor market changes and take proactive action. Digital marketing technologies allow organisations to easily sense and respond to market needs (Setia et al., 2013). Constant learning brings new knowledge that is essential to market orientation (Day, 2011) and value creation processes. Although generally there is evidence of strong informal linkages among SME agroprocessors, their customers and suppliers, (Mhazo et al., 2012) no evidence exist of their e-market sensing capabilities. Therefore, we posit our first hypothesis as: Hypothesis 1: E-market sensing capabilities positively influence agro-processors' a) intermediate market performance outcomes, and b) final market performance outcomes.
The second set of digital marketing capabilities are digital strategy development and execution capabilities. This is firm's ability to create and implement e-strategy for the attainment of marketing objectives (Chaffey, 2015;Vorhies & Morgan, 2005). Strategy formulation require an understanding of both internal and external environment (Barney & Hesterly, 2015;Chaffey & Ellis-Chadwick, 2016) as such calling for dynamic and adaptive marketing capabilities. Digital market strategy options can include customer propositions, customer acquisition efforts, customer conversion & experience ingenuities, development & growth, channel integration, and site improvements (Charlesworth, 2018). Failure to define clearly strategy can lead to missed opportunities, wrong direction, narrow integration, inadequate collection of customer data, and resource wastages (Chaffey, 2015). Vorhies and Morgan (2005) found strategy development and implementation capabilities to influence market performance. In another study, Abdullah et al. (2019) found e-strategy to influence consumers' online banking adoption. Marketers cannot afford to ignore the value of research and strategy in the digital marketing environment (WSI, 2015). However, there is no evidence or conceptualisation of digital strategy development and execution capabilities in SME agroprocessors yet strategy development, and execution can provide causal ambiguities that competitors cannot easily imitate. We therefore posit that: Hypothesis 2: Agro-processors' digital strategy development and execution capabilities positively influence their a) intermediate market performance outcomes, and b) final market performance outcomes.
The third capability we interrogated in this study was digital market innovation capabilities. According to Calantone et al. (2002), innovation is the development of novel ideas, processes, models, products and their subsequent acceptance and implementation. Innovation is an outcome of learning process (Romijn & Albaladejo, 2002) that emarket sensing capability can provide. Innovation capability are the skills and knowledge required to successfully recognise, grasp, and enhance prevailing technologies, and develop new ones (Romijn & Albaladejo, 2002). Barrett et al. (2015, p. 45) deriving from Yoo et al. (2010, p. 726) defined digital innovation as the novel blending of digital and physical components to create products (services included) as a result of data gathered from different sources to deliver services that eliminate industry boundaries. Therefore this study considers digital market innovation capabilities to be the abilities to create value through the development of new digital market ideas, processes, models and products utilising digital market data and technologies. Digital market innovation has the potential to contribute to the development of sustainable competitive advantage (Hooley et al., 2005), contribute to firm growth (Kyriakopoulos et al., 2015, p. 398) and drive development in developing countries regardless of infrastructure and other resource limitations (Barrett et al., 2015). According to Barrett et al. (2015) innovation in developing countries is not the same to that of developed markets due to resource differences. Barrett et al. (2015) found that innovators in developing countries are good at managing costs, small adjustments to products, & business models, and developing innovations using few resources because of the constraints that they face.
In view of these innovation capability contributions in resource-constrained environments, it is important to establish the contribution of such capabilities in the digital marketing environment from a Zimbabwean context. Academic research on digital market innovations and market performance of agro-processors in Harare is missing. The study consider digital market innovation capabilities to be a crucial resource as the capabilities are complex, have causal ambiguity, are hard to copy and non-transferrable thus valuable to small firms. We therefore posit that: Hypothesis 3: Digital market innovation capabilities positively influence agro-processors' a) intermediate market performance outcomes, and b) final market performance outcomes.
The fourth capability examined in this study is leadership capability. We consider leadership capability to be the ability to lead, manage, motivate, and coordinate activities within the organisation. Possession of exceptional human capital is not enough, there is need to bring that capital together to create value in the organisation. Managerial and organisational processes in an organisation influence its competitive advantage (Teece et al., 1997). Management of human capital and its development influence motivation and loyalty, which in turn affects strategy implementation. Unlike other studies, this study deliberately chose leadership instead of managerial capability because most SMEs do not have clear organisational structures that clearly define management roles. Instead, a bundle of skills is generally available and the owner usually makes all the strategic decisions. As such, it is crucial to refer to capabilities to lead and direct all other activities. Therefore, we proposed the following: Hypothesis 4: Leadership capabilities positively influence agro-processors' a) intermediate market performance outcomes, and b) final market performance outcomes.  (H2b). Third, the model shows the effect of digital market innovation on intermediate market outcomes (H3a) and final market outcomes (H3b). Lastly, the model shows the effect of leadership capability on intermediate market outcomes (H4a) and final market outcomes (H4b). In this study, intermediate market outcomes constituted of awareness, customer attitudes, product availability, brand associations, and customer satisfaction whilst final market outcomes are sales growth, market share, and profitability.

