Al-Driven Decision Support Systems for Business Strategy

 Corresponding author: Osita Victor Egwuatu Copyright © 2025 Author(s) retain the copyright of this article. This article is published under the terms of the Creative Commons Attribution Liscense 4.0. Al-Driven Decision Support Systems for Business Strategy Osita Victor Egwuatu * MBA (Information Systems), College of Business and Innovation, The University of Toledo, Toledo, Ohio United States. World Journal of Advanced Research and Reviews, 2025, 27(02), 1752-1769 Publication history: Received on 16 July 2025; revised on 24 August 2025; accepted on 26 August 2025 Article DOI: https://doi.org/10.30574/wjarr.2025.27.2.3065 Abstract This study explores how artificial intelligence (AI), specifically machine learning (ML), transforms Decision Support Systems (DSS) from descriptive tools into predictive and prescriptive engines for strategic decision-making. Using a case study in the retail sector, structured (sales, financials) and unstructured (reviews, social media) data were analyzed through supervised learning, natural language processing, and reinforcement learning models. Findings show improved predictive accuracy, customer retention, and sustainable pricing strategies compared to traditional IS/MBAs frameworks. The research contributes theoretically by extending DSS and IS literature, and practically by providing business leaders with actionable, AI-driven frameworks for long-term strategic agility. Keywords: Artificial Intelligence Governance; Enterprise Information Systems; Algorithmic Bias; Data Ethics and Privacy; Responsible AI Adoption; Regulatory Compliance 1. Introduction 1.1. Context: The Rise of Artificial Intelligence in Business Environments In recent decades, artificial intelligence (AI) and machine learning (ML) have moved from experimental technologies in academic and laboratory settings to mainstream tools driving innovation across industries. Businesses are increasingly harnessing AI not only to automate routine operations but also to enhance higher-order functions such as forecasting, risk assessment, and strategic decision-making. This shift reflects a broader digital transformation in which organizations compete not only on physical assets but also on their ability to generate, process, and interpret data. According to reports by McKinsey and Gartner, firms that embed AI into their business processes outperform peers in revenue growth and market positioning, suggesting a strong competitive premium for early adoption. In this evolving environment, managers face unprecedented complexity. Global supply chains, rapidly changing consumer behaviors, regulatory uncertainties, and disruptive competitors demand decisions that are faster, more accurate, and more adaptive than traditional methods allow. The sheer volume of structured and unstructured data available to organizations, from transaction records to social media sentiment, creates both opportunity and challenge. While this data can yield insights that inform strategy, extracting meaningful knowledge requires computational approaches far beyond the capacity of conventional analytics. 1.2. Problem: Limitations of Traditional Decision Support Systems Decision Support Systems (DSS) emerged in the 1970s as information systems designed to assist managers in making semi-structured or unstructured decisions. Early DSS incorporated databases, statistical tools, and what-if analyses, offering value by organizing information and providing frameworks for decision analysis. However, traditional DSS often fell short in two critical areas: predictive power and prescriptive guidance. World Journal of Advanced Research and Reviews, 2025, 27(02), 1752-1769 1753 First, traditional DSS were primarily descriptive. They summarized past performance and enabled scenario analysis but rarely anticipated future outcomes with accuracy. Their reliance on rule-based logic, deterministic models, or linear statistical approaches limited their ability to handle nonlinearities, high-dimensional data, and complex interdependencies that characterize modern business ecosystems. Second, these systems often lacked adaptability. Business environments evolve rapidly, yet conventional DSS models require manual updates and are not capable of learning dynamically from new data. As a result, managers using such systems risked basing their strategies on outdated assumptions or incomplete representations of reality. The growing sophistication of AI offers a compelling solution. Machine learning models, especially supervised and unsupervised learning algorithms, deep neural networks, and reinforcement learning systems, can identify subtle patterns in large datasets, update themselves as new information emerges, and generate predictions and recommendations with increasing accuracy. These capabilities shift DSS from being primarily descriptive to becoming predictive and prescriptive tools that directly inform strategic choices. 1.3. Aim: Toward AI-Driven Decision Support for Strategy The central aim of this research is to explore and demonstrate how AI-driven Decision Support Systems (AI-DSS), underpinned by machine learning models, can guide managerial decision-making at the strategic level. Unlike operational or tactical decisions, which often involve shorter time horizons and narrower scopes, strategic decisions concern long-term direction, resource allocation, competitive positioning, and overall organizational survival. These decisions require synthesizing vast amounts of heterogeneous information and evaluating multiple uncertain futures tasks for which ML-enhanced DSS are uniquely well-suited. Specifically, this study seeks to: • Examine how machine learning models can be effectively integrated into DSS architectures to enhance predictive and prescriptive capabilities. • Apply AI-DSS within a business case study to illustrate practical applications and outcomes. • Assess the benefits and limitations of AI-DSS for strategic decision-making, with particular attention to risks such as algorithmic bias, interpretability challenges, and organizational adoption barriers. By combining conceptual development with empirical illustration, the article positions AI-DSS as not merely a technological innovation