電子商務經營模式下之商品生命週期研究

Abstract

「商品」是零售業經營的核心，商品經營要能夠獲利，除了上市初期合理的成本、定價、毛利外，零售商對商品經營效率的追求更是最終是否獲利的關鍵，這關鍵在於商品經營要如何做到「高動銷、週轉快、不缺貨」。商品生命週期是一套描述商品從出生到死亡的理論架構，基於此架構，前人對於生命週期每個階段的經營策略多有研究。 過去針對商品生命週期的研究，多針對特定產業、特定類目、甚至特定商品的研究，推測是因為較少跨產業、跨類目、跨商品的完整數據。本研究立基於某台灣大型電商服務平台的數據，採用該電商服務平台2022年1月1日至2023年3月31日間的銷售資料，該平台服務的商家多、類目跨度大，因此可進行較大範圍的研究。另外亦抽出重點商家調研其橫跨3年的銷售數據，透過這些資料研究電商經營場景下各類商品的生命週期。本論文以k-means聚類法為基礎，結合動態時間校正法（Dynamic Time Wrapping, DTW）來整合具時間差的生命週期序列，再從類目、商店等不同面向切入，對商品的生命週期做分群並進行解釋。研究中也提出了有別於傳統生命週期可視化的方式，該方式有助於解決過去生命週期計算方式容易受短時間銷售影響導致趨勢不明確的問題。 研究最終總結了電商環境下商品生命週期的分群特性，並針對該特性的成因提出推論，最後基於最終分群的結果，提出對商品經營方法的建議，這些建議對於零售商制定電商的商品經營、庫存策略具有重要參考價值。

"Product" is the core of retail business operations. To ensure profitability in product management, retailers must consider reasonable costs, pricing, and gross margin during the initial launch. However, the pursuit of operational efficiency in product management is crucial for long-term profitability. This efficiency is achieved through high sales, fast turnover, and avoiding stockouts. The product lifecycle is a theoretical framework that describes the journey of a product from birth to death. Based on this framework, previous studies have extensively researched management strategies for each stage of the product lifecycle. Past research on the product lifecycle has mostly focused on specific industries, categories, or even individual products, likely due to the limited availability of comprehensive data spanning multiple industries, categories, and products. This study is based on data from a large Taiwanese e-commerce platform, specifically the sales data between January 1, 2022, and March 31, 2023. The platform serves numerous merchants across a wide range of categories, allowing for a broader scope of research. Additionally, key merchants were selected for an investigation using three years of sales data to study the product lifecycle within the e-commerce business context. The study employed the k-means clustering method as a foundation, combined with Dynamic Time Wrapping (DTW) to integrate time-adjusted lifecycle sequences. It approached the product lifecycle clustering and interpretation from various dimensions such as category and store. The study also introduced a visualization method that differs from traditional lifecycle visualization. This method helps address the issue of unclear trends caused by the influence of short-term sales on lifecycle calculations. The thesis ultimately summarizes the clustering characteristics of the product lifecycle in the e-commerce environment and makes inferences about the underlying causes of these characteristics. Based on the results of the final clustering analysis, recommendations are provided for product management strategies. These recommendations are of significant value for retailers in formulating e-commerce product management and inventory strategies.