A statistical approach to participant selection in location-based social networks for offline event marketing

Abstract

Offline event marketing has become increasingly popular. As a large amount of data from location-based social networks (LBSNs), such as Foursquare, Gowalla, and Facebook, becomes available, how to make use of these data to analyze users’ social behaviors is an important issue for offline event marketing. To provide some valuable guidance for businesses, this paper presents a statistical inference approach to optimally selecting participants who have a high probability of visiting an offline event. Technically, we formulate participant selection as a constraint optimization problem. In particular, our marketing cost model takes into account key factors such as distance, loyalty influence, and recommendation index. In addition, four participant-based strategies and a detailed algorithm are presented. Experiments on real-world datasets have demonstrated the effectiveness and efficiency of our proposed approach and the quantitative model.

Introduction

Markets that rely on media channels without being connected to the World Wide Web are called “offline” markets [17], [19], [20], [24], [29]. In other words, offline marketing utilizes offline media channels to create awareness of a company's products and services. Offline media campaigns include radio and print advertising, telemarketing, and television ads [29]. Given the Internet's tremendous rise in popularity [1], [8], [23], however, a company can accumulate data on a large number of users and merchandise through devices such as mobile phones [4], [6], [21], sensing devices [5], [8], [9], [32], and RFID [10], [11], [20]. These data have a wealth of marketing value by which to improve the effectiveness of advertising. In fact, offline event marketing and associated strategies have received increasing attention [17], [18].

An offline marketing event usually contains the following attributes [22], [30].

• Location: The address of a company or a store that holds the event in terms of its longitude and latitude. The place for holding an event is usually a business venue where products or services can be consumed by customers.

• Scale: The scale of an event is measured by the number of customers that are allowed to participate in the event. Sometimes, the more participants attending the event, the better. However, given the capacity of a venue, only a limited number of participants can be served. This work aims to invite a limited number of customers who are highly likely to participate in an event at a venue. At the same time, the expected number of visiting participants can equal a planned scale, called a scale goal.

• Served items: Products or services in an event as well as their corresponding prices.

• Others: A brief description of an event, such as the event duration, or some principles in selecting the participants to whom to present the event.

Except for advertising campaigns and creating promotional materials, the basic targets for any offline marketing strategy are to increase online traffic, overall sales, and profits [14], [16]. For this purpose, various strategies are available. Among them, three representative strategies are described as follows:

• Direct mail - one of the most commonly used mediums in offline marketing campaigns. Mailing lists based on demographic data should include customers who are most likely purchase the products. For instance, insurance companies may desire lists of customers who have recently purchased a new car or other things. This strategy only considers one factor. In fact, many factors can affect customers purchasing products, such as the travel distance of the customers purchasing the products and their interests.

• Discount pricing - Another way of attracting customers is to offer price discounts, such as coupons. Department stores, supermarkets and companies often insert their coupons into local newspapers or leaflets to advertise the sales of their products and services. Although offering discounts will increase the desire for customer consumption, this strategy is not feasible for consumers who do not need such products or services. This scheme increases some unnecessary costs of advertising.

• Loyalty programs - Loyal customers are an important resource for companies. To keep these customers and attract new customers, some preferential policies are introduced to encourage them to patronize businesses more frequently [32]. For example, a pet store may reward customers with a can of dog food on their tenth visit. Although marketers provide some preferential policies for encouraging customers, the probability of customers visiting companies may also be lower due to the long travel distance.

The essential problem of offline event marketing is how to propagate and advertise an event at a minimized marketing cost while fulfilling the scale and item coverage constraints. In other words, participants with a larger probability of attending an event are selected so that the resources invested in the event can be consumed as much as possible [7]. Some simple strategies, such as “nearest people first” (NF), random selection (RS), or first come first served (FCFS), cannot guarantee that the marketing cost is minimized. Instead, these strategies merely guarantee that the number of invited participants will not exceed the volume of an event as planned. A better strategy is “most potential people first” (MF). This assumes that the more potential a participant has, the more likely he/she will attend the event. Thus, the problem becomes how to measure the potential of a participant [25], [26], [27]. For offline event marketing, several important factors should be considered when applying the potential influence in participant selection: 1) distance: people are inclined to visit venues that are closer to their home locations. Thus, a far distance may cause a low visiting probability [15]; 2) item coverage: it would be better if all product items served by a venue could be consumed by the customers. Therefore, each item should be liked by the participants; 3) loyalty influence: if a customer is very loyal to a venue, then advertising to him/her is necessary because he/she is likely to come to the venue again [13]; and 4) influences from friends (called recommendation index): if a customer has many friends with considerable loyalty to a venue, then he/she has a high potential of visiting the venue because his/her friends may recommend the venue to him/her [25]. Our approach takes into account these factors for offline event marketing.

The main contributions of this paper are as follows.

• We form offline event marketing as a constraint optimization problem.

• We introduce the statistical inference approach by considering the distance factor, loyalty influence and recommendation index. Based on this approach, an effective algorithm is given.

• Four strategies that select customers based on three factors (distance, loyalty influence and recommendation index) are proposed:

• We conducted extensive experiments, which demonstrate the performance of the proposed algorithm and four strategies. The results show that strategy 3 based on unidirectional loyalty local influences and strategy 4 based on bidirectional loyalty local influences have better performance.

The rest of this paper is organized as follows. Section 2 reviews some related work. The system model and the problem of offline event marketing are well defined in Section 3. Section 4 presents an algorithm for minimizing the marketing cost. The experimental results are reported in Section 5. Finally, we conclude this paper in Section 6.