本研究主要探討因果效應模型在零售業中對於確認促銷導致的銷售侵蝕事件以及分析推廣活動所產生的影響。目前針對零售業使用因果效應模型分析銷售侵蝕以及推廣活動影響的研究並不常見，在過去的研究中針對此一議題大多是使用A/B測試進行研究，但在零售業中進行A/B測試需要實驗組與對照組，時間以及成本都較高，且在很多的狀況下實驗組與對照組之間可能還會有其他的因素影響。而透過因果效應模型可以更加的簡單針對零售業的各項促銷以及推廣活動進行評估，所以本研究希望透過台灣連鎖零售之個案，確認此一模型對於銷售侵蝕與活動影響評估之可用性。 本研究使用之因果效應模型使用貝氏結構時間序列模型進行因果推論，使用過去的銷售資料進行反事實的預測，推估促銷以及推廣活動可能造成之影響，研究發現使用此一模型可以確認促銷時造成的銷售侵蝕狀況，研究中的個案在分別在北區、中區、南區的零食類商品中發現數個銷售侵蝕事件，在不同的類別中皆有銷售侵蝕的狀況。研究也發現推廣活動對於相關商品會造成影響，根據研究中的個案發現在北區推廣活動對於相關商品造成的影響最明顯，在北區發現與推廣活動具有較強因果關係的商品有十多種，中區與南區與推廣活動舉有高度因果關係的商品則少於北區，此一研究結果可作為零售業在進行促銷和推廣活動時的參考，唯一需注意的是需要排除可能發生的其他因素影響，如產品的季節性影響等，以獲取更有價值之研究成果。 ;This study primarily investigates the application of causal impact models in the retail industry to confirm instances of cannibalization resulting from promotional events and analyze the effects of events on sales. Currently, research on using causal impact models to analyze cannibalization and the impact of events on sales in the retail industry is relatively rare. Previous studies mostly relied on A/B testing for this issue. However, conducting A/B tests in the retail industry can be resource-intensive in terms of time and cost, and in many cases, there may be other factors influencing the experimental and control groups. Through the use of causal impact models, it becomes more straightforward to assess various promotions and events on sales in the retail sector. This study aims to confirm the utility of this model through a case study of a Taiwan-based chain of retail stores. The causal impact model employed in this research utilizes a Bayesian structural time series model for causal inference. It leverages historical sales data for counterfactual prediction, estimating the potential effects of promotions events and events on sales. The study finds that this model can confirm instances of sales cannibalization resulting from promotions and assess the impact of events on sales on related products. several instances of sales cannibalization were identified in different snack product categories within the North, Central, and South regions, and this cannibalization was observed in various categories. The research also found that events on sales had an impact on related products. According to the results of case study, events on sales had a pronounced impact on related products and over ten product categories exhibited a strong causal relationship with events on sales in the North region. Meanwhile, the Central and South regions had fewer product categories with a high causal relationship with events on sales. These research findings can serve as references for the retail industry when conducting promotions and marketing activities. However, it is essential to account for potential external factors, such as seasonal influences, to obtain more valuable research results.
