- 學位論文
地方百貨應用大數據與機器學習的消費者分析：精準行銷與傳統行銷手法比較
Consumer Analysis Using Big Data and Machine Learning in A Local Department Store: A Comparison of Precision Marketing and Traditional Marketing Method
若您是本文的作者，可授權文章由華藝線上圖書館中協助推廣。
摘要
本研究旨在深入探討人工智慧 (AI) 和機器學習技術在地方 (社區) 百貨公司行銷策略中的應用，尤其關注其對活動投資回報率 (ROI) 和客戶轉化率的影響。隨著數位行銷的發展，百貨公司愈發依賴精準行銷來提高營收，然而傳統行銷方式依然在某些情境中具有一定優勢。本研究的核心在於比較傳統行銷方式與 AI 技術驅動的精準行銷方法，以了解這些技術如何在實際環境中為企業帶來更高的回報。 研究設計一個為期兩週的美食展活動，並設立兩種不同的折價券發送方式進行 A/B 測試。第一種為傳統行銷方式，消費者可在百貨公司的 APP 中限時搶折價券，兌換成功後可於指定櫃位消費使用。此方式依賴於大眾推廣，且根據過去數據顯示，兌換率約為 50%。第二種方式則是利用 AI 模型，通過機器學習預測模型篩選出最有可能參與活動的顧客，將折價券直接發送給這些高潛力顧客，並通過推送通知提醒他們使用。這種精準行銷方式旨在提高行銷效益，同時降低折價券發送和推廣成本。 在模型訓練與應用的過程中，我們選用了 XGBoost 作為主要的機器學習算法。XGBoost 具有優越的效能和靈活性，特別適合處理大規模數據，並且能夠有效處理異質數據與缺失值。 研究結果顯示，通過 AI 精準行銷策略，百貨公司能夠顯著提升客戶轉化率和 ROI。相比傳統行銷方式，精準行銷不僅能夠降低行銷成本，還能精確鎖定高價值客戶群體，進一步提高行銷活動的投資回報率。具體來說，AI 驅動的行銷策略能更有效地圈選出潛在顧客，這些顧客的購買行為預測更為準確，從而達到提升營收的目標。尤其是在折價券發放環節，AI 技術能夠自動化推送策略，使得每張折價券的效益最大化。 然而，本研究也發現，傳統行銷方式在吸引新顧客方面依然具有一定的優勢。儘管精準行銷能有效識別高潛力顧客，並提升回購率，但對於首次參與活動的顧客或非會員顧客來說，傳統的廣告推廣依然能夠帶來不錯的效果。因此，未來的行銷策略應結合兩種方法，充分發揮精準行銷和傳統行銷的互補優勢。
並列摘要
This study explores the application of Artificial Intelligence (AI) and machine learning technologies in the marketing strategies of local department stores, focusing on their impact on return on investment (ROI) and customer conversion rates. As digital marketing continues to evolve, department stores increasingly adopt precision marketing to boost revenue. However, traditional marketing methods retain certain advantages in specific scenarios. The core objective of this research is to compare traditional marketing approaches with AI-driven precision marketing techniques to evaluate their effectiveness in real-world environments. A two-week food festival event was designed as the research framework, featuring two coupon distribution methods via an A/B test. The first approach utilized traditional marketing, where consumers could claim time-limited coupons through the department store’s app, redeemable at designated counters. This mass-promotion strategy achieved a redemption rate of approximately 50% based on historical data. The second approach applied an AI model, specifically XGBoost, to predict and target high-potential customers. Coupons were sent directly to these customers, supplemented with push notifications as reminders. This precision marketing strategy aimed to enhance marketing efficiency and minimize coupon distribution and promotional costs. XGBoost was chosen as the primary machine learning algorithm due to its superior performance, flexibility, and ability to handle large-scale datasets, heterogeneous data, and missing values. The results indicate that AI-driven precision marketing significantly improves ROI and customer conversion rates. Compared to traditional methods, this approach reduces marketing costs while effectively targeting high-value customer groups, further optimizing the ROI of marketing campaigns. AI-based strategies demonstrated higher accuracy in predicting customer purchasing behavior, directly contributing to increased revenue. In addition, automated push strategies allowed for the maximization of each coupon’s effectiveness. The study also highlighted the enduring value of traditional marketing methods in attracting new customers. While precision marketing effectively identifies high-potential customers and increases repurchase rates, traditional approaches remain effective for engaging first-time participants or non-member customers. Consequently, future marketing strategies should integrate both methods to capitalize on their complementary strengths and achieve comprehensive marketing success.
參考文獻
Akter, S., & Wamba, S. F. (2016). Big data analytics in E-commerce: a systematic review and agenda for future research. Electronic Markets, 26, 173-194.
Al Khaldy, M. A., Al-Obaydi, B. A. A., & al Shari, A. J. (2023, May). The impact of predictive analytics and AI on digital marketing strategy and roi. In Conference on Sustainability and Cutting-Edge Business Technologies (pp. 367-379). Cham: Springer Nature Switzerland.
Aydin, Z. E., & Ozturk, Z. K. (2021, March). Performance analysis of XGBoost classifier with missing data. In 1st Int. Conf. Comput. Mach. Intell., no.
Babin, B. J., Darden, W. R., & Griffin, M. (1994). Work and/or fun: measuring hedonic and utilitarian shopping value. Journal of consumer research, 20(4), 644-656.
Chen, H. (2023). Enterprise marketing strategy using big data mining technology combined with XGBoost model in the new economic era. Plos one, 18(6), e0285506.