New business model for retailers in the AI era

Author: Zen and the Art of Computer Programming

1 Introduction

2020 marks the end of the fourth international trade war, followed by global economic recovery. Among them, applications in economic fields such as logistics and e-commerce are becoming more and more popular among consumers. However, from the perspective of the retail industry, due to the lack of objective data support and the dependence of other industries on the retail industry to earn profits, the retail industry is facing huge opportunities for change.
2019 is the first year of the development of China's retail industry. With the vigorous development of e-commerce and the rise of online platforms and APPs, offline physical stores and other Internet retail models, consumers' demand for online retail has increased by leaps and bounds. In 2019, China's total retail sales reached 4.2 trillion yuan. However, as the "epidemic" continues to spread, China's retail sector has also encountered unprecedented difficulties. Even so, judging from the growth of the overall business scale of the retail industry, the retail industry is still in a stage of rapid growth.
Under such circumstances, many retail companies have begun to explore new business models and try to use AI and machine learning to improve efficiency, reduce costs, and enhance brand awareness. These new business models have gradually formed a consensus and been recognized, such as AI O2O e-commerce, precise recommendations based on knowledge graphs, big data-driven cross-border mobile sales, IoT smart retail, etc.
This article will take the seven new business models in the retail field (AI O2O e-commerce, precise recommendations based on knowledge graphs, big data-driven cross-border mobile sales, IoT smart retail, smart warehousing, smart services, and digital transformation) as the starting point. Through the introduction of these models, it explains how the retail field can achieve a more sustainable, reliable, and secure retail ecosystem in the face of the global economic crisis and current retail industry development trends.

2. Core concepts and terminology

2.1 Blockchain

What is blockchain? Blockchain is a distributed database that stores a series of information blocks. Each block contains a record of a data writing operation, and these records are connected in series to form a chain. Blockchain can ensure that the database on each node is consistent, making the system highly secure, traceable and non-tamperable.

2.2 Smart Contract

Smart Contracts refer to computer protocols that can automatically execute contract obligations and are also a high-level programming language. It gives computers the ability to participate in smart contract execution without requiring third-party permission or concerns about trust issues. For example, smart contracts can be used to manage various business scenarios such as financial transactions, certificate records, and supply chains.

2.3 Knowledge graph

What is a knowledge graph? The knowledge graph is a graph database composed of a set of triples in the RDF (Resource Description Framework) structure. Each triple represents the relationship between a resource and its attributes. Knowledge graphs can be used for tasks such as knowledge retrieval, information extraction, open data analysis, and intelligent question answering.

2.4 Data analysis

Data analysis refers to the process of using statistics, mathematics and computer science techniques to process, refine, summarize, organize and express complex information to generate new data. Data analysis methods mainly include data acquisition, cleaning, conversion, integration, mining, presentation and other processes.

2.5 Natural language understanding

Natural language understanding (NLU) refers to the technology in which computers parse, understand and generate corresponding text output by parsing input text. Different from commands and queries in the traditional sense, NLU can understand the intentions, emotions, opinions, syntax, semantics and other factors in the text to make intelligent decisions.

2.6 Machine Learning

What is machine learning? Machine learning is a discipline that uses existing data to train computer models to complete specific tasks. It learns, summarizes, and summarizes from massive amounts of data, and finds hidden patterns, patterns, and relationships in the data.

2.7 Deep Learning

Deep learning is a type of learning algorithm. It builds a neural network and completes model parameter updates through backpropagation, thereby solving the gradient disappearance and under-fitting problems of manual neural networks.

3.AI O2O e-commerce

3.1 Mode description

"AI O2O e-commerce" is an e-commerce model based on blockchain technology and machine learning algorithms, which can help companies quickly achieve online shopping experience while reducing operating costs. This model can not only realize the user purchase process, but also improve the management efficiency of the entire mall through the integration of blockchain + artificial intelligence. "AI O2O e-commerce" is divided into modules such as intelligent product identification, e-commerce settlement system, member rights management, and mall promotion strategies.

