A Thousand Word Long Article | Understanding the General Knowledge of "AI+Cross border E-commerce" in One Article

Product Notes of You Yao

follow

2024-09-09

5 comments

3816 views

26 collections

23 minutes

  Two major challenges for B2B product managers: how to analyze and design products from multiple perspectives such as market, user, and business? How to effectively manage and promote project implementation?

This article will delve into the various roles in the cross-border e-commerce industry chain, analyzing how AI empowers these roles to improve efficiency, optimize experiences, and increase profits. From promotion, sales operations to infrastructure construction, the application of AI is gradually penetrating and reshaping every aspect of cross-border e-commerce. Through specific cases and product analysis, we will reveal the current situation, challenges, and future trends of the "cross-border e-commerce+AI" model.

According to iiMedia Research, the scale of China's cross-border e-commerce exports in 2022 is expected to reach 1.53 trillion yuan, a year-on-year increase of 10.1%. By 2024, the scale of China's cross-border e-commerce exports is expected to reach 2.95 trillion yuan, showing a significant growth trend.

1、 What are the roles in the cross-border e-commerce industry chain?

1. Platform side

According to the operation mode, China's cross-border export e-commerce can be divided into third-party open platforms and self operated platforms, with the classification standard being that self operated platforms have their own products and cargo rights.

Should product managers choose to work on the B-end?

In recent years, the Internet has experienced the upsurge of spending money on the market, and more and more realize that it is not enough to rely solely on the power of the C end. It needs to make up for the shortcomings of the B end as soon as possible. Should product managers choose to transition to the B-end?

View details>

According to the different target users, it can be divided into B2B and B2C. The following is the main platform information:

1) B2B third-party open platform

2) B2C third-party open platform

3) B2B self operated platform

4) B2C self operated platform

2. Brand merchant/manufacturer

From going global in the form of integrated industry and trade to going global in the form of brand power, various brand merchants continue to strengthen their digital capabilities, accelerate the integration of supply chains, and enhance brand influence. Cross border e-commerce brands/manufacturers will choose to join a combination of third-party open platforms and self built independent websites. The following is information on the main brand/manufacturer:

3. Service provider

According to the management chain of cross-border e-commerce sellers, cross-border e-commerce SaaS can be divided into three major scenarios: marketing, ERP, and supply chain.

1) Marketing service providers

According to product functions, cross-border marketing SaaS can be divided into four categories: website building, product selection, operation, and customer acquisition, covering various marketing needs of cross-border sellers from "0" to "1". Due to the interconnection of functions and data between different scenarios, manufacturers are also building product matrices towards one-stop marketing SaaS. Among them, the integration between operation and customer acquisition vendors is high, while the connectivity of website building vendors is strong. The following are the main marketing service providers:

2) ERP service provider

Enterprise resource management ERP is the core management work for sellers internally, focusing on "things", "people", and "finance". According to the direction of product function extension, cross-border e-commerce ERP SaaS can be divided into operational ERP and supply chain ERP. Operational ERP is based on the basic functions of ERP and extends functions such as data analysis and advertising to the marketing end. Supply chain ERP is based on the basic functions of ERP and extends functions such as sourcing, procurement, warehousing and logistics management to the supply chain end. The following is a brief overview of major cross-border ERP vendors:

3) Supply chain service provider

Cross border e-commerce supply chain SaaS covers the funding, logistics, and product flow links after users place orders, corresponding to three sub tracks: cross-border payment management, warehousing logistics management, and procurement sourcing management. The following is a brief overview of major cross-border supply chain service providers:

4) User reach channels

User reach channels are also crucial in cross-border e-commerce, and the following are the main user reach channels for cross-border e-commerce:

2、 What are the key demand scenarios that AI can empower?

The common goal pursued by all players in the cross-border e-commerce industry, whether it is third-party open platforms, self operated platforms, or brand merchants/manufacturers, is to maximize profits.

The profit of cross-border e-commerce can be broken down into the formula: "Sales profit=Traffic x Conversion rate x Customer price - Product cost - Other operating costs".

The precise content generation and delivery in the promotion process determine traffic acquisition, the automation and intelligence of the sales operation process determine conversion rates and unit prices, and the efficient solution of infrastructure determines the cost and expenses. These are the three most important links in the cross-border e-commerce industry, and the existing AI products mainly provide services around these scenarios.