Methodology
We conducted a cross-sectional survey on 298 SME agroprocessors in Harare, Zimbabwe. A survey was necessary because no data was available in public sources for secondary collection. Although expensive and difficult to collect (Moorman & Day, 2016), primary data remains viable option in this study.
A closed-ended questionnaire was the primary data collection tool. We distributed the questionnaire to senior marketing executives and owners or other senior managers in cases where no proper marketing departments existed. Senior executives were more knowledgeable about their marketing capabilities, processes, and performance therefore relevant to the survey. We adopted two approaches depending on respondents' availability, the interviewer approach and the drop-off and pick-up approach. We selected our respondents through two approaches, stratified simple random sampling in samples that had a well-defined sampling frames, and quota sampling where sampling frames were scarce.
All measures of the capability constructs were measured on a Likert scale ranging from one to five where one represented strongly disagree. Market performance constructs were measured on a one to seven Likert scale, and respondents judged their performance against the previous year. This subjective measurement or responses approach was useful because actual sensitive data could not be easily availed. Measures for all constructs were adapted from previous studies of Morgan et al. (2009), Barrett et al. (2015) and Calantone et al. (2002).
We analysed our data using STATA version 15. Initially we conducted descriptive analysis, followed by bivariate analysis and finally logistic regression to test our hypothesis. The results are in the following section.

Questionnaire validity
Several approaches were taken to ensure validity of the questionnaire since validity issues were central to the development and testing of the questionnaire. Validity was important in the study as it ensured the questionnaire measures what it was intended to measure (Saunders et al., 2016, p. 450). Since there is no single objective way of verifying what a researcher is measuring (Mooi et al., 2018), it was necessary to adopt multiple approaches. Wide literature review was one key element of our validity approaches. The wide literature review contributed to clearly defining research constructs. In addition, we sought expert opinion in the development and testing of the questionnaire. Expert opinion contributed to both face validity (expert validity) and content validity of measures. Content validity was essential to ensure the measurement questions in the questionnaire delivered adequate coverage of the required information needs (Saunders et al., 2016, p. 450). Further, we adapted market performance measures from O'Sullivan and Abel (2007), Hooley et al. (2005), Neely (2007), and Moorman and Day (2016). This contributed to construct validity. According to Garson (2013) construct validity is questionable if a researcher uses constructs that are at odds with literature because construct validity is indisputably connected to theory and hypothesis (Carmines & Zeller, 1979). Construct validity was also measured using Spearman's correlation. Before implementation, the questionnaire was pilot tested with workmates and conveniently selected agro-processors. The workmates who were researchers in marketing contributed to clarity, flow and general design of the questionnaire. This was part of face validity which sought to assess if the questionnaire made sense. Largely, pilot testing contributes to validity and reliability (Saunders et al., 2016).