3.2 Specific plans

3.2.1 Intelligent product identification

3.2.1.1 Product information classification

Users upload pictures or videos and intelligently identify product information in the images. Classify products through product information and prices, and determine charging standards based on the quality of the products.

3.2.1.2 Product query and recommendation

Product information is saved to the blockchain database through image recognition and OCR technology. At the same time, a graph engine is introduced to index and recommend product information to help users quickly find the products they want.

3.2.2 E-commerce settlement system

3.2.2.1 Real-time payment

Based on facial recognition, positioning, Alipay and other methods, a mobile cashier is provided, and the payment system automatically matches users and products to achieve real-time payment.

3.2.2.2 Member Rights Management

Provide users with various preferential mechanisms such as free shipping, discount coupons, and points return to achieve effective management of member rights.

3.2.3 Member rights management

Record users' consumption habits and preferences through blockchain, provide them with personalized product recommendations, and improve customer retention rates.

3.2.4 Mall promotion strategy

Develop promotion strategies to push products to customers, increase purchasing power, and improve user satisfaction.

3.2.5 Order management

All orders are tracked and managed to ensure order completeness and accuracy.

3.3 Technical implementation

• Intelligent product recognition: OpenCV, TensorFlow
• E-commerce settlement system: ZMQ, Flask, React.js
• Member rights management: MongoDB, React.js
• Mall promotion strategy: Apache Kafka, Spring Boot
• Order management: MySQL, Kafka, Zookeeper

3.4 Effect display

3.5 Applicable scenarios

3.6 Challenges and suggestions

4. Accurate recommendation based on knowledge graph

4.1 Mode description

In 2020, with the advent of the mobile Internet era, accurate recommendations in the e-commerce field have become a hot topic of focus. In the retail industry, customers usually contact retailers through search engines or phone calls to find the products they need, but in the e-commerce field, recommendation systems can use user behavior data to improve product recommendation results and increase user stickiness.
Traditional recommendation systems rely on the user's historical browsing history, search terms, preference bias and other characteristics to find relevant products through fuzzy matching. However, in the field of e-commerce, it lacks sufficient data support and cannot achieve good results. The recommendation system based on knowledge graph can integrate a large amount of user click data, purchase data, evaluation data, product description data and other information to achieve accurate recommendation of products.
The "accurate recommendation based on knowledge graph" model can help retailers generate user portraits by analyzing user purchasing behavior, product content descriptions and other data, and provide users with accurate product recommendations based on user portraits.

4.2 Specific plans

4.2.1 Data collection

Currently, there is a large amount of data that can be used to train recommendation models. However, collecting this data requires a lot of manpower, material and financial resources. In addition, because retailers have a strong awareness of personal privacy protection, users' personal information may be kept confidential. In order to better tap potential purchasing needs and preferences, more available data is needed.
Therefore, an accurate recommendation model based on knowledge graph first needs to collect massive user data. For retailers, users’ search, browsing, purchase behavior logs, product descriptions and other data can be collected directly from various platforms; for ordinary users, they can also view users’ shopping behavior, favorites, etc. through relevant websites or apps. Browsing history and other data.
Taking the e-commerce field as an example, in addition to order data, product reviews, browsing records, purchase data, etc. can also be collected for building a knowledge graph.

4.2.2 Entity discovery

1. User portrait: Analyze users’ preferences, collections, purchasing habits, occupation, place of residence and other characteristics, divide users into different groups, and build user portraits.
2. Entity discovery: Discover entities existing in user portraits through knowledge graph technology, such as products, customers, merchants, categories, etc.
3. Attribute extraction: Through rules and statistical methods, the attributes of entities, such as price, color, size, brand, etc., are discovered from users' search, browsing, and purchasing behaviors.
   Taking the e-commerce field as an example, you can discover the products that users like, frequency of purchases, frequency of reviews, the region where customers are located, product categories, on-shelf status, etc.

