基於搜尋移動路徑探勘消費者網路購物資訊行為

Abstract

本研究旨在了解在不同情境底下的購物任務中，消費者的行為是否具有差異。採用 ZOSTs 搜尋移動概念及 K-means 分群法，並設計具明確購物目標及不具明 確購物目標兩種情境之模擬購物實驗，分析消費者在購物網站上的搜尋瀏覽行為， 並且歸納出消費者不同的搜尋行為模式及特徵，藉此了解購物網站所提供之推薦 功能對購物決策的效益。研究結果發現，在具有明確購物目標的任務之下，消費 者能精準的鎖定在特定類型下的商品，透過網站中的關鍵字搜尋功能來找尋商品; 而在不明確的購物目標之下，消費者仰賴網站的推薦功能或是已有的商品分類來 幫助塑造商品的需求，產生多樣的搜尋瀏覽行為。另根據 20 位受試者的瀏覽數 據分群結果，區分出瀏覽型、明確型及其延伸之搜尋型消費者的存在，兩大類型 之消費者在「瀏覽不同類別次數」、「搜尋比較頁面佔比」、「活動/推薦頁面佔 比」方面具有顯著的差異。根據以上的實證研究結果，提出針對購物網站功能改善之建議，並提供賣家制定更精準行銷策略之參考。The aim of this research is to understand whether consumers' shopping behavior vary in the context of different shopping tasks. We first designed two types of shopping tasks: goal-oriented shopping and exploratory-based shopping. We then used zero- order state transition matrices (ZOSTs) and K-means clustering algorithm to analyze consumers' online shopping behaviors. Through clustering analysis, we identify different search patterns for two types of consumers with different tasks and examine the effectiveness of the recommendation functions (RFs) offered by the shopping website for online shopping decision making. The results show that the goal-oriented consumers tend to focus on specific type of products and use keyword search to find out what they want; while the exploratory-based consumers rely on the RFs or existing categories to help them clarify needs. In addition, there are significant differences between goal-oriented and exploratory-based consumers in “ category variety measure”, the percentage of pages that were “search result pages” and “RFs pages”. The findings can provide a reference for sellers to develop precision marketing strategies.