The impact of articial intelligence analytics in enhancing digital marketing: the role of open big data and AI analytics competencies

Purpose: The goal of digital marketing is to enhance strategic decision making to discover competitive and consumer behavior, and to provide early warnings of risks and opportunities. In the age of digitalization, digital marketing may be inuenced with the use of cognitive technologies and also open sources of big data. The aim of this paper is to understand the impact of articial intelligence analytics on digital marketing and the role of open big data and AI analytics competencies in this relationship. Design: A structural equation modeling of the 227 questionnaire was carried out. Findings: Our analysis shows that open big data has a signicant impact on digital marketing, and AI analytical competencies moderate the impact of open big data on digital marketing. Value: Theoretically, this study extends the literature on knowledge-based view and AI analytics in digital marketing. Practically, this study broadens understandings regarding the perquisites of digital marketing through the use of cognitive technologies and open big data


Introduction
The goal of marketing intelligence is to enhance strategic decision making to discover competitive and consumer behavior, and to provide early warnings of risks and opportunities [1]. In the age of digitalization, marketing intelligence is improved with the use of digital technologies, such as arti cial intelligence, big data analysis and the Natural Language Processing (NLP) [2,3]. Marketing intelligence is a knowledge-intensive trend that calls for structured collections of comprehensive data on consumers and competitors [4]. Cognitive technologies consisting of technologies such as arti cial intelligence, NLP, Natural Language Generation (NLG), image, voice and video recognition have disrupted work in the future and have introduced new concepts like "augmented intelligence" [5]. These technologies boost companies ' ability to process vast volumes of data in real time and increase business e ciency in the prediction of future consumer or competitor activities as well as marketing ecosystem transformation [6]. Such advantages have driven most companies to develop their marketing strategies by using cognitive technology.
Besides cognitive analytics, open Big Data as external knowledge sources will provide companies and marketing a promising future [7,8]. Open data have developed various marketing capabilities because they are quickly and rapidly usable, inexpensive and widely available [7]. In addition to cognitive analytics and open data, the third variable of this study is cognitive analytics competency inside businesses. It refers to the analytical competence of companies through the use of cognitive technologies and is called the fourth era in analytics or analysis 4.0. [9]. Cognitive technologies enable cognitive insight, process automation, and cognitive engagement [10]. In this study, we incorporated the framework developed by the [11] to analyze cognitive analytics competency. In their study, Ghasemaghaei et al., (2018) classify data analytics resources into data utilization, data quality, technical and human skills, and tools sophistication. However, considering the differences of cognitive analytics from the other types of analytics [9], in this study, we considered new types of analytics empowered by cognitive technologies as one of the constructs of the study. Unlike previous studies focusing on AI-enabled analytics [10], we also looked at other cognitive technologies and their impact on the abilities of companies in big data analytics. Cognitive technologies augment analytics by providing cognitive sensing, cognitive insight, cognitive engagement, and cognitive automation.
It is widely recognized that organizations have different levels of analytical expertise and vary in their ability to extract and embed different external data sources. Nevertheless, the in uence of companies ' cognitive analytics capabilities on marketing intelligence is lacking in knowledge. The second objective of this study is to ll this gap by understanding the moderating role of cognitive computing competencies in in uencing the use of cognitive analytics on marketing intelligence by enhancing open big data extraction.
In this study, we aim to understand 1) whether open big data as an external source of knowledge has a moderating role in the impact of using cognitive analytics on the marketing intelligence of a company; and 2) the role of cognitive analytics competency in enhancing marketing intelligence through increasing the use of open big data. The ndings of this study will help researchers and managers better understand the roles of competency in cognitive analytics and open data (i.e. as an external source of knowledge) in improving companies ' marketing intelligence.

Research Model
The aim of this study is to explore the relationship between use of cognitive analytics, open data use, cognitive analytics and marketing intelligence.
The research model of this research is shown in Figure 1.
H1: Open data increases marketing intelligence. acquired knowledge for strategic planning in the long and short term. [12]. Marketing intelligence increases as different data sources are combined [13,14]. Previous studies have shown that open data is important for intelligent purposes, for example in marketing practices [7]. These studies argue that marketers with open data access may have higher marketing intelligence levels (Fleisher, 2008). For instance, for the meta classi cation of gender predictions and the personalization of smart device offers, [15] discuss the use of consumer pro ling and microblog data in cognitive computing.
Nonetheless, several problems are associated with the use of open data and other digital data [16]. For example, [17] argue that the quality of open data is uncertain because it can be produced from a variety of data sources and that open data can be anything. Or the open data collected may be of little interest and bene t in meeting the marketing requirements [7].
H2: Cognitive analytics competency moderates the in uence of open data usage on rms' marketing intelligence, such that the effect is stronger with higher competencies.
Based on previous research, the adoption of advanced technologies has an impact on organizational intelligence [18]. Companies should integrate external data to improve the quality of marketing activities and to increase business, competitors and the environment knowledge. [19,20]. Previous research states that analytically advanced businesses integrate more data sources to improve customer engagement [21].
Cognitive technologies utilization, which is composed of cognitive automation [22,23], cognitive insight [24,25], cognitive sensing (Chen et al., 2018), and cognitive engagement [21] will assist rms to develop deeper understandings of their market and competitors [28,29]. Cognitive technology utilization is therefore an important factor in the rm's ability to perform cognitive analysis. Another important component of cognitive analytics competence is the maturity and sophistication of cognitive technologies. Higher sophistication cognitive technologies such as facial recognition, have improved neuro-marketing practices [30,31]; or Arti cial Intelligence Imagery Analysis has fostered big data analytics [32]. Therefore, cognitive analytics tools with higher maturity levels can help enhance consumer, business, and competitor understanding; therefore improving marketing intelligence.
The third element of the cognitive analytics capabilities of a company is the technical skills and human resource capability. Cognitive analytics capabilities of human resources is a vital factor enabling a rm to lead digital transformation [33]. If employees lack adequate technical skills in cognitive analytical tools, their attempt to gain meaningful insight from data will lead to false consumer and business interpretation. Competency in cognitive analytics such as cognitive analytical skills of employees or tools with a certain level of maturity may in uence the use of open data sources by an organization. Previous studies show that it is important to combine digital knowledge and human capital to gain meaningful insights from data [34]; and previous studies [35] highlight the complementarity of humans and cognitive technologies like AI to enhance organizational decision-making processes. Companies use cognitive analytics to provide improved analytical reports (such as automated, prescriptive and predictive analytics) and to derive value in real-time from large volumes of structured and unstructured data [36]. There is an increasing number of companies using cognitive analytics to achieve competitive advantage and improve customer engagement [21]. Big data, both unstructured and structured will allow companies to improve their marketing campaigns [37].
Open data as a form of external knowledge will allow companies to access and process large amounts of data that are easy to access and inexpensive [7]. Cognitive analytical tools like AI and NLP will introduce potential business values for businesses as they provide cognitive insight, automation and engagement [38]. Using cognitive analytics tools is a potential knowledge asset that may improve acquisition and processing of external knowledge in a rm. cognitive technologies, such as AI or NLP, will allow an organization to gain new insights to optimize its marketing practices [41].
The use of cognitive technology such as cognitive automation, on the other hand, would speed up the automated process of real-time data access, such as open data [42]. Previous studies suggest that factors such as the lack of skills and technical resources to use open data effectively challenge the position of open data [43,44]. The prior research conclude that the function of open data is challenged by factors including incompetency and lack of proper technical resources for successful use of open data.

Methodology
In this study, a survey questionnaire was used to investigate the mediating role of open data on the relationship between cognitive analytical use and marketing intelligence. We also investigated whether cognitive analytics enhances the intelligence of marketing by increasing the use of open data. We incorporated the literature on knowledge-based view and cognitive analytics competency to conceptualize and validate our research model. To this goal, we used partial least square by using the SMART PLS software.