4.2.3 Knowledge integration

1. Correlation analysis: Through rules or statistical methods, discover the correlation between entities appearing in user portraits, such as the interaction between customers and products, the correlation between product categories, and the transaction relationship between customers and merchants. wait.
2. Inference calculation: Based on the association relationship, different recommendation algorithms are used to calculate the user's interest in the product and provide recommendations to the user. Such as recall algorithm, sorting algorithm, collaborative filtering algorithm.
   Taking the e-commerce field as an example, after a user searches for a product, the recommendation system can analyze the user's search behavior, extract the user's keywords, calculate the most relevant products through an association algorithm, and then display them to the user.

4.2.4 Effect display

Taking Taobao as an example, the product details page can display the user's purchase history, comments and other data, match the recommended products with the user's purchasing behavior, and recommend products that may be of interest to the user.

4.3 Applicable scenarios

4.4 Challenges and suggestions

5 Big data-driven cross-border mobile sales

5.1 Mode description

2020 is a new century, and the global digital economy is sweeping the world at an unprecedented speed. With the continuous advancement of technology, the development of cross-border e-commerce is accelerating, and more and more cross-border companies are beginning to lay out their own business lines. However, today, cross-border e-commerce is still in its infancy. The operating costs and service levels of some emerging cross-border e-commerce platforms are too high, resulting in the operating costs and service levels of cross-border e-commerce platforms being unable to meet the needs of consumers. need.
The "big data-driven cross-border mobile sales" model can improve the operational efficiency of cross-border e-commerce platforms, reduce costs, improve user experience, and expand the supply chain system of the retail industry. The "big data-driven cross-border mobile sales" model provides a complete set of management tools to help cross-border e-commerce platforms solve business pain points, promote continuous innovation and upgrading of cross-border e-commerce platforms, and better serve consumers.

5.2 Specific plans

1. Convergence of core data sources: Collect and organize massive data from multiple data sources (including payment systems, supply chains, logistics, user data, product data, etc.) to form core data of the cross-border e-commerce platform.
2. Construction of cross-border network: Establish and maintain massive cross-border network connections to realize data exchange and communication between cross-border e-commerce platforms, within the platform and among suppliers.
3. Logistics and supply chain planning: Plan the logistics and supply chain of cross-border e-commerce platforms to ensure smooth, timely and accurate circulation of goods to consumers.
4. Application of big data analysis: Analyze and mine massive amounts of data to improve the management level and service capabilities of cross-border e-commerce platforms and bring new business value to the platform.
   Cross-border e-commerce platforms represented by Tmall and JD.com will provide relevant product consultation, price comparison, payment methods, logistics and distribution and other services on the platform, allowing consumers to enjoy a high-quality shopping experience.

5.3 Technical implementation

1. Data warehouse construction: Build a specialized data warehouse and store core data sources in the data warehouse.
2. Distributed computing framework: Build a computing cluster based on Spark, Flink and other frameworks to achieve batch data synchronization between data sources.
3. Data collection framework: Build a collection cluster based on Flume, Scribe and other frameworks to realize real-time data collection from data sources.
4. Data processing platform construction: Build a unified analysis platform to clean, transform, associate, aggregate, mine, and analyze real-time and offline data.
5. Design of service-oriented framework: Build service clusters based on Mesos, Kubernetes and other frameworks to achieve high-speed data processing and distributed computing of massive data.
6. Design of front-end display platform: Build a display cluster based on front-end technology stacks such as React and Angular, and provide a beautiful user interface to facilitate user access and operation.
   Taking Tmall as an example, Tmall will provide users with multiple functional pages such as "Shopping Cart", "My", "Order", and "Logistics". Users can manage shopping information, order information, address information, Delivery address, etc., thereby realizing the entire process of retail shopping.

5.4 Effect display

5.5 Applicable scenarios

5.6 Challenges and suggestions

6. IoT smart retail

6.1 Mode description

"IoT smart retail" is a new retail industry model. It uses intelligent terminal equipment and cloud computing technology to closely integrate Internet technology, IT technology with the physical world, and combines artificial intelligence, biometrics, machine learning, sensor networks, cloud computing, database and other technologies to achieve the goal through "big data" and "things". "Network" to realize the intelligence, refinement and efficiency of retail enterprises, and enhance the competitiveness, profitability and healthy development capabilities of the retail industry.
The "IoT smart retail" model can comprehensively process the retailer's internal and external data, as well as real-world data. Through network processing, it can improve the agility and accuracy of retail business, improve customer experience, and realize the competitive advantage of retailers. and customer satisfaction.

