The abundance of data in the form of customer transactions, customer search history, customer interactions and the computing power of modern computing systems provide a lot of new opportunities for data analytics in retail. The process of tracking customer data, forecasting of sales, building a pricing model for products and personalization of the website tailoring to the interest of the customer, all require data analysis. The data is obtained from various sources such as social media interactions, search history on the website, purchase history and even from third-party aggregators. Analyzing and processing this data provide new opportunities to any retail industry to achieve a competitive advantage over others. Personalized and customer-driven marketing has become very important in the retail industry. Amazon being an early adopter of this technology, is a prime example of how analytics has redefined the shopping experience in the retail industry. Implementing Machine learning, data collection using the Internet of Things and other technological advances in retail data can improve sales and attract new customers.
Keywords: Retail, Customer-Driven, Machine Learning, Internet of Things
Data acquisition using the Internet of Things (IoT) in the fashion industry (Chan, Lau & Fan, 2018) been one of the most important innovations in increasing the amount of relevant data that can be collected during a customer’s visit. Analyzing the behavior of a customer during an in-store purchase using sensing devices can reveal a customer’s interests and their choices. Using a number of sensing devices, the in-store customer behavior is captured, pre-processed and then transmitted over a wireless network to the cloud. A data analytics model built with fuzzy logic is developed to generate the data of a customer’s purchasing intentions. This approach will help retail stores to recommend products to customers and guide supply chain planning. Such an approach allows retailers to gain more sales thereby increasing their market value.
If you need assistance with writing your essay, our professional essay writing service is here to help!Find out more
The importance and usage of Business intelligence technologies in the retail industry have been discussed by Gang, Kai & Bei (2008). The increase in the amount of data available due to the advent of automation, new technologies and standards have made the decision-making process in business has become complicated. The key technologies used in business intelligence are Data Warehouse, Online Analytical Processing (OLAP), Data Mining and Release & Express technology. The main applications of a BI system are profit analysis and KPI Management, client service management and environmental analysis.
The traditional marketing strategies are data-driven and include Business analytics to improve customer relevancy and efficiency. Data-driven segmentation of customer behavior in the retail industry has been explained by Carmichael, Chen & Luo (2018). Customer segmentation has become an important part of marketing analytics because it allows the customers to be grouped based on their purchase behaviors, segment demographics, and behavioral evolvement. This segmentation is used to create tailored marketing campaigns based on the target customers to have an idea of the effectiveness of a campaign for each segment. Gathering enough data to analyze it for each segment has been the main limitation here.
Walmart, one of the biggest e-commerce stores, has a large amount of data to analyze. Using Big data analytics (Singh, Ghutla, Lilo Jnr, Mohammed & Rashid, 2017), analyzing the previous year’s data to predict
the next year’s sales are important for all retail companies. To predict sales, it is necessary to understand their business trends, the customer needs and manages the resources wisely. Big Data is taking over the traditional method of managing and analyzing data. It is important to take into consideration seasonality trends, randomness, and future forecasts in order to efficiently device ways to minimize the sale drop and maximize the profit to remain in the competition, it is important.
Recommendation systems using recommender algorithms (Chavan & Mukhopadhyay, 2017) play an important role in providing a personalized experience for the customer. Such systems use the past purchase history and the customer’s search data to provide relevant recommendations for the customer. All big companies such as Amazon, Netflix personalize the content for the user based on their shopping habits and behavior patterns. An effective recommendation system can increase sales manifold, by presenting users with items that would most likely need, before the user even recognizes they need it. The hybrid recommendation algorithms are more suitable in the e-commerce field. It improves the quality and efficiency by providing the user with a great shopping experience.
Data Analytics in Retail Stores
Data analytics has taken the traditional brick and mortar stores by storm, completely revamping their entire business model. It has brought in a different perspective to gauge customer needs and to optimize supply chain management and maximize profits. Moreover, it aids in maximizing profits by optimizing product placement, providing discounts and ensuring that loss due to unsold goods is minimized. Data analytics also helps in analyzing and understanding the sales pattern of each store, thereby learning the buying habits of people who frequent that store. This behavioral analysis will help businesses stock the store with products that sell fast, and also push similar products. Moreover, businesses can retain the customers by providing them with coupons or discount programs. A lot of companies now run membership programs, wherein a customer’s purchases are all linked to the same account, regardless of where the purchase was made – instore or online. This enables businesses to understand each customer well, and target sales appropriately.