but a transformative managerial tool. 1.4. Contributions: Theoretical and Practical This research makes contributions on both theoretical and practical fronts. Theoretically, it extends the Information Systems (IS) literature by integrating AI and ML into the longstanding DSS paradigm. While DSS research has traditionally emphasized data management, interface design, and decision heuristics, the infusion of ML models introduces new capabilities that redefine the boundaries of IS scholarship. This study also contributes to the emerging literature on digital strategy by situating AI-DSS as a key enabler of dynamic capabilities, resource orchestration, and organizational agility. Practically, the article proposes a framework for designing and implementing AI-driven DSS tailored to strategic management contexts. By presenting a real-world case study, it demonstrates how predictive modeling can be operationalized in managerial settings and how outputs can be translated into actionable strategic insights. Moreover, the discussion of risks and limitations provides managers with balanced guidance, helping them navigate ethical, technical, and organizational challenges. 1.5. Research Questions To operationalize its aims, this study is guided by the following research questions: • Integration Question: How can machine learning models be incorporated into decision support systems to enhance their ability to guide strategic decision-making? • Value Question: What tangible business benefits can organizations derive from adopting AI-driven DSS in strategy formulation and execution? World Journal of Advanced Research and Reviews, 2025, 27(02), 1752-1769 1754 • Risk Question: What risks and challenges arise from reliance on AI-driven DSS, particularly concerning algorithmic bias, interpretability, and managerial trust? These questions frame the investigation, ensuring that the analysis remains grounded in both technological capabilities and managerial realities. 1.6. Structure of the Article The remainder of the article is organized as follows. Section 3 reviews relevant literature on DSS, AI, and business strategy, identifying key gaps this study addresses. Section 4 presents the theoretical framework that links AI-DSS to established IS and strategy theories. Section 5 outlines the methodology, combining case study research with predictive modeling. Section 6 provides the case study findings, while Section 7 discusses results and implications. Finally, Section 8 concludes with a summary of contributions and suggestions for future research. 2. Literature Review 2.1. Evolution of Decision Support Systems (DSS) Decision Support Systems (DSS) have their roots in the broader field of Management Information Systems (MIS) that emerged in the mid-20th century. Early MIS in the 1960s and 1970s were primarily focused on electronic data processing and routine reporting, providing managers with periodic summaries of business information. The concept of DSS began to take shape as a response to the need for more interactive analytical tools to assist in decision-making. By the late 1960s and early 1970s, researchers introduced DSS as computer-based systems designed to aid managers in solving semi-structured and unstructured problems by combining data, analytical models, and user-friendly software interfaces. These early DSS were relatively simple – often built on basic models (like spreadsheets or rudimentary statistical programs) – but they laid the foundation for more sophisticated decision support by allowing “what-if” analysis and scenario planning beyond what standard MIS reports could offer. As computing technology advanced, DSS capabilities expanded through the 1980s and 1990s. One significant development was the rise of expert systems, which can be viewed as a form of knowledge-driven DSS. Expert systems emerged from artificial intelligence research and sought to capture human expertise in a set of rules and inference mechanismsdssresources.com. Unlike traditional model-driven DSS that followed predefined mathematical models, expert systems attempted to simulate human reasoning by using a knowledge base of facts and rules, along with an inference engine to apply logical rules to those factsdssresources.com. These systems could provide recommendations or diagnoses for specific problem domains (for example, medical diagnosis or mineral exploration) by mimicking the decision processes of human experts. In practice, expert systems represented an early infusion of AI into decision support, enabling computers to handle qualitative knowledge and inference. However, they were typically limited to narrow tasks and required extensive knowledge engineering (manual encoding of expert knowledge), which made them challenging to build and maintain. The 1980s also saw the proliferation of other DSS-related tools such as Executive Information Systems (EIS) for top-level dashboards and Group Decision Support Systems (GDSS) for collaborative decision meetings, each addressing different needs in the decision-making hierarchy. These varieties of DSS marked an evolution from basic MIS reporting toward more specialized support for decision processes at operational, tactical, and strategic levels. By the late 1990s and 2000s, business intelligence (BI) and data warehousing technologies became prominent, further transforming DSS. BI systems built upon data integration and data mining techniques to provide comprehensive analysis across different facets of the organization. They could handle large volumes of historical data, produce multidimensional reports, and uncover patterns or trends to inform decisions. In essence, BI broadened the scope of decision support by incorporating semi-structured and unstructured data (e.g. from transactions, customer interactions, etc.) and delivering insights through dashboards and visualizations. The focus was on enabling data-driven decision-making across all management tiers, improving not just operational decisions but also informing strategy by identifying key performance drivers. Most recently, the evolution of DSS has accelerated with the advent of advanced artificial intelligence and machine learning techniques essentially AI-driven DSS. Modern DSS are increasingly integrated with