Descriptive analysis
The results show that 54% of the respondents were male whilst 46% were females. Age composition of the respondents stood at 58% of the respondents in the 31-40 year age group whilst only 15% where aged 41-50 years. Results indicated that 28% of the respondents use websites and 84% did not have one. The majority of respondents, 35% had 6-10 years of existence in the agro-processing business. The results also show that 66% of the agro-processors that participated in the survey had 6-50 employees whilst 34% had 51-100. Food & beverage sector constituted 37% of the respondents followed by wood & furniture (34%), textiles and clothing (25%) and other (4%). Participants responded that they have on average 69% of both digital market innovation, and leadership capabilities of the required digital marketing capabilities. Respondents also indicated that they have on average 66% and 63% of digital strategy development & implementation and e-market sensing capabilities respectively. Generally, these results show that agro-processors have digital marketing capabilities (average 66.8%) for the execution of digital marketing activities. Table 1 shows the reliability analysis results. The researchers used the Cronbach's alpha index to test whether the research instrument (questionnaire) was reliable. In addition, the Cronbach's alpha test provides a guide to whether further advanced tests from the data are possible or not (Sekeran & Bougie, 2016;Saunders et al., 2016). All constructs in this study had Cronbach's alpha index above 0.75 indicating high reliability. The average alpha index was 0.895, which allowed the researchers to conduct further tests on the data. According to Sekeran and Bougie (2016), a Cronbach's alpha below 0.6 is poor, whilst an alpha around 0.7 is acceptable with alpha above 0.8 being very good.

Construct validity
Construct and criterion related validity was tested using Spearman correlation tests. The results showed a statistically significant association between digital marketing capabilities (DMC) and intermediate market outcomes (IO) and final market outcomes (FO) (p < 0.001). As DMC increases, IO and FO increases with correlations 0.350 and 0.227 respectively. According to Carmines and Zeller (1979, p. 23) if correlation results are positive and significant then there is evidence to support construct validity. Further, correlation contributed to testing criterion-related validity. Correlation had the power to tell if a variable predicted or was a valid measure for predicting market performance. According to Carmines and Zeller (1979, p. 18) variables that correlate highly predict outcomes better therefore are a good and valid measure. However, to achieve discriminant validity (Bagozzi et al., 1991) valid measures of unique concepts should not relate too highly. In this study, correlations were low therefore achieved discriminant validity. Overall, putting together validity approaches adopted in the entire methodology, the study was justifiably valid.

Digital marketing capabilities and intermediate outcomes
The results of t-tests showed that there was a difference in digital market innovation capabilities between those respondents >50% intermediate outcomes and <50% outcomes (p < 0.001), among those with >50% outcome they had 75% of the required digital market innovation capabilities compared to 40% among those with <50% income. There was also a difference in leadership capabilities between respondents who attained >50% outcomes and <50% outcomes (p < 0.001), among those with >50% outcomes they had 75% of the required leadership capabilities compared to 39% among those with <50% income. Differences between respondents >50% income and <50% income (p < 0.001) were also found in digital strategy development & execution and e-market sensing capabilities. Overall these results show that differences in digital marketing capabilities influence agro-processors' intermediate market performance outcomes.

Digital marketing capabilities and final market outcomes
There was a difference in leadership capabilities between those respondents >50% outcomes and <50% outcome (p < 0.001), among those with >50% final-outcomes they had 75% of the required leadership capabilities compared to 50% among those with <50% income. Results also indicated differences in digital market innovation capabilities between respondents >50% outcomes and <50% outcomes (p < 0.001), among those with >50% outcomes they had 75% of the required digital market innovation capabilities compared to 50% among those with <50% outcomes. Other differences where in digital strategy development & execution capabilities between those respondents >50% outcomes and <50% outcomes (p < 0.001), among those with >50% outcomes they had 72% of the required digital strategy development & execution capabilities compared to 50% among those with <50% final-outcomes. Finally, results indicated differences in e-market sensing capabilities between respondents >50% final market outcomes and <50% market outcomes (p < 0.001), among those with >50% outcomes they had 70% of the required e-market sensing capabilities compared with 42% among those with <50% final market outcomes. These results shows that differences in digital marketing capabilities influence agroprocessors' final market performance outcomes.