6.2 Specific plans

6.2.1 Physical layer:

1. Intelligent identification of items: Use IoT devices such as drones and barcode scanners to scan products and collect product information data to the IoT cloud.
2. Product correlation modeling: Model the similarity and correlation of commodities, establish a commodity correlation network, realize the integration of commodity data, and provide a semantic model for commodity data.
3. Personalized recommendation: Provide users with personalized product recommendations through item recommendations based on user preferences.
4. Item circulation route planning: Based on the item association network, logistics routes are optimized to improve commodity circulation efficiency.
   E-commerce platforms represented by Tmall upload the collected user data and item data to the IoT cloud at the physical level, and use drones, scanning guns and other IoT devices to realize product scanning, barcode generation, Item identification, product correlation modeling, logistics route planning and other functions.

6.2.2 Application layer:

1. Provide retail services: Provide diversified retail services through the Internet and IoT technology, such as "Taobao Customer Service", "Tmall Supermarket", "Double Eleven", "Points Exchange", etc.
2. Product recommendation: Combining user purchase, like, and browsing history with IoT data to provide users with more personalized product recommendations.
3. Logistics and distribution: Combined with IoT data, it provides users with real-time logistics information to achieve fast and accurate delivery of goods.
   At the application level, e-commerce platforms represented by Tmall provide retail services through the Internet and IoT technology, such as "Tmall flagship store", "movie ticketing", "train tickets, special sales", etc., to enhance the marketing of stores ability.

6.2.3 Data layer:

1. Data center construction: Build a data center, integrate various data sources, and ensure data security.
2. Data transmission encryption: Encrypt data transmission to prevent data leakage.
3. Data computing analysis: Use cloud computing and big data analysis technology to conduct big data analysis on data to improve data accuracy.
   At the data level, e-commerce platforms represented by Tmall build data centers to integrate various data sources to achieve data security, processing and analysis, improve data accuracy, and enhance data value service capabilities.

6.3 Technical implementation

1. IoT system development: Use open source hardware development boards such as Arduino and Raspberry Pi to connect microcontrollers, sensors and other hardware to realize the development of IoT systems.
2. IoT cloud platform construction: Establish an IoT cloud platform to realize cloud data collection, storage, calculation, analysis and other functions.
3. Cloud service development: Use cloud services to provide consumers with various retail services, such as "micro mall", "wool shopping", etc.
4. Message push technology: Use message push technologies such as SMS, WeChat, Weibo, and email to realize functions such as merchant notifications and user message subscriptions.
   E-commerce platforms represented by Tmall use Internet of Things technology to build an Internet of Things cloud platform to realize functions such as product scanning, item identification, product recommendation, and logistics distribution to enhance customer experience.

6.4 Effect display

6.5 Applicable scenarios

6.6 Challenges and suggestions

7. Smart warehousing

7.1 Mode description

In 2020, new technologies such as smart logistics, robotics, and drones have entered the retail field, and smart warehousing has also begun to enter a stage of rapid development. The "smart warehousing" model can realize the intelligent and precise allocation of logistics and warehousing resources of enterprises, improve operational efficiency, and enhance the competitiveness of the retail industry.
The "smart warehousing" model can combine cloud computing, big data, artificial intelligence and other technologies to maximize efficiency, reduce costs and save costs through automation, intelligence, precision and other methods from traditional material shipping methods.

7.2 Specific plans

7.2.1 Product inventory management

1. Commodity inventory forecast: Combined with big data technology, predict the inventory quantity of commodities.
2. Product shelf life management: Combined with quality assurance management, improve the shelf life of products.
3. Product safety management: Use Internet of Things technology to monitor the safety status of goods.
   Taking an international airline as an example, by predicting the inventory quantity of goods, improving the shelf life, and using Internet of Things technology to monitor the safety status of goods, conduct early stage management and control of production, improve production efficiency, and reduce work costs.