Sampling And Scaling
In this study, we used a survey questionnaire (Table 1). While some of the measurement scales for the constructs have been selected from the literature, others have been developed by the authors. Tool sophistication as the dimension of cognitive analytics competency is a formative variable.
The tools sophistication construct was developed and measured by using a 5-item scale. We developed and measured cognitive technologies utilization as a second order variable as a re ective variable which was formed by cognitive insight (5-item scale), cognitive sensing (3-item scale), cognitive engagement (3-item scale), and cognitive automation (4-item scale). Each of its item as rst order variables is re ective. Technical skills construct is a re ective construct and is measured using a three-item scale adopted from [11,45].We developed and measured marketing intelligence as a re ective variable (6-item scale). Open data usage was developed and measured as a re ective variable using a 4-item scale. The data quality item of the open data construct was adopted from [11,45].
We applied the [46] methodology to develop measurement scales for marketing intelligence, open data usage, tools sophistication, and cognitive analytics usage. After reviewing the literature on these three constructs, we developed some candidate measurement items by using the assistance of four faculties. To check the content's validity, we sent the instrument to 20 MIS experts to collect their views on acceptable items to be included. The average raw agreement rate between the judges was 92 per cent; indicating a high reliability level.
Using LinkedIn, we circulated our questionnaire to 568 middle and top-level managers around the world, who had ample experience and knowledge about cognitive computing and its effect on digital marketing. Of this number, 227 returned the questionnaire. Table 1 depicts the characteristics of respondent's rms and their demographic characteristics.

Results
We used SmartPLS version 3.2.8 [47] to check the proposed research model to implement structural equation modeling based on partial minus squares (PLS). PLS has gained attention among researchers [48] and is a well-established technique which can be used to test structural models [47]. Our research was carried out in two phases in which the rst stage included the evaluation of the measurement model, and the structure model was analyzed in the second stage.

Measurement Model
In the proposed research model, technical skills and capabilities, marketing intelligence use of cognitive analytics, use of cognitive technologies (as a second order variable formed by cognitive insight cognitive sensing, cognitive engagement and cognitive automation) were conceptualized as re ective constructs, while sophistication of tools open data quality and reuse of open data were considered as formative measurement constructs.
For re ective constructs, this study examined the convergent and discriminant validity and reliability of measurement scales. As show in Table 2, the convergent validity was adequate as standardized factor loadings were greater than 0.60 on its assigned construct and the average variance extracted (AVE) was greater than the benchmark of 0.5 [48]. The reliability of constructs was assessed by composite reliability (CR) coe cients (Table 2) and the ndings indicated that all re ective constructs was higher than the recommended value of 0.7 [49]. This study used the Fornell-Larker criterion and heterotrait-monotrait (HTMT) ratio to assess discriminant validity. As illustrated in Table 3 discriminant validity was sustained as the square root of the AVE was greater than all of the interconstruct correlations. In addition, the HTMT value obtained for each construct was under the prede ned threshold of 0.85, indicating the main constructs measured different aspects (Table 4) [50].   For formative constructs, this study tested potential multicollinearity among item measurement models by using a variance in ation factor (VIF) value.
As show in Table 5, the VIF value for all the formative variables was below the cut-off point of 3.3 [51], indicating that multicollinearity is not an issue for the formative constructs. Furthermore, to examine the cognitive technology utilization, cognitive analytics competency, and open data utilization as higher order formative constructs, we employed Bagozzi and Fornell's (1982) guideline. To create the composite indices, as Bagozzi & Fornell (1982) suggested, we multiplied weights of the indicators of each rst order variable and summed them up. Then, we used the composite indices to measure the higher order constructs. The VIFs of the higher order constructs were lower than 3.3, thus, multicollinearity was not a problem. Moreover, a signi cance level of at least 0.05 for each of these constructs ensured the construction of the composite latent construct. To sum up, the psychometric test of the measurement items displayed unidimensionality and consistency.