Hypothesis testing using logistic regression
The results in Table 2 are presented in two sections, first, digital marketing capabilities versus intermediate market outcomes and lastly, digital marketing capabilities versus final market performance outcomes. On the linkage between digital marketing capabilities and market performance, the results showed that there was a statistically significant association between intermediate market outcomes and leadership capabilities after adjusting for other study variable (p = 0.002), for a unit increase in leadership capabilities score, the odds of obtaining >50% intermediate market outcomes increases by 17%. This means agroprocessors are expected to witness increased intermediate market performance outcomes by 17% for every increase or improvement in leadership capabilities. Digital market innovation capabilities also had a significant association with intermediate market outcomes after adjusting for other study variables (p = 0.005), for a unit increase in digital market innovation capabilities score, the odds of getting >50% increases by 52%. This result means that an increase or improvement in digital market innovation capabilities is expected to result in a 52% increase of intermediate market performance outcomes. Digital market innovations therefore contributes the most to agro-processors' market performance. There was also a significant association between intermediate-outcomes and e-market sensing capabilities after adjusting for other study variables (p = 0.009), and for a unit increase in e-market sensing capabilities score, the odds of getting >50% decreases by 23%. The result means that although e-market sensing capabilities positively contribute to intermediate market performance, agro-processors' market performance outcomes are expected to decrease by 23% for every increase or improvement in e-market sensing capabilities. For digital strategy development and execution capabilities, the results showed that there was a statistically significant association between intermediate market outcomes and strategy development and execution after adjusting for other study variables (p = 0.005), for a unit increase in digital strategy development and execution capabilities score, the odds of obtaining >50% outcomes reduced by 30%. Therefore the results support all the hypothesis (H1a, H2a, H3a, and H4a) that linked digital marketing capabilities to intermediate market performance outcomes. On the linkage between digital marketing capabilities and final market performance outcomes, the results showed that there was a statistically significant association between final market performance outcomes and digital strategy development and execution capabilities after adjusting for other study variable (p = 0.019), for a unit increase in digital strategy development and execution capabilities score, the odds of obtaining >50% decreases by 11%. Therefore the results supported only hypothesis 2b.

Discussions and conclusions
The findings indicate that only 28% of agro-processors who participated in the survey have websites whilst 72% do not have. The results show a low website usage compared to existing research, which show that 60% of small business globally do not have websites (Nordahl, 2017 cited by Charlesworth, 2018). The low website uptake is contrary to findings that websites are part of the most used digital channels in SMEs (Heini & Heikki, 2015).
In the UK, about 2million, small businesses were found to be operating without websites yet websites could improve their revenues (Enterprise Management, 360, 2017). These results strengthen the argument that majority of SMEs utilised digital marketing poorly compared to big organisations (Heini & Heikki, 2015;Gilmore et al., 2007). Although the study found male respondents (54%), dominating compared to women (46%), previous research by FinScope (2012) found 53% of women to be engaged in agro-processing compared to 47% of men. The current study considered only participants who were in marketing executive positions or related, thus potentially excluding other women. However, the results shows that agro-processing provides a source of income to women (Ampadu-Ameyaw & Omari, 2015), constituting 59% of respondents in the 18-30 age group of this study. This implies women are getting into executive positions early than their male counterparts.
Results of the logistic regression shows significant associations between all tested digital marketing capabilities and intermediate market outcomes. This means agro-processors in Harare have the abilities and skills to convert their resources and execute digital marketing activities that create awareness, positive brand attitudes, and associations among other intermediate market outcomes. However, only digital strategy development and execution capabilities provided statistical evidence of significant association to final market performance outcomes of sales growth, profitability and market share. These findings are contrary to previous research that concluded that SMEs (agro-processors included) lack market information, intelligence, and market sensing capabilities (Gilmore et al., 2007;Mhazo et al., 2012;Heini & Heikki, 2015). Findings of this study support the capability approach that capabilities contribute to market performance, unlike marketing resources or assets, capabilities offer superior performance opportunities. The confirmation of hypothesis 1a that, "digital strategy development and execution capabilities positively influence intermediate market performance outcomes" is encouraging. This means agro-processors are developing and implementing digital strategies that are profitable in the sense that they are creating customer awareness, positive brand associations, positive attitudes, and availability of products and services. This a positive development considering that the Zimbabwean environment is very dynamic. This result confirms that agro-processors in Harare are able to design, and execute profitable digital marketing strategies. The findings support existing literature that strategy development and implementation capabilities influence market performance (Vorhies & Morgan, 2005). Organisations that fail to define clearly their strategies miss opportunities, and waste resources (Chaffey, 2015). In a study to assess adoption of online banking in Malaysia, Abdullah et al. (2019) found that e-strategy influence customer perceptions on intention to adopt online banking.
In addition, results of this study support the hypothesis that, "digital market innovation capabilities positively influence intermediate market performance outcomes". This means agro-processors in Harare are finding value in their pursuit of innovation capabilities. The results support findings of Barrett et al. (2015) that firms in developing markets innovate regardless of resource deficiences. The evidence of positive associations between leadership capabilities and intermediate market outcomes might signal that agro-processors in Harare have strong human capital. Although human capital is not enough on its own, it provides a strong base to build effective leadership capabilities (Teece et al., 1997). Corporate culture of SMEs is usually dominated by cultural perceptions and values of the owner or shareholders who tend to have more influence than in large corporates (Lynch, 2015). In some instances, these SMEs are owner managed, as such leadership capabilities become a key driver to success. The results affirms that agro-processors in Harare are able to bring together their human capital to coordinate, lead activities and processes in their organisations.
The final revelation that "e-market sensing capabilities positively influence intermediate market performance outcomes" is a positive outcome in the agro-processors capabilities research. Contrary to previous findings that agro-processors are not able to gather market intelligence (Zindiye et al., 2012) the study reveals that agro-processors are engaging in e-market sensing and benefiting from such capabilities. The power of digital technologies that agro-processors are implementing potentially explain this shift. Digital media facilitate easy market intelligence at a low cost (Charlesworth, 2018) thus reducing cost pressures for SMEs (Heini & Heikki, 2015). However, results testing the influence of digital marketing capabilities on final market outcomes support only one hypothesis that, "digital strategy development and execution capabilities positively influence final market performance outcomes". This implies that strategy development and implementation is critical to agro-processors. Capabilities remain valuable to Harare agro-processors though considering that they influence intermediate outcomes.
The study found that all the four digital marketing capabilities (strategy development & execution, innovation, leadership, and e-market sensing) to positively influence intermediate market performance measures. As such, the researchers concluded that digital marketing capabilities in agro-processors have a positive influence on intermediate market outcomes. However, only digital strategy development & execution capabilities showed positive influence on final market performance outcomes. We conclude that only digital strategy development and execution give agroprocessors positive final market performance outcomes of profitability, market share and sales growth. Although the other capabilities are good for intermediate outcomes, they do not directly influence final performance measures.
These conclusions imply that agro-processors must build digital marketing capabilities to improve both intermediate and final market performance outcomes. While intermediate market outcomes influence final market outcomes (Clark, 2007), there is need for identification and development of capabilities that directly influence final market performance outcomes.

