ПРИМЕНЕНИЕ БАЙЕСОВСКИХ СЕТЕЙ ДОВЕРИЯ ДЛЯ УПРАВЛЕНИЯ РИСКАМИ РЕАЛИЗАЦИИ GO-TO-MARKET СТРАТЕГИИ ИННОВАЦИОННЫХ ИТ-ПРОДУКТОВ В УСЛОВИЯХ ИМПОРТОЗАМЕЩЕНИЯ

Abstract:
The relevance of quantitative risk management for Go-to-Market strategy implementation of innovative IT products in the Russian B2B/B2G market under import substitution is substantiated. The purpose of the study is to develop and test a methodology for applying Bayesian belief networks to the quantitative assessment and management of GTM strategy implementation risks. The research hypothesis is that Bayesian belief networks, which account for causal dependencies between risk factors, provide a more adequate assessment of aggregate risk than the classical probability-impact matrix, which systematically underestimates it. The scientific novelty lies in adapting the Bayesian network apparatus to the task of GTM strategy risk management in the EdTech segment of the Russian IT market and in the comparative quantitative analysis of the predictive and diagnostic capabilities of the two approaches. The information base comprises materials of the Russian EdTech startup DISTECH LLC, expert assessments of five specialists, and the project financial modelling data. A reference Bayesian network is constructed covering nine risk factors across six categories (technological, market, financial, organizational-personnel, regulatory, and competitive) with formalized conditional dependencies; probabilistic inference is implemented in Python (the pgmpy library) using the variable elimination algorithm. The marginal probability of GTM strategy failure in the baseline scenario is 0.653, whereas the classical risk matrix creates an impression of manageable aggregate risk. It is shown that a combination of three mitigation measures reduces the failure probability to 0.213, while the diagnostic mode identifies the financial-personnel block as the most probable cause of failure. The practical significance of the results lies in their applicability by early-stage commercial IT startups for prioritizing mitigation measures and substantiating a financial reserve policy. Directions for further research related to the empirical verification of conditional probabilities and the use of hybrid and dynamic Bayesian networks are outlined.