Use case scenario of IoT in Retail
An excellent example of IoT in retail is Amazon Go. It is a first of its kind, experimental retail store, where there are no checkout counters in the store. The idea behind the store is that people should be able to walk in, grab what they need, and walk out without needing to wait to check out their items. The store has been automated completely. When a person enters the store, they need to check in with their Amazon Go app, by scanning a QR code. Once they do that, the system recognizes the person and ties them to an Amazon account. As soon as the customer exits the store, the products they purchased are identified, and the bill gets posted to their account. The store has strategically placed cameras and LIDAR sensor arrays to track every movement of a customer inside the store. These highly accurate sensors are trained to detect specific movements, the weight of the items that the customer picks, and they even know if the customer put it back on the shelf or not. The cameras use deep learning technology which allows the machine to see and identify the object that they are pointing to. These machines use advanced pattern recognition techniques to detect both the item and the person who took it, and track if the item was returned to the shelf. As soon as a person picks an item, the sensors detect the item and add it to the person’s cart. The weight of the item is also added to the item’s information. If the person returns the item to the shelf, the cart is updated accordingly. Using all these cameras and sensors, they are able to track their customers at the store and do auto billing of these items as soon as they walk out.
In order to achieve this level of automation, the deep learning engine behind the system will have to be trained with exhaustive amounts of data. Amazon employed a large number of human employees to train the algorithms and verify whether they identify the products correctly. The tremendous amount of data collected by these sensors are analyzed by Amazon’s data centers to continuously improve the performance, detecting accuracy and speed of the system. Amazon Go uses sensor fusionto combine data from all the sensors, including the weight sensors on the shelves to track individual products. The combination of sensors used makes up for an interesting mix of data available to the system, to “learn” the person and the product. This data is also utilized to periodically improve the algorithms that perform the learning.
Sensors, devices, and technologies used in Amazon Go
Light Detection and Ranging (LIDAR) is a remote sensing method that was initially used to examine the surface of the earth. With the advent of self-driving cars, LIDAR has gone mainstream, with these cars using LIDAR extensively for object detection, contour recognition, slope detection etc. LIDARs are used in Amazon Go store to detect objects and the spatial position of these objects. LIDAR bounces off light from objects and analyzes the reflected light to generate a three-dimensional image of the shape and characteristic of that particular object. LIDAR is capable of accurately identify objects along with their color and texture, and even the kind of material the object is made of. LIDAR combined with neural networks can detect almost every object with a high degree of accuracy. LIDAR is implemented in a variety of ways based on the application. The simplest implementation of LIDAR involves using solid-state lasers that have no moving parts. These lasers shine a focused beam of light on the target surface and analyze the reflected light to determine the characteristics of the target surface. Moreover, LIDARs can also be used to identify the spatial position of the object in a 3D space, and if the system knows the position of each and every object in the store, it is then trivial to recognize the object a customer picked up, with just LIDAR.
RFID is a mechanism by which electromagnetic fields are used to identify and track tags that are attached to objects. The tags contain electronically-stored information. In an Amazon Go store, all objects are tagged with RFID tags, containing information about that particular object. The RFID tags are used to expedite checkout, and also as a theft prevention mechanism. Since RFIDs can be read by a simple electromagnetic scan, they are widely used by a lot of other companies to provide a number of services. For instance, Neiman Marcus has “smart fitting rooms” which use RFID tags attached to clothes to allow customers to virtually change the color, texture of the clothes they try, to see which one fits them better. It also allows customers to virtually add accessories to their outfits to see how they look.