7.2.2 Supply chain transportation scheduling

1. Supply chain transportation planning based on graphical analysis: Combined with big data technology, it provides decision-making support for warehouses, procurement, logistics and other links by drawing supply chain roadmaps.
2. Automatically allocated supply chain transportation guidance: Combined with cloud computing technology, automatic transportation guidance for warehouses is realized.
3. Automatic sorting system for commodity tracking: Combining drones and robotics technology to realize automatic sorting of commodities.
   Taking an international airline as an example, it combines big data technology to draw supply chain roadmaps to provide decision-making support for warehouses, procurement, logistics and other links to improve operational efficiency; it also combines cloud computing technology to realize automated transportation guidance for warehouses and improve operations. The operational efficiency of the warehouse; combined with drones and robotics technology, can realize automatic sorting of goods and improve transportation efficiency.

7.2.3 Warehouse operation management

1. Warehouse management system: Build a warehouse management system to manage warehouse products.
2. Operator management: Divide warehouse management personnel into full-time, temporary, emergency and other levels to manage authority and enhance security.
3. Warehousing operation and maintenance automation: Combining cloud computing, big data, and robotics technology to realize automated operation and maintenance of warehousing materials.
   Taking an international airline as an example, a warehouse management system was built to manage warehousing products; warehouse management personnel were divided into full-time, temporary, and emergency levels to conduct authority management and enhance security; it combined cloud computing, big data, and robots technology to realize automated operation and maintenance of warehousing materials and improve the efficiency of warehousing management.

7.3 Technical implementation

1. Supply chain management system development: Use common supply chain management software to manage all aspects of the supply chain.
2. IoT device development: Use IoT-related open source software packages and SDKs to realize the development of IoT devices.
3. Development of commodity warehousing management system: Use software engineering methods and follow the product life cycle management process to realize the development of commodity warehousing management system.
4. Cloud platform construction: Build a cloud platform based on cloud computing platform to realize supply chain management, logistics, warehousing management and other services.
   Taking an international airline as an example, it uses general supply chain management software to manage all aspects of the supply chain; it uses open source software packages and SDKs related to the Internet of Things to realize the development of Internet of Things equipment; in accordance with the product life cycle management process, Realize the development of commodity warehousing management system; build a cloud platform based on cloud computing platform to realize the construction of cloud platform for supply chain management, logistics, warehousing management and other businesses.

7.4 Effect display

7.5 Applicable scenarios

7.6 Challenges and suggestions

8. Smart service

8.1 Mode description

In 2020, consumer demands are changing rapidly, and people have higher requirements for life. The development of artificial intelligence technology has made intelligent services an important market today. The "smart service" model can help retailers better serve consumers.
The "smart service" model can use artificial intelligence technology to improve user experience, improve service quality, optimize service methods, improve customer satisfaction, and gain good competitiveness.

8.2 Specific plans

8.2.1 Natural language understanding

1. Development of chatbots: Combining deep learning and NLP technology to develop chatbots that suit user tastes.
2. FAQ question and answer robot: combines big data and NLP technology to provide users with answers to frequently asked questions.
3. Message push: Use cloud computing, machine learning, and AI algorithms to improve user retention.
   Taking a well-known consulting company as an example, it develops a chatbot that suits users' tastes; combines big data and NLP technology to provide users with answers to frequently asked questions; and uses cloud computing, machine learning, and AI algorithms to improve user retention rates.

8.2.2 Entity recommendation

1. Product recommendation: Combining cloud computing, big data, recommendation systems and other technologies to recommend related products to users.
2. Vertical product recommendation: Combined with artificial intelligence technology, the user's preferences, wishes, preferences and other information are considered when recommending products.
3. Customer experience mining: Combining data mining, image processing, recommendation systems and other technologies to mine users' consumption habits and interests to provide users with better services.
   Taking a supermarket as an example, it combines technologies such as cloud computing, big data, and recommendation systems to recommend relevant products to users; it combines artificial intelligence technology to consider the user's preferences, wishes, preferences and other information when recommending products; it combines data mining and image processing. , recommendation systems and other technologies to explore users’ consumption habits and interests and provide users with better services.