Structural model
SmartPLS Version 3.2.8 [50] was also used to assess our structural model. As Fig. 2  and implies that the model has predictive relevance. Furthermore, a t index of standardized root mean square residual (SRMR) for the composite factor model was used [50] . As the SRMR was 0.061 and below the threshold of 0.08, the good t of the model was con rmed.
To further investigate whether the signi cant moderating effect of cognitive analytics competency is substantive, R2 changes resulting from the interaction effects should be examined [52], speci cally through the F-test [53]. Therefore, in Model 1 we included the main effects of independent variables, wherein Model 2 we added interaction effect of cognitive analytics competency to Model 1. The results showed that the interaction effect of cognitive analytics competency with open data usage signi cantly increased R2 of marketing intelligence by 2.27 percent (F =8.03, p < 0.05), indicating a small effect size (f2 = 0.23). Therefore, the results showed that the interaction effect increased R2 signi cantly, con rming the signi cance of the moderating effect [52].

Discussion
In the era of Big Data, many businesses and marketers are using sophisticated analytical tools, such as cognitive technologies, for collecting and analyzing high volume data from various sources in a variety of formats to improve their marketing processes. Open data is an important source of external knowledge for organizations, due to their accessibility. Our literature review found that the prerequisites for improving marketing through cognitive analytics have been ignored in most previous research. The present study aims to ll this research void and therefore to evaluate the role of cognitive analytical capability and open data use in the advancement of marketing utilizing cognitive analytical tools.
Our study revealed that open data usage has a signi cant effect on marketing intelligence; and increases marketing intelligence. This nding supports prior research veri ed that open data generates value for rms [54] and are bene cial in marketing analytics [7]. This study a rmed that cognitive Previous studies argued that using cognitive analytics will facilitate the competencies of rms in extracting and analyzing big data [55]. organization leveraging open data has a higher level of marketing intelligence than a business that uses cognitive analytics without using open data.
One possible future, therefore, is to understand the role of other sources of external knowledge in marketing intelligence.
Third, prior research has emphasized the use of big data analytics in digital marketing, and to our knowledge, we have not found any empirical study examining the moderating role of cognitive analytics competence in digital marketing. Our study showed that while cognitive analytics competence moderates the relationship between open data and digital marketing, it does not moderate the relationship between cognitive analytical use and open data. The role of cognitive analytical skills in marketing should therefore be further explored in possible future studies.

Practical contribution
Many companies invest signi cantly in cognitive analytics technologies to support their marketing activities. On the other hand, companies and governments have been under considerable pressure to exchange their data and to set up open databases in public spaces. Our research results will broaden business views on the role of open data in advancing digital marketing, using sophisticated analytical tools. On the other hand, our work highlights the importance of organizations ' willingness to use cognitive analytics, employee recruitment with experience in advanced analytics, such as vision analysis, and the use of high-maturity analytical tools to enhance their marketing practices. Our research encourages companies to spend more in hiring skilled professionals and delivering successful and specialized training. Studies have con rmed that cognitive analytics tools such as AI can help companies recognize quali ed personnel automatically via social networks such as LinkedIn. Through the ndings of this study, businesses may understand that cognitive analytics tools can be used as appropriate to improve cognitive insight cognitive engagement cognitive sensing and cognitive automation. It reduces costs and leads to high returns on investment for corporations in the implementation of their marketing activities.

Limitations and future studies
This study has its own limitation. We explored the role of open data as an external knowledge asset on digital marketing; so one possible future research can be exploring the impact of big data as an external knowledge asset on digital marketing. In addition, we focused mainly on the use and competence of cognitive analytics; therefore, one possible future strand could be the analysis of the impact of cognitive technologies; or the review of the impact of each individual technology on cognitive technologies.  Structural model results