1. Promotion: Precise content production and delivery

To some extent, e-commerce is actually a way of monetizing traffic, and traffic is the first step in reaching users, determining the scale at which enterprises can reach consumers. Cross border e-commerce mainly has the following promotion methods:

AI can focus on empowering the content production and precise delivery modules in the promotion process. While improving efficiency, the ultimate goal is to enhance marketing effectiveness and increase traffic conversion rates

1) Content production

Almost all promotion channels involve the production of promotional copy, promotional images, and promotional video content, and require the generation of personalized marketing content based on the tone of the promotion platform and the target user group. At the same time, it is necessary to solve the cross-border multilingual problem. AIGC models such as ChatGPT have strong text generation capabilities and can achieve low-cost and high-efficiency content production.

2) Accurate targeting

By utilizing AI capabilities and collecting relevant data and information from consumers, such as purchase history, browsing history, search history, etc., precise analysis of consumers' consumption habits, preferences, needs, etc. is conducted to build a consumer database. Based on the interest characteristics presented by consumers, accurate advertising and marketing content is delivered, including channel selection, keyword research, bid management, etc.

2. Sales Operations: Automation and Intelligence

AI can focus on empowering modules such as product selection analysis, product listing, operational management, and customer service in the sales process, improving overall conversion rates and average customer value

1) Product selection and development

Merchants can use AI to optimize their product selection models, improve product selection efficiency and accuracy, and achieve more efficient product selection decisions from dimensions such as sales routes, advertising platforms, shelf time, and product features.

2) Product listing

Merchants can use generative AI to extract product selling points from user purchasing motivations, product advantages and defects, customer expectations, usage scenarios, customer situations, and other aspects, and generate competitive product detail pages (including product descriptions, high-quality product images, and videos);

Based on analysis of consumer evaluations, product performance, and competitors, automatically generate reasonable pricing recommendations for products, and adjust pricing strategies in a timely manner through real-time evaluation of market dynamics and monitoring of competitive prices.

3) Operations Management

The platform can utilize AI to enhance the correlation between the platform and users, and target potential consumer groups for precise marketing of products.

4) Customer Service

Intelligent customer service can be utilized for pre-sales guidance and recommendation, in sales Q&A, after-sales service, to enhance user experience and strengthen conversion.

3. Infrastructure: Efficient Solutions

AI can also provide efficient solutions in warehousing, logistics, and payment settlement modules, improving efficiency and reducing related costs

1) Warehousing and Logistics

Based on the collection and analysis of weather, traffic, and other information, predict potential logistics risks, plan reasonable delivery routes, and significantly improve order delivery efficiency; Real time monitoring and intelligent management of the complex logistics process enables businesses and consumers to accurately grasp logistics dynamics and provide efficient delivery solutions.

2) Payment settlement

Reduce cross-border transaction risks faced by e-commerce merchants and platforms, such as exchange rate fluctuations and payment fraud.

3、 What are the typical "cross-border e-commerce+AI" products?

1. Platform side

1) Amazon

2) AliExpress

Alibaba's AliExpress announced that the cross-border merchant AI tool Pic Copilot is free of charge. As a mapping tool developed by Alibaba International based on deep e-commerce insights training, Pic Copilot can quickly generate marketing selling point maps that are suitable for different national markets and languages, improving product click through rates. In overseas market deployment tests, the average click through rate of product selling point maps generated by Pic Copilot has increased by more than 7%, significantly better than traditional advertising images.

3）ebay

In 2023, a beta version of the AI tool "magical listing" was released, allowing sellers to take or upload photos in the eBay app, and AI will fill in product details based on these images. At the same time, the image background removal tool has been improved, allowing sellers to share better product images.

4) Alibaba International Station

5) China Manufacturing Network

On May 15, 2023, China Manufacturing Network officially launched the exclusive AI assistant "Michael" for foreign trade enterprises to both buyers and sellers worldwide. It has been upgraded and iterated three times so far. The nine core functions have covered the entire scenario of foreign trade transactions, including marketing promotion, brand going global, and procurement negotiation, with full chain AI service support capabilities.