AI algorithms, predictive models, and big data analytics to provide far more powerful insights and recommendations than earlier generations of decision support tools. These AI-driven systems can analyze massive, complex datasets in real time, learn from new data, and even perform autonomous decision-making in certain contexts. For example, incorporating deep learning has given DSS “quantum leap” improvements in predictive accuracy and adaptability, enabling more precise forecasts and World Journal of Advanced Research and Reviews, 2025, 27(02), 1752-1769 1755 pattern recognition than traditional statistical models. Today’s AI-driven DSS can continuously refine their recommendations as new data flows in, and they often include natural language interfaces or conversational agents that make them easier for managers to interact with. Crucially, the integration of AI has extended decision support into areas that require higher-order cognition such as strategic planning and risk assessment – which earlier MIS and DSS struggled to address. In fact, modern AI capabilities are credited with offering organizations a competitive edge by facilitating faster data-driven decisions, enhancing operational efficiency, and even creating personalized customer experiences at scale. The capabilities of AI-driven DSS range from real-time analytics and innovative product development to risk prediction and long-term strategic forecasting. As organizations integrate AI into their decision support infrastructure, they not only optimize internal processes but can also create higher barriers to entry for competitors, effectively positioning themselves as industry leaders through superior decision intelligence. 2.2. Machine Learning in Business Decision-Making Machine learning (ML), a core subset of AI, has become a driving force in contemporary decision support systems and business analytics. While earlier DSS relied on predefined models or expert-crafted rules, ML techniques enable systems to learn patterns from data and improve over time without being explicitly programmed for each scenario. In business, ML is applied across a wide range of functions from forecasting and marketing to operations and customer service essentially bringing a predictive, adaptive edge to decision-making processes. Below, we discuss key ML approaches in business and their impact: Predictive Analytics: One of the most common applications of ML in business is predictive analytics, which involves using historical data and statistical algorithms (including machine learning models) to predict future outcomes. Predictive analytics transforms raw data into forward-looking insights, helping companies forecast trends, customer behaviors, and risks so they can act proactively rather than reactively. For instance, organizations use predictive models to anticipate sales demand, identify which customers are likely to churn, or detect fraudulent transactions before they occur. By leveraging techniques such as regression analysis, time-series forecasting, or machine learning classifiers, businesses can align their strategies with likely future scenarios. The benefit is improved decision accuracy and timing – companies can address opportunities and challenges proactively, gaining a competitive advantage by staying one step ahead of market changes. In fact, firms that effectively use predictive analytics often report better alignment of their operations with strategic goals and an enhanced capacity for risk mitigation through early warnings. In summary, predictive analytics powered by ML allows data-driven foresight in decision-making, from finance (e.g. credit scoring, investment predictions) to supply chain (demand forecasting) and marketing (targeted campaigns). Natural Language Processing (NLP): NLP is a branch of AI/ML focused on enabling computers to understand and generate human language. In business, NLP techniques unlock the value of vast amounts of unstructured text and speech data – customer reviews, social media posts, support emails, call transcripts, reports, and more. By processing this data, NLP can reveal insights that would be hard to obtain otherwise. Applications include sentiment analysis (gauging customer opinions and brand sentiment), automated customer service chatbots and virtual assistants, machine translation for global operations, and text analytics for tasks like contract analysis or resume screening. NLP thus helps organizations listen to and respond to stakeholders at scale. Importantly, NLP has become “indispensable for maintaining a competitive edge in today’s dynamic business environment”. It allows companies to rapidly analyze public perception and market trends from textual data, personalize content or recommendations for users, and streamline operations such as document processing and reporting. For example, deploying NLP-driven chatbots can provide instant 24/7 customer support, improving service quality while reducing costs – an advantage in competitive markets. Similarly, sentiment analysis can alert firms to emerging issues in customer satisfaction or product reputation in real time, enabling a fast strategic response. The integration of NLP into business processes ultimately fosters a deeper connection between companies and their customers by bridging human communication with machine intelligence, leading to more informed decisions and tailored experiences. Reinforcement Learning (RL): Reinforcement learning is an ML paradigm where an autonomous agent learns to make sequences of decisions by interacting with an environment and receiving feedback in the form of rewards or penalties. Over many iterations, the agent learns an optimal policy (strategy) to maximize cumulative rewards. In a business context, RL is especially powerful for complex, dynamic decision problems where there may not be a single-step prediction target, but rather a need to optimize long-term outcomes through a series of interdependent decisions. Use cases include dynamic pricing strategies, real-time supply chain and logistics optimization, adaptive control systems in manufacturing, recommendation systems that adjust to user behavior, and any scenario where decisions have a delayed impact that needs to be learned. The appeal of RL in business is that it can discover novel and adaptive solutions in environments too complex for rule-based programming. Unlike traditional programs, an RL agent is not explicitly told how to react to every situation; it learns from experience and exploration. This means RL can sometimes outperform World Journal of Advanced Research and Reviews, 2025, 27(02), 1752-1769 1756 human decision-makers in high-dimensional problems by considering a vast range of possibilities and learning from trial and error. For example, retailers face rapidly changing consumer preferences and market conditions that make static forecasting difficult. 2.3. Information Systems and Business Strategy: Linking Technology Adoption to Competitive Advantage Information Systems (IS) and information technology more broadly have long been recognized as key enablers of competitive advantage in business. A rich body of literature examines frameworks and models that explain how adopting technology can translate into improved performance, market position, and strategic differentiation. Fundamentally, these frameworks argue that technology is not merely a support tool for executing strategy, but can be an integral part of shaping and enhancing a firm’s strategy. One foundational perspective is Michael Porter’s view on IT in competition. Porter’s frameworks (such as the Value Chain and Five Forces) highlight how IS can create advantages either by lowering a firm’s cost structure or enabling differentiation. For example, integrating IS into the value chain can streamline processes (procurement, logistics, production, marketing, etc.), yielding cost leadership advantages, or enable superior customer insights and product innovations, yielding differentiation advantages. Classic cases often cited include Walmart’s use of information systems for supply chain optimization (achieving low-cost leadership) and Amazon’s use of data-driven personalization (differentiating through customer experience). These examples underscore that alignment between technology and business processes can directly bolster a company’s competitive strategy. IS can improve operational efficiency, enhance customer service, foster innovation, and enable faster decision-making all identified as key sources of competitive advantage in modern markets. Beyond individual cases, formal models like the Strategic Alignment Model (SAM) by Henderson and Venkatraman provide a theoretical framework for linking technology adoption to competitive advantage. The SAM posits that to fully realize value from IT investments, an organization must achieve alignment between its IT strategy and its business strategy. In other words, technology initiatives should be directly driven by business objectives and vice versa. This model identifies multiple domains (business strategy, IT strategy, organizational infrastructure, and IT infrastructure) and argues that coherence across these domains is critical. The rationale is that misalignment for instance, adopting a cutting-edge technology without a clear business strategic need will yield suboptimal results, whereas tight alignment can produce synergistic gains. Henderson and Venkatraman developed SAM specifically to address the “growing need for organizations to effectively exploit IT capabilities for competitive advantage and manage the increasing complexity of aligning technology with business goals”. It has since become a cornerstone in IS strategy research and practice, reinforcing the idea that technology adoption must be guided by strategy (and can even inform new strategic opportunities) to create sustainable success. Another important perspective comes from the innovation and diffusion of technology angle. Early adopters of transformative technologies can often gain a temporal competitive edge, a concept related to first-mover advantage. Businesses that are quick to embrace emerging technologies (such as AI, cloud computing, or IoT in recent times) may reap benefits like efficiency gains, new product or service models, and positive branding as innovators. These benefits can translate into market share growth or profitability bumps that laggards struggle to match. For example, companies that invested early in big data analytics capabilities were able to better understand customer trends and optimize operations ahead of their competitors, sometimes dominating their sectors as a result. As one industry commentary noted, early adopters of AI “shape their industries… set trends, improve services, and force competitors to catch up”, illustrating how being at the forefront of tech adoption can redefine competitive dynamics. A frequently cited case is Netflix, which built its recommendation system (an AI-driven engine) early on; this not only improved customer retention through personalized content, but also set a new standard in the entertainment industry that others had to follow. Likewise, Tesla’s aggressive adoption of AI for self-driving features and data collection has given it a lead in autonomous driving data that traditional automakers are racing to close These examples highlight how technology adoption timing is a strategic consideration: adopting too late can mean playing catch-up in capabilities and facing higher switching costs, whereas adopting at the right time (with the right implementation) can yield a defensible advantage, at least until the rest of the industry catches on. It is important to note, however, that not all technology adoption automatically confers long-term advantage. Some scholars (e.g., in the resource-based view of the firm) argue that for an IS or technology capability to provide sustained competitive advantage, it should be valuable, rare, inimitable, and supported by the organization (the VRIO criteria). In practice, this means that simply buying the latest software or hardware that competitors can also purchase may offer only a transient boost. The differentiating factor often lies in how technology is implemented and integrated with unique business processes, human expertise, and data assets. For instance, a company’s proprietary dataset combined with a World Journal of Advanced Research and Reviews, 2025, 27(02), 1752-1769 1757 custom ML algorithm can be a unique asset that rivals cannot easily replicate. Additionally, organizational change management and culture play a role: firms that successfully adapt their workflows and train their people to leverage a new system fully will derive more strategic value from it than those that do not. This aligns with the view that competitive advantage arises not just from technology itself but from embedding that technology in complementary organizational resources and strategies. 2.4. Research Gaps: DSS for Strategic Management Decisions Despite the advancements in decision support technologies, there remain notable gaps in the research and application of DSS, particularly regarding support for strategic-level decision-making. Traditionally, DSS research and implementations have been skewed toward assisting operational and tactical decisions those that are more structured, frequent, and data-intensive (e.g., scheduling, inventory management, budget allocations). These are areas where ample historical data and clear criteria allow analytic models to thrive. In contrast, strategic decisions (such as setting longterm objectives, entering new markets, or transforming business models) are often semior unstructured, involve significant uncertainty, and rely on higher-level judgment and external information. They are typically made by senior executives and have broad, long-term impacts on the organization. Most existing DSS tools and case studies do not extensively address this strategic decision space, creating a gap in both research and practice. A review of the literature reveals that few decision support systems have been explicitly designed to support high-level strategic management decisions – especially those in fast-changing, “high-velocity” environments. For example, Ladd et al. (2013) point out that very few commercial DSS offerings at the time could adequately support high-velocity strategic decision requirements out-of-the-box. This means that even as businesses face environments where strategic agility is crucial, their information systems are often not up to the task of providing the needed decision support. The consequences of this gap can include executives relying on intuition or incomplete data for strategic choices, or conversely, being overwhelmed by information without a framework to analyze it for long-term planning. 3. Theoretical Framework 3.1. Linking DSS to IS Theories The study of Decision Support Systems (DSS) has long intersected with Information Systems (IS) theory, which provides conceptual foundations for understanding how technology adoption generates value in organizations. Two theoretical lenses are particularly useful in framing AI-driven DSS: the Technology-Organization-Environment (TOE) framework and the Resource-Based View (RBV). The TOE framework explains technology adoption as the product of three interacting contexts: technological readiness, organizational structure and culture, and environmental pressures. From a DSS perspective, TOE suggests that the successful deployment of AI-driven decision support depends not only on the technical feasibility of integrating machine learning models but also on whether the organization has the managerial commitment, skills, and cultural openness to adopt data-driven decision-making. Environmental factors such as competition, regulatory demands, and industry volatility further influence whether firms perceive AI-DSS as a necessity for survival or differentiation. Thus, TOE situates AI-DSS adoption within a broader system of drivers and constraints, highlighting that the effectiveness of decision support is not purely technical but socio-technical. The Resource-Based View (RBV) provides a complementary perspective by focusing on the internal capabilities of the firm. According to RBV, organizations achieve sustained competitive advantage through resources that are valuable, rare, inimitable, and organizationally embedded. Data assets and machine learning capabilities increasingly fit this description. While many firms may access off-the-shelf DSS tools, the unique integration of proprietary data, domain expertise, and customized AI models can yield decision support capabilities that competitors cannot easily replicate. In this sense, AI-driven DSS can evolve into a strategic resource that underpins competitive advantage. RBV also underscores the role of organizational processes and human capital in fully leveraging such systems: a firm’s ability to train its managers to interpret and trust AI insights, and to improve its DSS continuously, is critical to sustaining its advantage. Together, TOE and RBV offer a dual lens: TOE explains the conditions that enable adoption, while RBV explains how adoption can translate into enduring strategic value. This theoretical synthesis is especially useful for analyzing AIdriven DSS because it situates them both as technological artifacts embedded in organizational environments and as potential strategic resources that can shape long-term competitiveness. World Journal of Advanced Research and Reviews, 2025, 27(02), 1752-1769 1758 3.2. Extending Simon’s Decision-Making Phases with Machine Learning Herbert Simon’s classical model of decision-making, comprising the phases of intelligence, design, choice, and implementation, remains foundational in IS and DSS research. Machine learning technologies significantly extend and enrich each of these phases. Intelligence Phase: Traditionally, this phase involved gathering and scanning information to identify problems or opportunities. ML algorithms now automate and amplify this phase by sifting through vast amounts of structured and unstructured data, detecting anomalies, trends, and signals that human managers might overlook. For example, clustering algorithms can segment customer bases, while anomaly detection can flag emerging risks in supply chains. Design Phase: In Simon’s model, the design phase entails developing possible courses of action. ML enhances this by generating data-driven models of alternative scenarios. Predictive models can simulate demand under different pricing strategies, and reinforcement learning can suggest adaptive pathways by experimenting with possible decision sequences. In this way, ML augments the creative process of design with empirical and simulated evidence. Choice Phase: The choice phase involves selecting among alternatives. Here, machine learning contributes by providing probabilistic predictions and optimization outputs that clarify trade-offs. Decision trees, ensemble models, or neural networks can offer ranked recommendations with confidence intervals, allowing managers to weigh decisions against quantified risks and benefits. Importantly, ML systems can present not just a single “best” option but a spectrum of choices optimized under different constraints. Implementation Phase: Finally, the implementation phase focuses on executing decisions and monitoring outcomes. ML models extend this phase through continuous learning and feedback loops. As outcomes unfold, reinforcement learning agents or online learning algorithms can update their policies, adjusting strategies in near real-time. This creates a dynamic implementation process where DSS not only guide decisions but also refine themselves as decisions are enacted and outcomes observed. Through this integration, ML technologies transform Simon’s decision model from a relatively linear, human-centric process into a cyclical, adaptive system where data and algorithms continuously feed back into each stage. The result is a DSS paradigm that is not just supportive but collaborative, working alongside human decision-makers to navigate uncertainty and complexity in strategic contexts. 4. Methodology 4.1. Research Design This study adopts a mixed-method research design, combining a case study approach with predictive modeling to investigate the integration of machine learning into Decision Support Systems (DSS) for strategic business decisionmaking. The case study provides rich, contextualized insights into how AI-driven DSS can be embedded within an actual organizational setting. At the same time, predictive modeling demonstrates the technical application of machine learning techniques to real-world business data. This dual approach ensures both theoretical grounding and practical validation, addressing not only the managerial aspects of decision support but also the computational feasibility of deploying ML-based DSS. The methodological choice is guided by the recognition that strategic decision-making cannot be fully captured by quantitative models alone. A case study allows for exploration of organizational, cultural, and managerial dynamics that shape the adoption and use of DSS. Predictive modeling, on the other hand, illustrates the tangible capabilities of machine learning in enhancing foresight and prescriptive guidance. Together, they create a comprehensive methodology that reflects the socio-technical nature of AI-driven DSS. 4.2. Industry and Firm Selection For this research, the retail industry has been selected as the empirical context. Retail offers a fertile setting for studying AI-driven DSS for several reasons. First, the industry is data-rich, generating vast amounts of structured data (e.g., sales transactions, inventory logs, customer loyalty programs) and unstructured data (e.g., product reviews, social media engagement). Second, retail firms face highly strategic decisions such as market expansion, pricing strategy, and customer segmentation that can directly influence competitive positioning. Third, the retail sector has been at the forefront of adopting analytics and AI, making it a relevant environment in which to study advanced DSS integration. World Journal of Advanced Research and Reviews, 2025, 27(02), 1752-1769 1759 Within this industry, the case study focuses on a mid-sized omnichannel retail firm that operates both physical stores and an e-commerce platform. This firm was chosen because it faces strategic questions about resource allocation between online and offline channels, customer retention in the face of increasing competition, and the optimization of pricing and promotion strategies. Such strategic challenges provide an opportunity to explore how ML-driven DSS can inform long-term decisions rather than merely operational efficiencies. 4.3. Data Sources The study employs both structured and unstructured data, reflecting the multifaceted information landscape of strategic business decision-making. 4.3.1. Structured Data Sales Transactions: Daily records of sales across different product categories and store locations, including variables such as quantity sold, revenue, discounts, and channel (online vs. offline). • Financial Data: Periodic profit-and-loss statements, marketing expenditure records, and inventory costs. • Customer Demographics: Information from loyalty programs, including age, gender, location, and purchase frequency. • Unstructured Data: Customer Reviews: Product reviews collected from the firm’s e-commerce platform and major online marketplaces. • Social Media Data: Tweets, posts, and comments mentioning the brand or its competitors, providing insights into sentiment and brand perception. • Market Intelligence Reports: Textual reports from industry analysts and market research agencies that highlight external environmental factors. The combination of structured and unstructured data ensures that the DSS can integrate both quantitative performance indicators and qualitative market insights, enhancing its ability to support strategic decisions. 4.4. Predictive Modeling Approach The predictive modeling component applies supervised machine learning algorithms to structured business data, complemented by text analytics techniques for unstructured data. The selection of models is guided by their interpretability, predictive power, and relevance to business applications. 4.4.1. Supervised Machine Learning Models: • Regression Models: Linear and logistic regression are used for baseline predictions, such as forecasting sales demand or predicting customer churn probabilities. • Decision Trees: Useful for segmenting customers and identifying key decision rules driving outcomes. • Random Forests and XGBoost: Ensemble methods that improve predictive accuracy by aggregating multiple decision trees. These models are particularly effective in handling nonlinear relationships and large feature sets common in retail data. • Time-Series Forecasting Models: For predicting future sales trends, models such as ARIMA and Prophet are employed, augmented with ML-based feature engineering to account for promotions, seasonality, and external shocks. 