Computer Vision is a technique which is used to automatically identify objects from a video or an image. Machine learning is employed to train a model to identify the features in the images. This needs a lot of data and used to be traditionally dominated by supervised learning. With advances in deep learning methodologies, newer models are being developed which are used to extract and identify images from an unlabeled dataset, enabling unsupervised machine learning to happen. The models are usually trained with a multilayer convolutional neural network, by feeding it large swathes of data. The frames of a video are passed into the layers of neurons. Each layer of the neural network and the neurons in them are assigned with a filter that checks for unique features in the images. Each filter is moved over the image and to obtain a confidence value. Likewise, all these features with numerical values are trained, and an overall feature matrix for a particular object is “learned”. As the data available to the model increases, the accuracy of the model also increases. The model also uses the data it records to continuously refine and tune itself. Computer vision is used to identify both the objects, as well the humans who enter the store.
The process of combining data from one more sensor is called sensor fusion. Sensor fusion is often employed to combine or “fuse” a number of measurements from individual sensors, into one holistic measurement for the object. The data from sensors such as weight, image, position and other information is then used to accurately identify the object. Amazon Go store uses sensor fusion to fuse data from the cameras, LIDAR, weight sensors, RFID, and other sensors located throughout the store to accurately pinpoint the person, and the object they pick up. Weight sensors are used to identify the weight of the objects, and spatial recognition is used to detect if the object is in the customer’s cart, or if it is replaced. All these data are combined together to add or remove an item from the virtual cart of the customer. The idea behind the technique is simple, however, the implementation of such a system at scale is not. The fusion process has to be precise and as accurate as possible, in order to minimize losses due to misclassification. Moreover, these systems have very low latency constraints and need to be as fast as possible to enable near real-time updates of a customer’s virtual cart.
Personalizing online shopping based on customer behavior in-store
In today’s hyper-competitive world, a business has to put in a lot of efforts in order to retain customers. Businesses strive to customize the experience to suit each and every customer. Businesses that have both an online presence and a physical store gain a significant advantage, as they get to collect data from customers when they are in-store. There are a number of points at which customer interaction with products can be captured in a store. However, most businesses stick to the one at the checkout counter, where an option is provided to the customers to link all their purchases to a business specific identity. For instance, GAP sells custom credit cards, that it can use to track people’s purchases across GAP and its sister brands. It offers additional benefits to customers who own such cards, to entice customers to buy them. Most people would like to shop at the comfort of their home and get all they need in a single click. Integrating a customer’s in-store preferences with their e-commerce shopping experience and behavior will help retailers personalize the products for them. It also helps businesses to recommend products to customers based on analytics. The in-store shopping experience also is made interactive by asking the customer to connect their personal devices to the in-store network, and then recommending what the customer can buy when they are at the store.
Our academic experts are ready and waiting to assist with any writing project you may have. From simple essay plans, through to full dissertations, you can guarantee we have a service perfectly matched to your needs.View our services
The in-store experience can be further expanded by adding intelligent accessories to the stores, such as smart mirrors in fitting rooms, a in-store digital assistant that helps the customer in shopping etc. Similar to what Amazon Go uses, RFID, Augmented reality, UPC scanning and other smart technologies can be used to create an interactive experience for the customer.
Role of analytics in e-commerce
Data analytics is used heavily in e-commerce to personalize the shopping experience of users. The success of e-commerce depends heavily on returning users. Given a large number of companies that engage in e-commerce, the competition is heavy, and every company has to do as much as it can do to retain customers.
Users will return to a site if and only if the e-commerce store offers the customer more value than they expected. The system needs to understand what a user is looking for, in order for the system to make relevant recommendations. Hence, businesses collect a lot of data from the user when they use the store and use that data to suggest products to the users. Almost all interactions of the user, such as clicks on a product, searches for a product, adding a product to the cart, purchasing a product etc. are logged and the resulting data is mined to understand the users’ preferences. The metrics being tracked also includes the amount of time a user spends on a particular product page, which will help businesses understand what the user is looking for, in a much-nuanced way. It is essential to filter out the noise in these data points in order to build an effective recommendation system. The system can be made as a “learning” system, by rewarding it if the user purchases, or adds a recommended item to a cart, and punishing it if the user clicks on “not relevant”. By doing so, businesses can fine-tune their recommendation systems, and understand their users better.
