news

한어Русский языкEnglishFrançaisIndonesianSanskrit日本語DeutschPortuguêsΕλληνικάespañolItalianoSuomalainenLatina

Author: summer
Email: [email protected]

While there was a fierce battle on the European Cup stadium, another game was taking place outside the stadium on the e-commerce platform.

"During this period, the store traffic has increased rapidly, and the total sales have more than doubled compared with the same period last year," said Fang Ping, a 26-year-old Yiwu sports goods merchant. According to AliExpress data, after the official announcement of the "Euro Cup" sponsorship, AliExpress's football goods sales in May increased by 80% year-on-year.

In fact, cross-border e-commerce has maintained a growth trend in the past two years, not just during the European Cup. Taking 2023 as an example, data from the Ministry of Commerce showed that my country's cross-border e-commerce exports reached 1.83 trillion yuan, a year-on-year increase of 19.6%.

Compared with the slowdown in the growth of the domestic e-commerce industry, the huge potential overseas and the lowering of the threshold for going overseas through full-hosting and semi-hosting services have led merchants to choose to go overseas to find new continents. Among them, small and medium-sized merchants have become the well-deserved main force.

According to iMedia Consulting data, medium, small and micro enterprises account for 39.4%, 17.5% and 13.6% of the companies that have gone overseas respectively. At the same time, China's overseas business is spreading from mature regions such as Europe, the United States, and Southeast Asia to emerging regions such as the Middle East and Africa.

With new people going overseas and new markets going overseas, the difficulty of doing business has also been increased.

However, things soon took a turn for the better. Since last year, facing a series of problems ranging from language to culture, operations, and compliance, more and more businesses have begun to try to take advantage of AI and go global.

On July 16, at an AI-themed sharing session at Alibaba International, Zhang Kaifu, vice president of Alibaba International Digital Business Group and head of AI business, revealed that in the past year, Alibaba International has established an AI team, tested AI capabilities in more than 40 scenarios, empowered 500,000 small and medium-sized businesses, and optimized 100 million products. Merchants' demand for AI is also growing. Data from the past six months shows that on average, merchants' calls for AI double every two months.

"The smaller and medium-sized businesses are, the more they can benefit from the application of AI," said Zhang Kaifu.

1

When Yiwu's foreign trade bosses began to collectively roll out AI

Yiwu bosses are the most mysterious group of people in the e-commerce industry. They can always keep up with the trends and stay at the forefront of the US elections and world events. Yiwu people have naturally not missed the AI ​​opportunities brought by the explosion of big models, and even started doing business during the day and taking "AI extracurricular classes" at night, taking the lead.

Zhou Zhou, who runs a hat accessories business in Yiwu, is a true portrayal of this. Zhou Zhou, 37 years old this year, has experienced several changes in the e-commerce boom. She started out as a foreign trade business, and then four years ago, when exports were affected by the overall environment, she began to focus on domestic platforms for wholesale. In the past two years, she has followed the trend of full and semi-trusted services and started B2C cross-border e-commerce.

After Alibaba International AI launched AI tools for material processing last year, Zhou Zhou started to try them out. She found that these AI functions are very suitable for businesses that want to switch from domestic to overseas operations.

"Now, I can directly transfer domestic product pictures and information to AliExpress, let AI help me translate product titles, and convert product pictures into local versions of multiple countries, such as Spanish-speaking regions, etc." Zhou Zhou said.

In short, it is to use AI to rewrite all product information so that overseas consumers can understand it and want to place an order.

The most direct change is to rewrite the Chinese characters in the products into local styles according to the language of local consumers. Secondly, the same product has different selling points in different countries. AI can optimize title keywords and detailed information to maximize the attractiveness of the text.

"Before AI, modifying a detail page and translating it into different languages ​​was a painful process. We had to extract the text one by one, repair the images, translate, and format the page. If we wanted to do it well, we would need to hire 1-2 employees just to upload the best content. Uploading 10 pages a day was considered good."

Now that AI is used, Zhou Zhou can quickly move pictures from domestic e-commerce platforms to overseas platforms. "They are basically all usable, especially the multi-language processing of product titles, which is much more accurate than the Google Translate we used before."

In addition, He Bing, a Yiwu businessman who sells ball games, mentioned, "The modification of text and image materials does improve the effect, but I think the most powerful one is the generation of model product pictures."

He Bing mentioned that, generally speaking, merchants in the industrial belt will not hire foreign models to do product pictures. It costs at least 200 yuan to make a set of product pictures in China, but it will cost several thousand yuan to hire foreign models to shoot. If you hire an e-commerce graphic designer, the average monthly salary expenditure is 8K-15K yuan.

"It's OK for B2B business, but for B2C business, consumers will pay more attention to visual effects." He Bing began to use the AI ​​model function provided by Alibaba International AI to create product pictures. "For example, tennis and golf scenes, plus an image of a European or American person wearing sportswear, the texture is immediately improved. We can even select AI models with specific images based on the aesthetic tastes of local consumers and sell them globally with one click."

It is worth noting that in the process of material processing, after a wave of generative AI, AI in the e-commerce field can do more than just translation, image and text generation, and has also integrated industry experience and decision-making capabilities. In other words, AI not only performs basic processing on product images, but can also understand consumer needs, adjust titles, details, and design images, and improve conversion efficiency.

The same decision-making ability is also reflected in intelligent customer service and chargeback defense agents.

Taking AI customer service as an example, Alibaba International has equipped all fully managed and semi-managed products with AI customer service on its cross-border e-commerce platform AliExpress. Data shows that in June, the conversion rate of pre-sales inquiries for products equipped with AI customer service increased by 29%.