8.2.3 Potential customer analysis

1. Establishment of user portraits: Combining big data and NLP technology to establish user portraits.
2. Potential customer identification: Combine artificial intelligence, neural network, and deep learning technology to identify potential customers.
3. Customer psychology analysis: combine big data, statistical analysis, and NLP technology to analyze customer psychology and improve services.
   Taking a B&B hotel as an example, it combines big data and NLP technology to establish user portraits; it combines artificial intelligence, neural networks, and deep learning technology to identify potential customers; it combines big data, statistical analysis, and NLP technology to analyze customer psychology and improve services.

8.3 Technical implementation

1. Development of intelligent question and answer system: using natural language understanding technology to develop a conversational question and answer robot.
2. Intelligent search engine development: Use search engine technology to develop an intelligent search engine.
3. Business system construction: Use object-oriented technology to build business systems and improve business efficiency.
   Taking an e-commerce company as an example, it used natural language understanding technology to develop a conversational question-and-answer robot; it used search engine technology to develop an intelligent search engine; it used object-oriented technology to build business systems and improve business efficiency.

8.4 Effect display

8.5 Applicable scenarios

8.6 Challenges and suggestions

9. Digital Transformation

9.1 Mode description

In 2020, digital transformation is becoming an important topic in the retail field. In this process of digital transformation, the main role of the retail industry has changed from consumers to merchants. The "digital transformation" model can help retailers carry out digital transformation and build their own brands.

9.2 Specific plans

9.2.1 Brand development

1. Brand positioning: Choose a brand positioning with market development potential and develop unique products and services.
2. Brand value statement: Make a brand value statement to clarify the brand values ​​and convey the brand concept.
3. Brand image building: Create brand image, shape brand style and create atmosphere.
   Taking a certain enterprise as an example, it chooses a brand positioning with market development potential and develops unique products and services; makes a brand value statement, clarifies brand values, and conveys brand concepts; creates a brand image, shapes the brand style, and creates an atmosphere.

9.2.2 Marketing strategy

1. Marketing plan: Develop a marketing plan suitable for retailers to increase retail revenue.
2. Brand marketing activities: attract user attention and enhance brand awareness through various marketing activities.
3. Online and offline marketing linkage: Through the combination of online and offline, all-round marketing in the retail industry is achieved.
   Taking a catering group as an example, it formulates a marketing plan suitable for retailers to increase the revenue of the retail industry; through various marketing activities, it attracts user attention and enhances brand awareness; through the combination of online and offline, it achieves all-round marketing in the retail industry.

9.2.3 Full life cycle management

1. Brand Management: Responsible for brand operations and marketing to enhance brand influence.
2. Product management: Responsible for the pricing, promotion, sales, etc. of products to provide consumers with high-quality services.
3. Service management: Responsible for ensuring service quality and providing high-quality services to consumers.
   Take an e-commerce company as an example. It is responsible for brand operation and marketing to enhance brand influence; it is responsible for the pricing, promotion, and sales of goods to provide consumers with high-quality services; it is responsible for ensuring service quality and providing consumers with high-quality services. Quality service.

9.3 Technical implementation

1. Information system construction: Use ERP, CRM and other information systems to realize the integration, management and analysis of retailer information.
2. Operation platform construction: Use cloud computing, big data and other technologies to build a retail operation platform.
3. Channel marketing construction: combine online and offline marketing linkage to achieve all-round marketing in the retail industry.
   Taking a home appliance retailer as an example, it uses ERP, CRM and other information systems to realize the integration, management and analysis of retailer information; uses cloud computing, big data and other technologies to build a retail operation platform; combines online and offline marketing linkage to realize retail sales All-round marketing for the industry.

9.4 Effect display

9.5 Applicable scenarios

9.6 Challenges and suggestions