2. Brand merchant/manufacturer

1) Huakai Yibai

Huakai Yibai is a representative of Amazon's general category sellers, with hundreds of thousands of SKUs and a large number of stores, covering more than 100 sub categories, and has a practical need for the large-scale operation capability of AI. The company adds 200000 to 300000 SKUs annually, and the number of online stores sold on Amazon in 2021 was 678. The company's self-developed "Yibaiyun" digital operation system, combined with ChatGPT, has achieved functions such as intelligent publishing, intelligent price adjustment, intelligent advertising, and intelligent stocking

Intelligent Price Adjustment: Continuously upgraded through independent research and development, achieving rapid price adjustment in batches. As of December 2022, intelligent price adjustment has achieved 6.1 billion price corrections on product pages through algorithm technology support;

Intelligent publishing: The number of monthly listings per capita has significantly increased, combined with the continuous upgrade of ChatGPT, greatly improving the exposure and conversion rate of the company's products;

Intelligent advertising: Continuously upgrading advertising system algorithms, steadily improving the conversion efficiency of promotional expenses;

Intelligent stocking: independently developed MRP system, significantly improving inventory age structure and turnover rate.

2) Jihong Corporation

Actively promote the application and implementation of AIGC technology in various business scenarios to help reduce internal costs and increase efficiency:

Optimizing intelligent product selection: The company independently developed the e-commerce text vertical classification model ChatGiiKin-6B based on Transformer, improving the efficiency and quality of large-scale product selection, and automatically generating matching advertising copy. As of 2023, the e-commerce text vertical model ChatGiiKin-6B has helped the company launch over 100000 new products, increasing the efficiency of product selection decisions, product selling point sorting, and advertising copy output by 95% year-on-year;

Optimizing intelligent advertising placement: The company has launched the Jike Yin AI intelligent advertising assistant ("G-king"). As of 2023, G-king has helped optimization engineers reduce an average of about 3.62% of ineffective advertising budgets, and the work efficiency in multi platform advertising placement operation processes and advertising placement has increased by more than 50% year-on-year;

Optimizing intelligent design and material generation, intelligent translation: As of 2023, the company's e-commerce intelligent design and material generation vertical model GiiAI has helped the company obtain over 50000 graphic/video materials, helping designers improve efficiency by 40% year-on-year in material design and short video editing, and helping small language translation teams improve video material content translation by 90% year-on-year;

Optimize intelligent customer service: The company's cross-border social e-commerce operation management system has been integrated with ChatGPT, and has taken over customer product inquiries and casual conversations for over 300000 customers, saving a total of 5000 customer service hours.

3) Anke Innovation

The company has been deeply involved in the field of algorithmic AI for many years, and its business is highly compatible with AIGC technology. In the future, it is expected to:

Launch intelligent 3C products that combine new technologies such as Chatgpt to enrich the product matrix. AIGC can serve as an important auxiliary tool for smart home appliance technology, enhancing interaction with customers and forming a product ecosystem.

Open the traffic portal to alleviate the pressure of marketing expenses. The company's advertising and marketing expenses account for a relatively high proportion, squeezing profit margins. Chatgpt and similar models are expected to become new traffic entry points, significantly optimizing brand company expense ratios.

3. Service provider

New products/features that have been launched by service providers in combination with AI on the track:

Emerging vertical AI service providers:

Advertising platforms: Google, Meta, Microsoft, etc

4、 Short term risks and opportunities of "cross-border e-commerce+AI" landing

1. In the short term, the combination of AI and manual labor is still the main way for AI products to be applied in cross-border e-commerce

At present, the "illusion" problem inherent in large models and the insufficient fine-tuning training have limited their accuracy. Various AI products still have problems in real scenarios, and any slight "unreliability" in the entire business chain may bring serious consequences. The combination of AI generation and manual adjustment can reduce the risks that AI may bring and improve efficiency.

In product design, based on the current state of AI capabilities, it is necessary to consider the convenience and ease of manual adjustment after AI generates generated content. At the same time, the manually adjusted data can be integrated into the model for continuous iteration and evolution, thereby continuously improving the usability of the product.

In terms of delivery method, you can refer to the image generation service process of Yunxiang Intelligence. After AI generates the effect image, professional manual adjustment is carried out, and finally the image is delivered to ensure the quality and effect of the final deliverables.