Limitations and further research
The researchers acknowledges some limitations to this study. Market performance studies are difficult to conclude in cross-sectional surveys. It is often difficult for a research to delineate cause and effect of interventions in a cross sectional study. In the context of the current study, it was impossible to link objectively identified market performance outcomes to digital marketing capabilities. As such, the researchers relied on subjective measures. Although, the researchers sought responses from two senior executives, the approach still give problems as judgements always differ. The use of a questionnaire as the sole data collection instrument brings weaknesses to the study. The questionnaire did not give the researchers a chance to probe certain responses especially in cases where the senior executives had tight schedules and requested to complete the questionnaire at their own time. The questionnaire itself was long (10-pages) for respondents to remain focussed on the questions. The researchers identified lack of concentration in the last segment where tendencies of straight lining were high. In addition, lack of complete sampling frame was a limitation to the study. The mixed sampling approached posed challenges of identifying respondents in sectors that had no complete sampling frames. The researchers developed constructs for this study from non-digital marketing resources and capability studies. The lack of existing well-defined constructs that the researchers could test or extend to the agro-processing sector posed a limitation of generalisability.
For future research, while intermediate outcomes influence final market outcomes, there is need to link scientifically the two outcomes. For example, how much awareness is required to create a certain level of sales growth? Marketing researchers must test capabilities identified in this study in other markets and contexts. More research is required in the digital marketing capability area since the application of established capability frameworks has contextual limitations. More still, there is no agreement among scholars on resource and capability definitions, classifications and impact to market performance. There is need for more research that employ new marketing research techniques such as experimentation. Researchers must adopt more scientific, robust and objective means of establishing the cause and effect relationship between capabilities and market performance. Such studies can benefit from employing longitudinal perspectives so that researchers capture effects over a long period. In addition, studies of this nature must capture both objective and subjective data. By so doing researchers eliminate bias of both respondents and the researcher. Overall, there is need for more research in developing markets particularly in the African context that focus on digital marketing capabilities, activities, and market performance. Knowledge development in the digital marketing space is weak and the gap continue to grow with every technological development.