Chargeback defense is another pain point that is unique to cross-border e-commerce. Generally, it is because customers call the platform after making a purchase to raise questions about the payment, and then Visa, Master and other card organizations will send an email to the merchant, requiring the merchant to send an email in time to defend and provide evidence, otherwise the money for the order will not be returned. Due to the need to provide financial evidence, small and medium-sized merchants often lack the ability to understand and handle it.

"Out of 500 orders a month, five may be rejected," said Zhou Ming, a merchant who sells small household wooden products. "Before, we treated rejections as a fixed cost and did not deal with them."

This year, when Zhou Ming saw the rejected order, he was going to complain a few words and close the page as usual. However, he saw an option "Download Template" next to the order. "Click download and a file opposing the rejection will be directly generated." He used this file to reply to the rejection email and soon succeeded in his defense.

Another merchant mentioned, “Because AI can understand the content of the email, it can find relevant evidence to write this reply email, and the approval rate on the credit card company side is very high, even higher than that written by a real person.”

According to test data, AI Chargeback Defense Agent can help Chinese cross-border merchants on Alibaba International’s various platforms recover RMB 20 million in losses in a year and protect the rights and interests of merchants.

At present, AI capabilities based on large models, which understand the industry, can analyze, write and draw pictures, have greatly reduced the difficulty of operating for small and medium-sized businesses. This is equivalent to allowing every business with shortcomings to gain a more stable ability to go overseas.

1

Good AI will become the foundation of e-commerce platforms

For the entire cross-border e-commerce model, the effectiveness of AI goes beyond this.

It is understood that since April last year, Alibaba International's AI team has started AI testing on more than 40 key scenarios of cross-border e-commerce, covering the entire chain of cross-border e-commerce, including product graphics, marketing, search, advertising, SEO, customer service, refunds, store decoration, etc.

The test results show that in scenarios such as AI customer service and marketing push, AI performs very well and can bring double-digit results. In addition, in most scenarios, AI can bring 1-30% improvement to merchants in terms of conversion rate, click-through rate, consumer satisfaction, etc. For example: AI automatically generated marketing materials led to a 5% increase in advertising ROI.

The improvement in effect is because, unlike previous algorithms that analyzed a small amount of user behavior data, the large model with multimodal capabilities has taken a clear step forward in understanding a small amount of user behavior data.

Taking the "search, promotion, and advertising" in the marketing process as an example, "Overseas, platforms use multimodal models to understand products and add the understood information to 'search and promotion'. The gains are sometimes even greater than those in China," said Zhang Kaifu.

This is because traditional "search promotion" engines do not rely on the understanding of data through multimodal models, but on specific, quantifiable behavioral data analysis. However, compared with the richness of domestic e-commerce behavioral data, the overseas e-commerce market is larger, with lower penetration rates and more dispersed demand patterns. User behavior data is not complete, and the original algorithm is not accurate enough in processing data.

Today, the multimodal understanding capabilities of large models can provide more accurate insights into users and improve conversion rates.

In order to better meet the needs, AI can work on both the supply and demand sides: on the one hand, on the supply side, based on multimodal understanding, it can identify product images and give more accurate product labels and logos, such as marking a top as straight, black, with pockets, and pure cotton; on the other hand, it is the technical processing of user searches, turning user intentions into more precise instructions, and finally making the two accurately matched.

In particular, the matching of cross-border e-commerce is a cross-language environment. There is noise between the Chinese expressions of merchants and the foreign language needs of consumers. AI builds a bridge in the middle, which can weaken the noise as much as possible and improve the effect.

In the entire process of AI+e-commerce business, unlike other models, Alibaba International AI exists as a foundation.

It is understood that at present, various AI capabilities have been applied on platforms such as AliExpress, Lazada, Trendyol, etc., covering all platforms of Alibaba's international business, serving 500,000 merchants and 100 million products.

After all, e-commerce businesses have a lot of common experiences and processes. For example, merchants need daily sales data analysis, and when launching new products, they need to consider new product explosions... In the process of AI+ business, Alibaba's many international business platforms do not have to reinvent the wheel. They can directly provide services with a large AI base and connect to every link in the daily business process of merchants, so that merchants do not even have a strong "I feel like I'm using AI." After all, for merchants, technology and products are not the focus. The focus is to truly solve the business pain points from a functional perspective.

AI has always been likened to water and electricity. This does not just mean that AI is as indispensable as water and electricity and becomes the basis for the operation of the system. It also means that AI should penetrate into every specific scenario like water and electricity.

As Zhang Kaifu said, "We want to provide everyone with a shared AI infrastructure, and then solve the AI ​​needs of all these businesses in e-commerce."

Of course, as the first batch of systematic AI capabilities implemented on the entire platform, merchants also have some concerns during the application process. For example, when applying large models to process localized information and using intelligent customer service, whether the AI ​​is accurate enough or whether it will talk nonsense.

In this regard, Luo Weihua, head of the algorithm of Alibaba International AI business, mentioned that "the previous neural network translation was to accurately translate the input information word by word. It is essentially impossible to 100% overcome the illusion by migrating to the new large-model-based translation. However, at present, the Alibaba International AI business team is doing three-stage optimization: ensuring that the data volume is large enough, strengthening feedback reinforcement learning, and combining RAG as a backup to improve accuracy."

Last year was the first year of big models, and general big models were frantically increasing parameters and model capabilities; this year, big models began to accelerate their integration with the industry and entered the application stage.

At the current stage, an in-depth understanding of the business and practical consideration of the path of AI+business are the key factors that widen the gap between platforms.

For more information on AI e-commerce going global, please click to follow the official account below