Wi-Fi Based Retail Analytics For Improved Customer Experience

Executive Summary

Footfall retail data and indoor location finding by using WI FI is considered as most emerging technology in tracking industry. The main inspiration of this project is the passive localization and tracking of customer in indoor environments using Probe Requests send by customers’ smartphones. These probe requests are later filtered to get MAC Address of devices and used for device tracking. Indoor customer localization enables retailers to know about customer traffic surrounding a store and converting it into in-store visits. Realtime retail analytics will allow retailers to make operational and strategic decisions that will optimize staff, and improve merchandising and store displays. The exact location of a customer can be found out by using localization algorithm with the help of RSSI (Received Signal Strength Indication). The system will provide an insight into customer behavior, shopping patterns and dwell time inside a building. (Tšernov, n. d. )

Scope, Introduction and Background of the Project

Scope of the Project

A positive customer experience not only satisfies customers, but also generates additional revenue, which in turn benefits retailers. Considering this factor, our project is typically focusing on retail stores for improving customer experience. Wi-Fi based retail analytics will provide visual customer location and detailed analysis of customer visitation which is important for retail stores to cope up with the competitive environment. This system will provide benefits to multiple teams simultaneously including marketing, merchandizing and IT. Other than retail stores, this system can be used to observe the visitor traffic across the industries including Automotive, Transport, Hospitality, Tourism, Property, Banking and Airports.

Introduction

A Wi-Fi tracking based retail analytics is used to localize and track customer journey data with the help of passive network sniffers. Sniffers are small Wi-Fi like devices used to capture packets of data across a network. Measured packets of data (probe requests) are used to identify the zone with the help of mac addresses where the Wi-Fi device is located. It provides information related to footfall analytics which allow retailers to make smarter decisions and manage their systems more efficiently. It’s a challenge for retailers to know about customer trends and needs. Understanding customers’ behavior through physical analytics can provide crucial insights to the business owner in terms of effectiveness of promotions, staff allocation and efficiency of services. By further looking into customer dwell (stay) time, footfall data gives intuition into customer experience and satisfaction. The field of retail analysis has gone beyond ordinary data analysis or simple counting, now the world is using data mining and data discovery to produce applicable business intelligent insights. (IEEE Computer Society, 2016)

Literature Review

Wi-Fi based localization and tracking technology can be used to identify each customer's pathway behavior in any retail store. With the swift development and widespread deployment of advanced wireless technologies now a days Wi-Fi signals are not only limited to the internet as a communication medium when propagating indoors the Wi-Fi signal will be re-modulated by human actions carrying the information of human devices. GPS is, however, a accurate solution for relatively open and outdoor environments. Prasithsangareee et al. state that GPS is not accurate for indoor localization, because satellite signals are blocked inside the building. Further, GPS signal is available less of the time in a user’s one day normally. But as we know people spend much of their time in indoor environments where GPS is not suitable. The Wi-Fi localization system is accurate as compared to GPS for indoor localization environments. An approach for indoor localization represents the system to localize the users within the network. However, the indoor locations are more complex as compared to outdoor.

In this case, there are various barriers, for example, obstacles, equipment, human beings, which lead to multipath defects. The received signal strength (RSS), Mac address, carrier signal phase of arrival (POA), and time of arrival (TOA) of the received signals are the parameters used for find location. Koyuncu and Yang developed that many solutions are available for the indoor or outdoor localization of devices that are based on triangulation, trilateration and multilateration methods using Wi-Fi signals, which provide location information. Kapadia et al. study deeply the privacy issues that cause from the localization of the devices. Following aspects of privacy must be approved the usersHow users are comfortable with their data that is to be shared?As there are any privacy concerns the users have?If they are, then how private users can be motivated by their personal information?According to Euclid Analytics, they transfer unidentified device data into customer’s useful information. Euclid analyzes the customer behavior on the basis of Wi-Fi data, shopping habits of customer, and calculates performance decisions.

Their calculations and store conversions, average store time, participation rate, bounce rate, loyalty, daily sales, conversion rate, sales transactions, average weekly sales, external opportunities, store operations Time optimization, cross-related shopping. They do not release any information related to their clients. (Vrije Universiteit Amsterdam, 2014)There are many existing projects related to location aware systems. Chon et al. developed a system that is called "Life Map" which represents a Wi-Fi based data provider and a cost-effective system used for collecting indoor location environments. Life Map uses sensors in the smartphone to provide indoor context data. Further, Rekimoto et al. created system called “LifeTag" carries a small Wi-Fi sensing device that synchronously captures surrounding Wi-Fi fingerprint data (Wi-Fi mac address and received signal strength). Later, this collected information is transferred into actual locations. The application used in our location based system is retail that analyzes the performance of a new Wi-Fi tracking system for device localization. In the literature, most projects for Wi-Fi localization are small projects that do not deal with large amounts of data. Previous products usually used sensor technology and camera for localization purposes which is so costly. So, to make our product less costly and efficient, we will use radio maps to access the locations and provide a low-budget economical product favorable for retailers in particular. On the technical side, this project proposes a strategy for analyzing the performance of the Wi-Fi tracking systems. From a business perspective, this Wi-Fi based tracking system is clearly based on the framework for indoor localization fingerprint algorithm.

Current State of the Art

Understanding the impact of marketing and operations on retail and sales is complex. The retail landscape is evolving rapidly, so retailers and store owners need a solution with trustworthy data to help them make informed decisions based on real facts that have a positive impact on sales. With Insights into customer data effective marketing an operational strategy can be developed. There are many existing systems that are performing well in the field of footfall retail analytics some of the best available market solutions are, ShopperTrak Retail Traffic Solution they are incorporating broader market benchmark, promotional, and other data sets, by which retailers will be able to convert real time data into meaningful insight to sell more and faster.

Another one is IPSOS Retail Performance providing the services of people counting and footfall solutions since 1989. It has become a leading name in retail monitoring technology. BLIX Traffic is also a real time footfall traffic analytics company for unparalleled insight into customer engagement and behavior. It has a quick and easy setup get installed in less than an hour using the existing Wi Fi network of store. BLIX Traffic collects anonymous customer data from Wi Fi enabled smartphones without depending upon any mobile application to be installed first. V Count is the another leading manufacturer of 3D people counting and customer counter, retail analytics system, visitor counter, people counter for stores. V Count is used in more than 100+ countries and installed in over 25,000 points around the world. It captures the complete customer journey from entrance to exit. These are some of the best available market solutions tracking and customer passively or actively based on MAC Address of customer’s mobile device.

Challenges

Following are the challenges faced during our project:1. In the retail store, which is the possible domain of our system, Wi-Fi sniffing stations will be set there for capturing the probe requests sent from the phone. However, the real world conditions are not that easy. Wi-Fi enabled handsets send packets randomly in the store. Thus, the proposed system will be requiring a procedure for efficient packet capturing and to improve the accuracy of RSSI. Another key challenge is the association and processing of data to get accurate results. The analysis on the big date would take a definite time for data load. The system is related to the concerns over customer’s data privacy. Thus it will be a huge task of assuring end users that their personal information is safe. Since in most of the cases Wi-Fi is 2. 4GHz based, so channel management would also be one of the challenges to be faced. This affects the Wi-Fi systems performance. (The Shortcomings Of WiFi RTLS, 2017)

Motivation and Need

Mobile phones are key to gathering footfall data. They act as just as a people counter works. Wi-Fi is hundred times more energy efficient than Bluetooth. Wi-Fi devices can communicate up to 11mbps which is 11 times faster than Bluetooth. So using Wi-Fi retail analytics is efficient this way. The motivation for doing this project was to make an attempt to solve a challenging problem and convert it into a commercial product. Our inclination towards this idea is to find ways for exploring this technical and growing area of computing. The project is necessary for following reasons: (Glaser, 2016)

Accurate people counting is considered to be a tough part. Almost all the existing systems are using different kinds of sensors or cameras and their system only works for fixed areas within the range of those sensors. In order to avoid and get rid of these limitations we plan to design a system which in not dependent on sensors. (IEEE Xplore Digital Library, 2015)Most of the existing solutions available in the market are costly so there is a need of some system which can be affordable for average retailers. In Pakistan there is no such application developed so it possesses a uniqueness in this specific area. Many existing systems are performing very well in the field of footfall retail analytics and enhancing the customer experience along with retailers’ profit but there are some limitations for these systems i. e. they are actively localizing a person which makes the system dependent on a mobile application to track people. This means that the user has to install the application first, only then he will be able to get benefit from active localization. Whereas our system is not dependent on any mobile application. Other solutions are using personal and private information of the customers just to make money without customer’s permission while our system is not breaching customer’s privacy in any way.

Objectives of the Project

Following are the objectives of our project:

People Counting: Customer visitation by hour, day and week. Unique Customers: How many of the customers are unique. Returning Customers: How many return and how often. Window Conversion: How many passersby enter store. Dwell Time: How long customers spend in a store. Optimize Staffing: Optimize shifts to align staff with peak hours. Conversion Rate: The percentage of customers that purchase. Location Profiling: Discover popular zones in store. Compare campaign effectiveness. Set benchmarks for rents. Use the system as a fire security. Compare performance of multiple zones. Real-time occupancy: Live Heat MapPeak hours location wise. Real-time Customer Tracking: Track and record the route and activities of customers inside stores progressively in real-time. This will enable retailers to understand where they are lingering, and where they are buying. Street Counting: Know how many potential visitors pass by your stores and your streettostore conversion rate. Queue Management.

Development Approach

Development Methodology

The methodology which is adopted in our project is SCRUM. SCRUM is one of the variant of agile methodology. To complete the key tasks, iterative approach is used in this methodology. This iterative approach has small iterations, which are known as SPRINTS. In simple words, the time frame in which the divided work has to be completed is called SPRINT. When one SPRINT is completed, the next one starts. In the SCRUM methodology, every team member has pure focus on the project only. When one sprint is completed, a meeting is held to discuss about the tasks offered and the changes and revisions to be made in the sprint. This process is followed throughout the project.

Risk Analysis

Risks of the Project

Following are the list of risks that have been identified as potentially affecting the project:1. The first risk associated is that hackers may set their sniffers in the respected area and they would capture the probe requests wirelessly whenever they want. (Privacy and Anonymity in the Information Society (PAIS), 2016)Although there are many benefits for customers and retailers both but privacy issues arise at the same time. Since the system accesses location data and activities of the customers.

To use the data, a customer must trust the service provider; however, due to the vast amount of data and its significance, it is unrealistic to expect a system to be able to face all the different adversarial attacks against their database systems. The proposed system is concerned about added security threats on the network. Customers are concerned about being physically tracked and about exposure of their private data. While localization solutions don’t send much data and aren’t connecting to Wi-Fi as much as capturing it, there are still security concerns about having unmanaged Wi-Fi end nodes on the network. Wi-Fi is made for data transfer, but if more access points are used solely for the purpose of localization, the higher levels of interference could cause our data system to suffer.