本次參加的主要目的是發表論文「Phone-based Data Collection for Understanding Consumer Flow Behavior to Physical Stores」，此文 之共同作者為中研院博士後研究游創文與 台大資工系朱浩華教授，為一跨領域研究，摘要如附件。由於本研討會本質仍屬行銷與消
(英文) The 18th Recent Advances in Retailing & Services Science Conference
(European Institute of Retail and Services Studies, EIRASS)
發表論文 題目
(中文)
以行動電話自動搜集消費者前往零售店動線研究 (英文)
Phone-based Data Collection for Understanding Consumer Flow Behavior to Physical Stores
研究關注的重點。
3
Phone-based Data Collection for Understanding Consumer Flow Behavior to Physical Stores
An important question in consumer behavior research is how to systematically and quantitatively determine patterns in consumer behaviors that can facilitate understanding of where, when, and how consumers purchase products and services at (non-online) physical retail shops. Collecting naturalistic data on real consumers who shop at retail stores is often one of the most challenging and expensive parts of consumer behavior studies. This research proposes phone-based data collection to consumer behavior research.
Mobile phones have become indispensable part of our everyday lives.
New mobile phones are equipped with sophisticated sensing, computing, and communication capabilities. For example, new smart phones have a variety of sensors including GPS, accelerometer, digital compass, Wi-Fi, and cell-ID sensors that can detect consumers’ locations and movements.
By taking advantages of phones’ ubiquitous presence with consumers and phones’ sensory capability to observe consumers everywhere, it is
possible to leverage and organize these phones (which are user-owned and –maintained) and to build naturalistic and low-cost data collection systems that capture spatially relevant information of user behavior at large scales [Madan 2010][Gonzalez 2008]. Such data collection of user behavior enables geographical and quantitative analysis of where, when and how urban consumers visit their neighborhood convenient stores (CVS) invisibly and non-intrusively, i.e., without disruption to human natural behaviors.
We believe that phones provide opportunities to outsource the process of collecting customer flow data to any local residence who owns and/or carries a mobile phone and is also a customer of neighborhood CVS stores. Outsourcing data collection to consumers can significantly reduce the cost for consumer behavior researchers to run quantitative marketing studies. Furthermore, phones provide opportunities to automate the data collection process by embedding smart sensing, detecting, and logging of customers’ CVS trips in the phones. Automating data collection does not only enable gathering of consumer behavior naturally without
interrupting users’ activities, but also reduces underreporting and recall errors found in the traditional self-reporting, face-to-face interview, and surveying methods.
We have developed a phone-based data collection system. This system works by enabling consumer behavior researchers to recruit qualified residence, who live or work in an area of interest, to participate in the data collection process. Participants first download an application to their phones, in which the phone application embeds automated sensing to detect trips to CVS outlets and also logs CVS patronage data in their phones. The phone application runs in the background and does not disturb participants’ normal phone’s usage. Periodically, participants upload data from their phones to a data repository on a server. For security and privacy purpose, the phone application must ask and obtain user permission prior to any data uploading. At the end of uploading data, participants can optionally help in correcting any mistake made by automated sensing and/or label meta-data description (e.g., purchased items, purchased amount, etc.) about their CVS visits. Then, the server
processes data stored in the data repository, while summarizing and visualizing customer flow behavior to the consumer behavior researchers.
To encourage participation in data collection, consumer behavior researchers can set incentive policies that reward micropayments to participants based on the quality and quantity of their uploaded data.
We have deployed and tested the system by collecting real customer flow data from 42 participants who made 394 pedestrian trips to three
competing CVS stores situated within the same neighborhood area. To compare the data collected using our phone-based data collection system to those using the traditional data collection method, we also ran a pen and paper survey that involved face-to-face interviews with 90 customers of these CVS stores. Preliminary results from this comparison user study showed that (1) the phone-based data collection system achieved over 90% accuracy in detecting CVS store visits, and (2) consumer flow data obtained from our phone-based data collection system had little
difference to consumer flow data obtained from the pen and paper survey involving face-to-face interviews in term of distributions of consumer inbound/outbound directions.
This research promotes this new area of applying everyday phone sensing to consumer behavior and marketing research and opens a new door on practical use of phones from everyday consumers to automatically sense and report consumer behavior. Our future work will consider security and privacy issues associated with phones collecting consumer data as well as new incentive policies put in place, such as location-triggered coupon delivery to phones, to encourage everyday consumers to participate in data collection. Our future work will also explore scalability potentials in phone-based data collection from recruiting a large number of
participants who will cover a large geographical area and shop at different varieties of stores.
We believe that the development of phone-based data collection systems will lead to better quantity and quality of consumer data available to physical store retailers for mining and understanding of their customers, i.e., comparable to the wealth of consumer data collected and mined from online customers and available to online retailers.
References
Gonzalez, M., Hidalgo, C., and Barabasi, A.-L. Understanding Individual Human Mobility Patterns. Nature, 453:779–782, 2008.
Madan, A., Cebrian, M., Lazer, D., and Pentland, A. Social Sensing to Model Epidemiological Behavior Change. In Proceedings of 12th ACM International Conference on Ubiquitous Computing (UbiComp’10), ACM, New York, NY, USA, 291-300, 2010.