Analytics plays a major role in supply chain management, recommender systems, marketing, and product development. An E-commerce system provides a virtual environment to buy products, and hence most marketing is in the virtual environment. The online marketing teams work on placing ads for relevant keywords on Google or other websites. The collected data from the users is used to identify the keywords to target, in order to maximize visits to the store. They analyze the funnel of new prospect customers and maximize the likelihood of a customer clicking on an ad.
When a customer visits the website or a mobile application of a retailer, relevant information such as, the customer’s location, what they search for, how they landed on the website, what they bought, the price of the item and the quantity. Since all of his information is stored, it is easier to analyze the behavior pattern of the customer on the website and provide personalized recommendations. If a customer spends a lot of time on a product’s page, it could either mean that the customer is interested in the product, or they aren’t actively looking at their device at the moment. The algorithms need to be able to differentiate between the two scenarios in order to provide effective recommendations. There are multiple ways to tune the algorithms, of which some methods ask the user to provide feedback on the quality of recommendations. Care has to be taken to keep this feedback collection mechanism to be minimally intrusive so that it does not disrupt a user’s usual shopping experience.
To provide better recommendations, visitor data is segmented into new visitors and existing customers. Since the companies know the history of user searches and purchases, they are able to provide automated recommendations in real-time. It is also helpful to “activate” users who have been inactive for a long period, by providing them with exclusive deals and offers. An excellent example of maximizing sales by providing rewards is the Starbucks mobile app, where daily offers are sent as push notifications. Moreover, Starbucks has a rewards system where the user is rewarded with “stars” for every dollar spent. The users can use these “Stars” to get discounts on their next purchases. A lot of companies display customized offers and product recommendations to each customer to enhance the overall user experience. The price of the products displayed is often adjusted in real-time based on competitor’s product prices, demand, and marketing scenario. Based on the historical data on how the company performed, it is possible to forecast the sales and demand for future years. Having this intelligence is a huge plus, as the companies can now stock their warehouses based on projected demand. Amazon uses this data in an interesting way. When new products such as phones launch on Amazon, they prepare their delivery warehouses in advance, dispatching enough units to each warehouse based on expected demand. When the sale starts, and customers purchase, Amazon then handles the last mile delivery in a shorter span of time, thereby maximizing customer satisfaction.
The customer’s journey on the website is often represented in the form of a funnel. This hypothetical funnel has a lot of entry points where users enter the process. These entry points include – Visiting the website directly, landing on the website through a search, clicking on an ad, visiting from a social media post etc. Once the user enters the funnel, there are a series of well-defined actions that the user can take, and the end result is the user checking out or purchasing a product. Funnel analysis involves analyzing the series of events (a well-defined flow), for example, the checkout process, that leads to a goal. During each step in the funnel, there’s a possibility of losing a number of customers. The funnel analysis process is used to calculate the conversion rate of a customer, that is, how many people successfully go to the checkout page and buy the product. By employing all the above-mentioned techniques, e-commerce companies can stay a level above the competition and deliver a good, satisfactory user experience.
Data analytics is a field that has just started to grow. The amount of innovation that can be achieved in this space is just monumental, and the current products and services are barely scratching the surface. The advent of the Internet of Things has made innovations like Amazon Go possible. It is only a matter of time before other retailers follow suit, making similar, if not better fully automatic stores. As the number of users increases, the data available to the businesses also increase, thereby enabling the businesses to further refine their offerings. Businesses will be capable of predicting when the user will need a particular product based on their purchasing habits. They can also offer to auto-deliver such products if the user consents to it. This will take the current subscription-based delivery model to a predictive and adaptive need-based delivery model.
Retailers get feedback from the users either directly, or indirectly by sending surveys, scraping Facebook comments, likes, and reviews. This is a very inexact science as there’s no proper way to connect user profiles to the internal databases of the company. Moreover, not all customers leave feedback. So, the satisfaction quotient of a product cannot be reliably estimated. In the future, businesses can use sensors and cameras inside the physical store to determine the user’s reactions and use visual recordings to determine their mood. As a result, businesses can use this data to understand the users who visit their retail stores. Big data also plays an important role in supply chain logistics and management. Using data analysis, it is possible to identify which product is popular in which region and identify the best route to deliver the product.