4.5. Unstructured Data Analysis Natural Language Processing (NLP): Sentiment analysis is performed on customer reviews and social media posts to extract customer sentiment scores. Topic modeling (e.g., Latent Dirichlet Allocation) is used to identify emerging themes in customer concerns and market discussions. Text Classification Models: Supervised classifiers categorize reviews or social posts into themes (e.g., product quality, customer service), which are then linked to structured performance metrics. 4.6. Reinforcement Learning for Strategic Scenarios For strategic decisions such as dynamic pricing or promotional allocation, reinforcement learning is employed. The RL agent simulates different pricing strategies over time, receiving feedback in terms of revenue and customer retention. World Journal of Advanced Research and Reviews, 2025, 27(02), 1752-1769 1760 Over multiple iterations, the model learns the optimal balance between short-term profit and long-term customer loyalty. The integration of these models provides a holistic DSS framework: predictive analytics for foresight, NLP for capturing external perceptions, and reinforcement learning for exploring adaptive strategies. 4.7. Evaluation Metrics To assess the effectiveness of the predictive models and the overall DSS framework, multiple evaluation metrics are employed: 4.7.1. Accuracy and Precision For classification tasks (e.g., churn prediction, sentiment classification), accuracy, precision, recall, and F1-score are measured. These metrics ensure that the system not only makes correct predictions but also minimizes false positives and false negatives, which is critical in managerial contexts. 4.7.2. Forecasting Accuracy For time-series models, metrics such as Mean Absolute Error (MAE), Root Mean Squared Error (RMSE), and Mean Absolute Percentage Error (MAPE) are used to evaluate forecasting performance. Lower error rates indicate stronger predictive reliability. 4.7.3. Return on Investment (ROI) Impact Beyond technical metrics, the DSS is evaluated in terms of its business impact. Simulated scenarios assess how the recommendations of the DSS influence key performance indicators such as revenue growth, customer retention, and market share. ROI is calculated by comparing the gains from DSS-informed decisions against the costs of system implementation and operation. 4.7.4. Managerial Usability and Trust Because strategic decisions require human judgment, the DSS is also evaluated qualitatively through manager feedback. This includes assessing the interpretability of the model outputs, the clarity of visualizations, and the extent to which managers trust and adopt the DSS recommendations. 5. Case Study 5.1. Business Environment The case study centers on a mid-sized omnichannel retail firm operating in a highly competitive consumer goods sector. The firm manages approximately 80 physical outlets across major metropolitan areas while also maintaining a rapidly growing e-commerce platform. Its product portfolio spans apparel, household essentials, and consumer electronics. Annual revenues exceed $500 million, but profitability margins are under pressure due to intense price competition, rising customer acquisition costs, and supply chain disruptions. The firm’s strategic challenge lies in balancing investments between its brick-and-mortar operations and its online platform. While online sales have grown steadily driven by digital marketing and changing consumer habits physical stores remain the firm’s core revenue base. Executives must decide how to allocate resources between these channels, specifically: • Whether to accelerate market expansion by opening new stores in secondary cities, or • Whether to channel investments into digital strategies, such as personalized promotions and dynamic online pricing. The decision is inherently strategic: it affects long-term positioning, capital expenditure, and the firm’s ability to defend its market share against both traditional competitors and digital-first entrants. World Journal of Advanced Research and Reviews, 2025, 27(02), 1752-1769 1767 [2] Kitsios, F., & Kamariotou, M. (2018). Decision support systems and strategic planning: Information technology and SMEs’ performance. International Journal of Decision Support Systems, 3(1/2), 53–70. https://www.inderscienceonline.com/doi/10.1504/IJDSS.2018.094260 ResearchGate [3] Batz, A. (2025). Integrating machine learning into business and management research: Intellectual landscape and strategic considerations. Humanities and Social Sciences Communications. https://doi.org/10.1038/s41599025-04361-6 Nature [4] Habel, J., Alavi, S., & Heinitz, N. (2023). A theory of predictive sales analytics adoption. AMS Review, 13, 34–54. https://doi.org/10.1007/s13162-022-00252-0 SpringerLink [5] Sarker, I. H. (2021). Machine learning: algorithms, real-world applications and research directions. SN Computer Science, 2, Article 160. https://doi.org/10.1007/s42979-021-00745-1 PMC [6] Heinrich, K., Janiesch, C., & Krancher, O. (2025). Decision factors for the selection of AI-based DSS… PLOS One (duplicate entry omitted) [7] Sprague, R. H., & Carlson, E. D. (1982). Building Effective Decision Support Systems. Prentice-Hall. [8] Keen, P. G. W., & Scott Morton, M. S. (1978). Decision Support Systems: An Organizational Perspective. Addison-Wesley. [9] Arnott, D., & Pervan, G. (2005). A critical analysis of decision support systems research. Journal of Information Technology, 20(2), 67–87. [10] Power, D. J. (2002). Decision support systems: Concepts and resources for managers. Greenwood. [11] Negash, S. (2004). Business intelligence. Communications of the Association for Information Systems, 13, 177–195. [12] Wixom, B. H., & Watson, H. J. (2010). The BI-based organization. International Journal of Business Intelligence Research, 1(1), 13–28. [13] Provost, F., & Fawcett, T. (2013). Data Science for Business. O’Reilly. [14] Waller, M. A., & Fawcett, S. E. (2013). Data science, predictive analytics, and big data: A revolution that will transform supply chain design and management. Journal of Business Logistics, 34(2), 77–84. [15] Bihani, P., & Patil, A. (2019). Managing