電子商務網站最大價值顧客之分析 —消費者購買集中度指標之模型建構

Abstract

對於電子商務網站而言，如何更清楚地界定其最大價值顧客？是那些經常拜訪該網站卻購買很少，還是那些較少光臨卻每次購買很多的網友？若從網友的搜尋傾向來分析，是那些常常造訪者，還是那些每次來總是流覽許多網頁的網友？這篇論文即是採用消費者購買集中度的觀點，來探討上述議題。本文將消費者購買集中度視為行銷衡量指標，計算網友在不同搜尋傾向上的購買集中度比值，透過相互比較來發展出診斷最大價值顧客行為之策略。本文也建構一個衡量消費者購買集中度的二維構面，分別從網友的搜尋傾向來分析其在購買次數與購買金額上的購買集中度，藉以確認其最大價值顧客的線上行為模式。 本文運用comScore的資料庫，所分析的電子商務網站是選取那些購買頻次較高的產品類別，它們是書藉雜誌類、音樂類與美容保養類網站。研究發現，(一)購買集中度的計算，當分別考慮是否包含無任一購買記錄的網友時，有些類別的網站會出現不同的結果，此意味著排序在最前面百分比的顧客，其購買行為模式會依計算的基礎而不同。此結果同時出現在書藉雜誌類的兩個網站，但音樂類僅出現在columbiahouse網站，或許這是由於整個產品類別的特性，及網站特性的因素使然。 (二) 透過購買集中度的比較，行銷工作者可以擬定未來該網站相對其主要競爭者，想要達到最大價值顧客所佔排序在前的百分比為多少。(三) 透過購買集中度比值的比較，行銷工作者可以判斷該網站相對其主要競爭者，排序在前最大價值顧客的搜尋傾向為何，做為其擬定行為策略的專注焦點。例如，在美容保養類網站，avon的最大價值顧客相較於melaleuca就有較明顯的搜尋傾向差異，不論是在購買次數或是購買金額上。 總而言之，即使在所觀察的期間，不論是購買次數或購買金額，消費者購買集中度皆沒有明顯的差異，但只要能持續追蹤自身及整個產品類別購買集中度的變化，並隨時偵測該比值與主要競爭者的差異，購買集中度指標對於行銷策略的發展仍有其助益。

While examining the most valuable customers for an e-commerce site, should one focus on the customers who visit the site frequently but spend less, or those who visit the site infrequently but spend more on each visit? In terms of the consumer search propensity, should we focus on the customers who visit the web site frequently or those who make in-depth search for each visit by having a larger number of page views? To answer these questions, this dissertation take a perspective of customer concentration which can serve as a measuring metric to develop diagnostic strategies from the comparisons of different concentration degree ratios counted by means of customer’s search propensity, i.e., the number of repeat visits and the number of page views on each site visit. We also provide a framework to measure the concentration degree from two dimension, they are the aspects of customer value (purchase frequency and monetary value) as well as the customer’s search propensity, respectively. Therefore, our proposed model can identify the behavior patterns of the most valuable customers for an e-commerce site. We apply the modeling approach to the comScore database with a focus on frequently purchased product categories such as books, music, and health & beauty. From the empirical results, we have three main findings. First, some of the concentration degree ratios are different when calculated by including zero class versus by excluding zero class, which means there are different behaviors of search propensity for top percentile of customers while we calculate the concentration degree from different bases. These results occur at both amazon.com and barnesandnoble.com in the category of books & magazines, but just occur at columbiahouse.com in the category of music. They might be taken as product-specific feature, or website-specific feature, respectively. Second, from the comparisons of concentration degree curve, marketers can choose which extent of concentration degree is their targeting top percentile of customers in the future by comparing with its main competitor. Third, from the comparisons of concentration degree ratios, marketers also can decide which search propensity of their top percentile of customers is their focus by comparing with its main competitor. For example, in health & beauty category, both of the deviations of two concentration degree ratios in avon.com are manifestly greater than those of concentration degree ratios in melaleuca.com, no matter purchase frequency or monetary value, which reveals that the marketing strategy based on customer’s search pattern should clearly focus attention more on one of the search propensities. In sum, even though there might be no significant difference among concentration degrees no matter on purchase frequency or monetary value in observed time period, it is useful for strategy development to keep on tracking the changes of concentration degree for itself and the whole product category, and monitoring the comparisons of concentration degree ratio with the target competitors.