Yunxiang Smart Image Generation Service Process

2. In the medium to long term, the ceiling of "AI+cross-border e-commerce" lies in whether it can create conversion rates similar to or better than real people

In various aspects of cross-border e-commerce, AI products can quickly generate images, videos, product descriptions, and copy, as well as provide multilingual customer service and other services, thereby reducing labor costs and improving operational efficiency. However, for enterprises, the cost of saving manpower itself is very difficult to calculate clearly, and there is uncertainty. At the same time, deep cooperation from employees is still needed in the process of using AI products. They need to 'translate' their knowledge and abilities into instructions for AI. In fact, workers are somewhat skeptical and wary of such a new technology that may replace them.

But if it can generate incremental space, such as if I have the opportunity to help the merchant earn an extra 1 million, the other party will be very willing to promise to pay. If AI products can create conversion rates similar to or even better than real people through content creation, product display, and precise traffic acquisition in the promotion and sales operations, it will bring about an increase in sales revenue and huge potential for growth. To create better conversion rates, either provide products that implement end-to-end AI optimized workflows, or specialize in highly segmented scenarios.

At present, AI products focus more on single function applications in a single scenario, and there are very few products that achieve end-to-end AI optimized workflows. The improvement of conversion rates in cross-border e-commerce is interrelated and requires the integration of data from the entire chain.

Taking advertising products in the promotion process as an example, the current common product functions are relatively shallow because they are basically based on the rapid generation of marketing materials and then placed. The entire advertising placement may involve key aspects such as early stage advertising strategy formulation (promotion channel selection, budget allocation, promotion product allocation, etc.), keyword research, material production, bidding management, effect monitoring and optimization. At the same time, the advertising placement effect is also influenced by product details and website operation strategy marketing. If AI products can integrate data from various links and provide full chain services, it will greatly improve the conversion rate of advertising.

There are also companies that have entered into niche areas and taken a successful step towards commercialization, such as Kuang Ye Media from Hangzhou. They have received orders from Alibaba's middle platform, some of which are image materials for external placement on AliExpress, and some are image replacement for internal product detail pages. Unlike in the market, when solving the problem of image materials, companies like Qianmian AI mainly generate clothing model images. Huang Guanyu, the cross-border marketing manager of Kuangye Media, told the brand factory that their main focus is on the ability to produce images for the jewelry category.

3. In the promotion process, traffic acquisition needs to consider the transformation of search engine AI and other AI platforms

Gong Yi from Cross Sea Technology believes that in the next three to five years, or even shorter, the e-commerce industry will achieve over 50% of high-quality articles (including keywords) generated or assisted by AI, with over 90% of the processes being automated. With the improvement of AI generation efficiency, consumers may face the problem of information overload. Therefore, AI tools targeting the B2C industry may customize information according to consumer needs, providing them with condensed and algorithmic information.

At the same time, current e-commerce websites and search engines dominate the main sources of e-commerce traffic. With the implementation of AIGC applications, product sharing, professional evaluation websites, and AIGC platforms may become new traffic entry points. They can collect product information based on big data across the network to complete recommendations, save consumer selection costs, and profit through commissions. Cross border e-commerce sellers who are able to master AIGC search recommendation rules and write product titles/descriptions/detail page copy in the early stages of AIGC application may be able to obtain early zero cost/low-cost traffic dividends and take advantage of the situation to become protagonists.

4. Compliance, intellectual property, and controllability of large models are the main factors that may hinder the adoption of AI based products on the B-end.

Compliance: Whether using ChatGPT or other foreign underlying models, users generally worry about whether they meet regulatory requirements.

Intellectual Property: Currently, the legal definition of intellectual property rights for AI generated content is not clear. Although logically speaking, content generated using AI can also be considered original works, many large companies still maintain a cautious attitude and adopt a wait-and-see attitude towards intellectual property issues. This cautious observation limits the willingness of B-end users to adopt AI based products.

The controllability of AI big models: For example, in terms of material generation, although big models have rich knowledge and diverse art styles, they must be connected to real-life products in order to truly have value. However, it is difficult to train large models in many business scenarios.

This article was originally published by @ You Yao's product notes on Everyone is a Product Manager. Reproduction without permission is prohibited

Image from Unsplash, based on CC0 protocol

The viewpoint of this article only represents the author himself, and everyone is a product manager. The platform only provides information storage space services.

万字长文｜一文了解“AI+跨境电商”通识