churn using predictive analytics. Journal of Database Marketing & Customer Strategy Management, 26(3), 161–172. [16] Feldman, R., & Sanger, J. (2007). The Text Mining Handbook: Advanced Approaches in Analyzing Unstructured Data. Cambridge University Press. [17] Davenport, T., & Ronanki, R. (2018). Artificial intelligence for the real world. Harvard Business Review, 96(1), 108–116. [18] Sutton, R. S., & Barto, A. G. (2018). Reinforcement Learning: An Introduction (2nd ed.). MIT Press. [19] Silver, D., et al. (2016). Mastering the game of Go with deep neural networks and tree search. Nature, 529(7587), 484–489. [20] Hu, J., & Jennings, N. R. (2020). Multi-agent reinforcement learning for dynamic pricing. International Journal of Forecasting, 36(4), 1337–1348. [21] Gabel, T., & Nematbakhsh, R. (2021). Reinforcement learning for pricing and inventory control: A retailer’s perspective. European Journal of Operational Research, 293(3), 1077–1092. [22] Tornatzky, L. G., & Fleischer, M. (1990). The Processes of Technological Innovation. Lexington Books. [23] Wade, M., & Hulland, J. (2004). Review: The Resource-Based View and information systems research: Review, extension, and suggestions for future research. MIS Quarterly, 28(1), 107–142. [24] Teece, D. J., Pisano, G., & Shuen, A. (1997). Dynamic capabilities and strategic management. Strategic Management Journal, 18(7), 509–533. [25] Venkatraman, N., & Henderson, J. C. (1993). Real strategies for virtual organizing. Sloan Management Review, 34(2), 33–48. [26] Porter, M. E., & Millar, V. E. (1985). How information gives you competitive advantage. Harvard Business Review, 63(4), 149–160. World Journal of Advanced Research and Reviews, 2025, 27(02), 1752-1769 1768 [27] Bharadwaj, A. (2000). A resource-based perspective on information technology capability and firm performance: An empirical investigation. MIS Quarterly, 24(1), 169–196. [28] Sambamurthy, V., Bharadwaj, A., & Grover, V. (2003). Shaping agility through digital options: Reconceptualizing the role of information technology in contemporary firms. MIS Quarterly, 27(2), 237–263. [29] Melville, N., Kraemer, K., & Gurbaxani, V. (2004). Information technology and organizational performance: An integrative model of IT business value. MIS Quarterly, 28(2), 283–322. [30] Huang, M.-H., & Rust, R. T. (2021). Artificial intelligence in service. Journal of Service Research, 24(1), 3–21. [31] Ghahremanlou, A., & Zhang, J. (2020). AI-enabled Decision Support Systems: A case study in logistics. Decision Support Systems, 133, 113283. [32] Ladd, T., March, S. T., & Carr, J. (2013). Decision support in high-velocity environments: A framework for operational DSS. Communications of the Association for Information Systems, 33, Article 17. [33] Lyytinen, K., & Newman, M. (2008). Explaining information systems change: A punctuated socio-technical change model. European Journal of Information Systems, 17(6), 589–613. [34] Tafti, A. P., & Dewan, R. (2020). Business analytics and corporate performance: An integrated framework. Journal of Business Analytics, 3(4), 225–241. [35] Berman, S. J. (2012). Digital transformation: Opportunities to create new business models. Strategy & Leadership, 40(2), 16–24. [36] Davenport, T. H., & Kirby, J. (2016). Only Humans Need Apply: Winners and Losers in the Age of Smart Machines. Harvard Business Review Press. [37] Brynjolfsson, E., & McAfee, A. (2014). The Second Machine Age: Work, Progress, and Prosperity in a Time of Brilliant Technologies. W.W. Norton. [38] McKinsey Global Institute. (2020). The Age of Analytics: Competing in a Data-Driven World. McKinsey & Company. [39] Manyika, J., et al. (2017). Harnessing Automation for a Future That Works. McKinsey Global Institute. [40] Doshi-Velez, F., & Kim, B. (2017). Towards a rigorous science of interpretable machine learning. arXiv preprint arXiv:1702.08608. [41] Ribeiro, M. T., Singh, S., & Guestrin, C. (2016). "Why should I trust you?": Explaining the predictions of any classifier. Proceedings of the 22nd ACM SIGKDD International Conference on Knowledge Discovery and Data Mining, 1135–1144. [42] Mittelstadt, B. D., Allo, P., Taddeo, M., Wachter, S., & Floridi, L. (2016). The ethics of algorithms: Mapping the debate. Big Data & Society, 3(2), 2053951716679679. [43] Amershi, S., et al. (2019). Guidelines for human-AI interaction. Proceedings of the 2019 CHI Conference on Human Factors in Computing Systems, Paper 144. [44] Brown, T. B., et al. (2020). Language models are few-shot learners. Advances in Neural Information Processing Systems, 33, 1877–1901. [45] Bommasani, R., et al. (2021). On the opportunities and risks of foundation models. arXiv preprint arXiv:2108.07258. [46] Chen, M., et al. (2021). Evaluating large language models trained on code. arXiv preprint arXiv:2107.03374. [47] Teal, G., & Gartner, W. B. (2022). Real-time analytics for dynamic decision-making: A conceptual framework. Journal of Business Analytics, 5(1), 48–63. [48] Lustig, F., & Nabeth, T. (2023). Generative AI for decision support: A taxonomy and research roadmap. Decision Support Systems, 158, 113838. [49] Watson, H. J. (2018). The role of real-time analytics in strategic decision making. Journal of Decision Systems, 27(sup1), 5–10. [50] Żbikowski, K., Ostapowicz, M., & Gawrysiak, P. (2021). Deep reinforcement learning for resource allocation in business processes. arXiv preprint arXiv:2104.00541. ResearchGateU.S. Geological SurveyarXiv World Journal of Advanced Research and Reviews, 2025, 27(02), 1752-1769 1769 [51] Théate, T., & Ernst, D. (2020). An application of deep reinforcement learning to algorithmic trading. arXiv preprint arXiv:2004.06627. arXiv [52] Charpentier, A., Elie, R., & Remlinger, C. (2020). Reinforcement learning in economics and finance. arXiv preprint arXiv:2003.10014. arXiv [53] Francois-Lavet, V., Henderson, P., Islam, R., Bellemare, M. G., & Pineau, J. (2018). An introduction to deep reinforcement learning. arXiv preprint arXiv:1811.12560. arXiv+1 [54] Lifewire. (2023, June 21). What is reinforcement learning? Lifewire. https://www.lifewire.com/what-isreinforcement-learning-7508013 lifewire.com [55] Decision support system. (2025). In Wikipedia. https://en.wikipedia.org/wiki/Decision_support_system Wikipedia