- Bradlow, E. T., Gangwar, M., Kopalle, P., & Voleti, S. (2017, 03). The Role of Big Data and Predictive Analytics in Retailing. Journal of Retailing, 93(1), 79-95. doi: 10.1016/j.jretai.2016.12.004
- Carmichael, G., Chen, Y., & Luo, C. (2018, 05). Data-driven segmentation of consumers’ purchase behaviour in the retail industry. 2018 4th International Conference on Information Management (ICIM). doi:10.1109/infoman.2018.8392838
- Caro, F., & Sadr, R. (2018, 10). Internet of Things (IoT) in retail: Bridging supply and demand. Business Horizons. doi: 10.1016/j.bushor.2018.08.002
- Chan, C. O., Lau, H. C., & Fan, Y. (2018, 05). IoT data acquisition in fashion retail application: Fuzzy logic approach. 2018 International Conference on Artificial Intelligence and Big Data (ICAIBD). doi:10.1109/icaibd.2018.8396166
- Chavan, R., & Mukhopadhyay, D. (2017, 02). A comparative study of recommendation algorithms in e-commerce. 2017 International Conference on I-SMAC (IoT in Social, Mobile, Analytics and Cloud) (I-SMAC). doi:10.1109/i-smac.2017.8058331
- Gang, T., Kai, C., & Bei, S. (2008, 09). The research & application of Business Intelligence system in retail industry. 2008 IEEE International Conference on Automation and Logistics. doi:10.1109/ical.2008.4636125
- Gayathri, B. (2017, 02). Efficient market basket analysis based on FP-Bonsai. 2017 International Conference on I-SMAC (IoT in Social, Mobile, Analytics and Cloud) (I-SMAC). doi:10.1109/i-smac.2017.8058287
- IoT Based Retail Shopping System Using NFC. (2018, 03). International Journal of Modern Trends in Engineering & Research, 5(2), 159-163. doi: 10.21884/ijmter.2018.5053.bup6w
- Li, Y. (2015, 12). Adaptability Analysis of System for Retail Chain Enterprise Management. 2015 International Conference on Intelligent Transportation, Big Data and Smart City. doi:10.1109/icitbs.2015.22
- Mavroudis, V., & Veale, M. (2018). Eavesdropping Whilst You’re Shopping: Balancing Personalisation and Privacy in Connected Retail Spaces. Living in the Internet of Things: Cybersecurity of the IoT – 2018. doi:10.1049/cp.2018.0018
- Patil, A. P., Deepshika, M. P., Mittal, S., Shetty, S., Hiremath, S. S., & Patil, Y. E. (2017, 08). Customer churn prediction for retail business. 2017 International Conference on Energy, Communication, Data Analytics and Soft Computing (ICECDS). doi:10.1109/icecds.2017.8389557
- Seetharaman, A., Niranjan, I., Tandon, V., & Saravanan, A. S. (2016). Impact Of Big Data On The Retail Industry. Corporate Ownership and Control, 14(1). doi:10.22495/cocv14i1c3p11
- Singh, M., Ghutla, B., Jnr, R. L., Mohammed, A. F., & Rashid, M. A. (2017, 12). Walmart’s Sales Data Analysis – A Big Data Analytics Perspective. 2017 4th Asia-Pacific World Congress on Computer Science and Engineering (APWC on CSE). doi:10.1109/apwconcse.2017.00028
- Wu, C., Zeng, Y., & Shih, M. (2015, 06). Enhancing retailer marketing with an facial recognition integrated recommender system. 2015 IEEE International Conference on Consumer Electronics – Taiwan. doi:10.1109/icce-tw.2015.7216881
Cite This Work
To export a reference to this article please select a referencing stye below:
Related ServicesView all
DMCA / Removal Request
If you are the original writer of this essay and no longer wish to have your work published on UKEssays.com then please: