Abstract
For consumer market, it is important fact that the BtoC-EC (business-to-consumer electronic-commerce) market continues to grow. One of the effective marketing strategies is “reciprocal customer transfer” which a store introduces the store’s customers to another store to obtain customer loyalty. In this study, we analyze for effective reciprocal customer transfer between a golf course reservation site and a golf EC site to promote customer’s upsell. Firstly, we divided golf courses into 32 categories and the golf items into 40 categories. Next, we extracted the user’s usage history of these two services. Then, using these data, we carried out the time series association analysis. As a result of the analysis, we found some characteristic rules.
You have full access to this open access chapter, Download conference paper PDF
Similar content being viewed by others
Keywords
1 Introduction
In recent years, the BtoC-EC (business-to-consumer electronic-commerce) market has been expanding steadily. According to the definition of the Ministry of Economy, Trade and Industry, this market is composed of three fields: the field of merchandising, the field of service, and the field of digital. Markets themselves continue to be booming, and there can be no major issues that will impede market growth. However, the growth rate of the merchandising field in 2017 is 7.5%, which is slower than the growth rate of 10.6% in 2016 [1]. We thought that one of the future effective strategies to keep current growth in BtoC-EC market is that “reciprocal customer transfer”. It is a strategy to expand services by introducing customers to each other among different companies and services. For example, Softbank was deploying a service that triples a T point at Family Mart for its mobile phone subscribers [2], and JCB is making up a mechanism of reciprocal customer transfer with Person Holdings [3].
2 Purpose of This Study
In order to promote reciprocal customer transfer, it is necessary to analyze the relationship between the two services. In this research, we aim to activate reciprocal customer transfer between golf tool EC site and golf course reservation site. For that purpose, we analyze the mutual use situation between both services for the golf tool EC site and the portal site which manages the golf course reservation site.
3 Target Data
In this study, we focus on a golf course reservation history and a golf EC site purchase history from January 2016 to December 2017 of users possessed by a management company of both golf course reservation site and golf goods EC site. The target user is restricted customers who registered from January 2016 to June 2017 and come under the following formula (1) value is between 0.4 to 0.6.
There were 7947 users. The target golf courses are the average handicap of the user can be calculated out of the golf courses. Then, the number of golf course becomes 1867.
4 Analysis
In this section, we explain own analysis procedure.
4.1 Flow of Analysis
Firstly, we classified golf courses using their features such as average handicap of reserved users, the average utilization price per capita and the average review score. Secondly, we classified the product into some product categories. Next, we compiled the product by brand for each product classification. Then, we classified brands into 3 or 4 clusters based on average price and total sales volume for each product classification. Thirdly, we combined categorized golf courses and golf items with customer’s transaction history. Finally, we carried out time series association analysis using the customer’s transaction history.
4.2 k-Means Clustering
The k-means method is one of typical methods of non-hierarchical cluster analysis. By minimizing the evaluation function φ expressed by the following equation, it is divided into arbitrary k clusters.
\( \varvec{x}_{j } (j = 1, \ldots ,n) \) is the value for \( j \), and \( n \) is the number of cases. Also \( \varvec{c}_{i} \) is the center of cluster \( i \left( {i = 1, \ldots ,k} \right) \) [4].
4.3 Time Series Association Analysis
Time series association analysis [5] which is an analysis to extract effective association rules considered time transition of record. In usual association rule analysis, rules relating to simultaneous events are extracted. However, in time series association rule analysis, an association rule of events occurring at different times is searched in accordance with the temporal flow of the data designated by the time series. As a result, it is possible to clarify the temporal change in purchase behavior such as which product the person who purchased a certain product purchase next. Indicators are confidence, support and lift. The following will explain the indicators. Count (X) is the number of transactions including antecedent X, Count (X∩Y) is the number of transactions including both antecedent X and consequent Y, and U is the total number of transactions.
-
Confidence
Confidence shows the ratio at which antecedent X and consequent Y appeared at the same record among transaction data in which antecedent X appears. Confidence is calculated as follows.
If the value of this indicator is large, it can be said that it is a rule that the relation between the antecedent and the consequent is strong.
-
Support
The degree of support shows the proportion of combinations that will be the antecedent X and the consequent Y among the whole. Support is calculated as follows.
A large value of this indicator indicates that the rule is frequently performed. Conversely, if the value of this indicator is low, the rule is considered to have happened by chance and it is judged that it is not useful.
-
Lift
Lift expresses whether the assignment of the rule which is the antecedent X and the consequent Y increases more than the assignment of the rule leading to the consequent Y in the whole, by the ratio thereof.
If the value of this indicator is large, it can be judged that there is a relationship between X and Y. Because it is judged that Y was purchased on the antecedent that X was not purchased for Y alone reason.
5 Result and Discussions
In this section we show the analyzing result and discuss about them.
5.1 Result of Classification of Golf Course
The golf courses were divided into 32 categories based on three values which are the average handicap of users who reserved, the average utilization price per capita and the average review score. We divided the average handicap of users into four categories according to the quartile. From the top it was “for senior”, “for semi-senior”, “for semi-beginner”, “for beginner”. We divided the average utilization price per capita into four according to the quartile. From the top it was “expensive”, “relatively expensive”, “relatively low-price”, “low-price”. We divided the average rating points below the average and below “high” and “low”. We made 32 classifications by multiplying these three variables.
5.2 Classification Result by Product Category
We divided products into 11 categories based on the genre of the EC site. Next, we classified brands belonging to each the genre into 3 or 4 categories based on the average price and the number of sales of brand products. Also, I used the k-means method for classification for each category (Table 1).
We define brand classification with high sales quantity as “popular brands”, define low-price brand classification as “inexpensive brands”, define brand with a high price as “high brands”, the remaining brand classification was defined as “standard brands”.
5.3 Results and Discussion of Time Series Association Rule Analysis
We extracted rules with the support over1% and a lift over 1 among the obtained rules. In order to make rank the extracted rules, we calculated the standard score of each evaluation indicator and summed up then. Then we evaluated the rule with this value high as a useful rule. The characteristics of the obtained rule are shown below.
6 Results and Discussion Concerning the Rule from the Purchase of Goods to the Reservation of Golf Course
-
Rules from purchasing balls and accessories to booking golf courses (Table 2)
From these rules of purchasing balls and accessories to reserving golf courses, I found the following. First, purchase of a popular brand led to a highly rated golf course. Next, purchasing low-price brands led to semi-beginners or beginners centric, relatively low-price and low rating golf courses booking. Also, purchasing standard brands led to golf course for semi-senior, relatively expensive and high rating golf courses booking. Finally, purchasing high brands led to semi-seniors or seniors centric, relatively expensive and high rating golf courses booking. From the above, it is considered that there is a proportional relationship between the price range of purchased balls and accessories and the price, reputation and player level of the golf course to be used.
-
Rules from purchasing clubs to booking golf courses (Table 3)
We summarize the results and discussions about rules from purchasing clubs to golf course reservations. First, golf courses linked from the purchase of popular brands were semi-beginners centric, the low-price and the evaluation was low. second, golf courses linked from the purchase of standard brands were semi-senior-centric and expensive. Finally, golf courses linked from the purchase of low-price brands were semi-senior-centric and the evaluation was low. Purchasing clubs and golf courses for semi-beginners and semi-senior players are highly relevant. In addition, we obtained not the rules from the purchase of low-price brands to the reservation of the beginner-centered golf course but the rules from the purchase of low-price brands to the reservation of the semi-senior centered golf course. The reason is considered as follows (Fig. 1).
It is understood that the proportion of wedge is higher for low-price brands than for whole. Therefore, it can be inferred that player who is in the progress process stick to this tool more than beginners. The wedge is a club used for shots within 100 yards, approaches from around the green, and bunker shots.
From Fig. 2 it can be seen that the price of a low-price brand wedge is not cheaper than the whole wedge. Based on the above, we think that the ratio of wedges is high in the low-price brands is the reason for obtaining rules from the low-price brands to golf courses for semi-senior.
Results and Discussion Concerning the Rule from the Reservation of Golf Course to the Purchase of Goods
Overall, there were many rules linked to the purchase of balls, accessories and menswear from booking of golf courses. As there are no distinctive and definite rules compared to each golf course classification, we focus on the consequent of purchasing goods and consider the difference between the antecedent of the reserved golf course. First, we consider the rules that have the consequent of purchase to balls and accessories.
-
The rule from the reservation of golf course to the purchase of balls and accessories (Table 4)
Figures 3, 4 and 5 show the relationship between the characteristics of golf courses as the prerequisite and the brands of balls and accessories as consequent, using the maximum value of the total of the standard scores. The horizontal axis are each brand classification, and they are arranged in descending order of the average price. Also, the vertical axis shows the maximum value of the total of standard score. From Fig. 3, it can be seen that low-price brands have the highest total standard score in the golf course for beginner except that the popular brand which is a large volume of sales. Also, it can be seen that high brands have the highest total standard score in the golf course for senior except that the popular brand which is a large volume of sales. According to Fig. 4, in the low-price or relatively low-price golf course, it can be seen that the total of the standard score is the highest for low-price brands except for standard brands. Also, in high price golf courses, it can be seen that the sum of the standard points is the highest for the high price brands. From Fig. 5, it can be seen that low-price brands have the highest total standard score in the low rating golf course except that the popular brand which is a large volume of sales. Also, it can be seen that high brands have the highest total standard score in the high ratings golf course except that the popular brand which is a large volume of sales. From the above, the following can be considered in purchasing a ball on the antecedent of a golf course reservation. First, regardless of the characteristics of the prerequisite golf course, balls and accessories of the popular brands are most likely to be purchased most. Next, it is thought that the price, evaluation, the level of the player’s golf course used, and the price range of the balls and accessories are proportional.
-
The Rule from the Reservation of Golf Course to the Purchase of Men’s Wear (Table 5)
Figures 6, 7 and 8 shows the relationship between the characteristics of golf courses as the prerequisite and the brands of men’s wear as consequent, using the maximum value of the total of the standard scores. The rows on the horizontal axis are each brand classification, and they are arranged in descending order of the average price. Also, the vertical axis shows the maximum value of the total of standard score. There were no rules on high brands among men’s wear brand classifications. Looking at Fig. 6, the following can be seen. First, the standard brand showed the highest value in any type of golf course. Next, it can be seen that low-price brands have the highest total standard score in the golf course for beginner except that the popular brand which is a large volume of sales. In addition, standard brands showed the highest value in golf courses for senior. Figure 7 shows that low-price brands showed the highest value for low-price golf courses. Also, standard brands and popular brands showed high values for expensive golf courses. In addition, standard brands as the price of golf courses rises, the total of standard scores is rising. From Fig. 8, the following can be seen. Firstly, the score of the popular brands is high regardless of the evaluation of the golf course. Secondly, the difference of the standard score between ones which are low rating and golf courses which are high rating is not seen, but for popular brands, golf courses with high ratings have a higher standard score than ones with low ratings. From the above, the following was found out.
First of all, we think that it tends to lead to the purchase of low-price brand men’s wear from reservations of low-price and golf courses for beginner. Next, we think that it is easy to be associated with the purchase of standard brands men’s wear from reservations of expensive golf course or golf course for senior.
7 Consequent
In this research, we analyzed the relationship between the purchase of golf tool and the reservation of golf course at the EC site to promote reciprocal customer transfer between these two services. Through this analysis, the relationship between the golf course and each product was understood. By using the results obtained in this study, it is considered that reciprocal customer transfer can be encouraged between the golf course EC site and the golf course reservation site. Therefore, we believe that it is possible to expand customer share and advance upselling, which can lead to an increase in site loyalty.
We are considering the following as future works. First, some subjectivity is included in interpreting association rules. Hence, we need to interpret objectivity by using other analysis methods. Second, we should analyze more detail by using access logs to purchase and reservation, then we many obtain more effective rules by customer site wage.
References
Ministry of Economy: Trade and Industry Ministry of Economic Affairs Commerce Transaction Bureau. Infrastructure development of data driven society in Japan (e-commerce market research) (2018, in Japanese). http://www.meti.go.jp/press/2018/04/20180425001/20180425001-2.pdf
Nihon Keizai Shimbun website (in Japanese). https://www.nikkei.com/article/DGXLASDZ22HMR_S5A720C1TJC000/. Accessed 29 Jan 2019
Sankei Shimbun website (in Japanese). https://www.sankei.com/economy/news/181108/prl1811080407-n1.html. Accessed 29 Jan 2019
Onoda, T., Sakai, M., Yamada, S.: Experimental comparison of clustering results for k-means by using different seeding methods. In: Conference of the Japanese Society for Artificial Intelligence of Japan, vol. 25, pp. 1–4 (2011, in Japanese)
NTT DATA Mathematical System, Inc.: Visual Mining Studio Manual (2018, in Japanese)
Kim, M.: Data Science by R. Morikita Publishing (2012, in Japanese)
Acknowledgment
We thank Golf Digest Online Inc. for permission to use valuable datasets and for useful comments. This work was supported by JSPS KAKENHI Grant Number 16K03944 and 17K13809.
Author information
Authors and Affiliations
Corresponding author
Editor information
Editors and Affiliations
Rights and permissions
Copyright information
© 2019 Springer Nature Switzerland AG
About this paper
Cite this paper
Hirota, K., Otake, K., Namatame, T. (2019). Reciprocal Customer Transfer Analysis at Golf Course Reservation Service and Golf Goods EC Site. In: Meiselwitz, G. (eds) Social Computing and Social Media. Communication and Social Communities. HCII 2019. Lecture Notes in Computer Science(), vol 11579. Springer, Cham. https://doi.org/10.1007/978-3-030-21905-5_28
Download citation
DOI: https://doi.org/10.1007/978-3-030-21905-5_28
Published:
Publisher Name: Springer, Cham
Print ISBN: 978-3-030-21904-8
Online ISBN: 978-3-030-21905-5
eBook Packages: Computer ScienceComputer Science (R0)