NEURAL NETWORK TECHNOLOGIES IN THE MARKETING MANAGEMENT SYSTEM OF ENTERPRISES

Abstract

The article examines the theoretical foundations of neural network technologies application in the marketing management system of industrial enterprises. The relevance of artificial intelligence implementation as an adaptive marketing tool under logistics crisis conditions caused by the wartime state in Ukraine and global supply chain disruptions is substantiated. It is established that traditional marketing  analysis tools demonstrate insufficient efficiency under conditions of sharp market uncertainty growth, demand volatility, and sales channel reorientation, which are characteristic of crisis phenomena. A typology of neural network architectures according to marketing management functions is systematized: LSTM and GRU for demand forecasting, CNN for consumer segmentation, Transformer for NLP competitive environment monitoring, and Autoencoder for anomaly detection in logistics flows. An original concept of the Neural Network Marketing Management System (NNMMS) operating at three levels — operational, tactical, and strategic — is proposed and described. A comparative analysis of traditional and neural network-based management systems is conducted across seven criteria: decision-making speed, forecasting accuracy, disruption response, unstructured data processing, scalability, implementation cost, and ERP/CRM integration. It is established that LSTM model implementation reduces demand forecasting error from ±15–25% to ±5–8%. A SWOT matrix for neural network technology deployment in industrial enterprises’ marketing activities is developed, considering wartime conditions and the European integration context. Three priority directions for digital transformation of marketing functions are identified: real-time LSTM-based analytics, NLP-driven competitive environment monitoring, and autoencoder-based logistics anomaly detection with potential reduction of supply disruption losses by 18–27%. The research results form a methodological basis for further development of neural network management systems and marketing resilience models for industrial enterprises under crisis conditions.