In retail pricing strategy, a critical challenge is grouping products that respond similarly to market conditions. While traditional approaches rely on simple correlations or domain expertise, they often fail to capture complex temporal dependencies and true causal relationships. We present a novel framework that combines causal discovery with time series clustering to automatically identify products with similar sales drivers. Our approach analyzes Multivariate Time Series (MTS) data, incorporating both endogenous factors (prices, promotions, product interdependencies) and exogenous variables (climate, events, competitor behavior), using PC-MIC (Partial Correlation with Mutual Information Coefficient) to infer causal relationships. Experimental results on real retail wine sales data demonstrate that our method improves forecast stability and produces interpretable product groups that align with domain expertise.