华立

Baidao Helps Holley Explore AI

Customer background

Hong Kong Huali Technology Co., Ltd. was founded in 2004 and is headquartered in Hong Kong. It is a well – known cross – border e – commerce enterprise in China. Since its establishment, Huali Technology has been committed to innovation and development, and has pioneered the “pan – supply chain, pan – channel” solution. Through continuous R & D investment in aspects such as self – built platforms, self – owned brands, self – built supply chains, and self – developed ERP, this solution has efficiently solved the problems of multi – category, multi – supplier, multi – platform, multi – warehouse, multi – logistics, multi – country, multi – language multi – thread high – concurrency cross – border export B2C retail. The self – owned brand matrix of the company covers more than ten categories including home furnishings, outdoor products, health and beauty products, auto parts, photographic equipment, audio – video products, clothing, toys, computer and mobile phone peripheral accessories, etc. It has hundreds of thousands of products, and its sales network covers 220 countries and regions around the world. In recent years, with the successful operation of the self – built platform, the sales volume of Huali Technology has been growing continuously.

Business requirements and challenges

As a cross – border e – commerce enterprise, Holley Technology has a vast product library and an extensive sales network. In the increasingly complex market demands and the ever – changing competitive environment, especially when managing the inventory of its huge product catalog, in order to improve operational efficiency and profitability, Holley Technology needs a solution that can accurately predict future sales volumes to optimize inventory management and increase the sell – through rate.

Huali Technology has encountered challenges in sales forecasting during its rapid expansion, especially considering the following points:

• Data noise: Outliers and seasonal fluctuations are mixed in the massive transaction data, increasing the difficulty of prediction and affecting the accuracy of the prediction model.
• The influencing factors are complex: numerous factors such as weather, season, price, marketing activities, and consumer preferences can affect sales volume, making it difficult to predict.
• Multi – region prediction: Globalized businesses require precise understanding of consumption habits and market dynamics in different regions.
• The impact factors vary across different categories: Commodities of different categories are affected by various factors to different degrees, and it is necessary to build targeted prediction models.

solution

This solution has elaborately designed an end – to – end intelligent sales volume prediction system. It deeply integrates Google Cloud technology, aiming to create a seamless link from data capture to decision – making support and comprehensively upgrade the intelligence level of its supply chain.

• Real – time data capture: Use Google Cloud Pub/Sub to collect the transaction data streams of Huali Technology’s self – built e – commerce platform in real – time, ensuring the timeliness and integrity of each transaction information and laying a solid foundation for subsequent analysis.
• Efficient Data Processing: Utilize the elastic computing power of the Dataproc cluster and leverage tools like Spark to efficiently clean and preprocess massive amounts of data. Remove outliers, standardize the format, and simultaneously extract key features such as seasonal trends and price fluctuations to prepare a high-quality dataset for model training.
• Automated model building: After importing data into BigQuery for storage and fast querying, the machine learning framework of Vertex AI is adopted. In view of the product classification of Holley Technology and the diversity of the global market, the prediction model is automatically trained and optimized, greatly shortening the model iteration cycle and improving the prediction accuracy.
• Closed – loop from prediction to action: Deploy the trained model on Vertex AI to output real – time prediction results, seamlessly connect to Huali Technology’s inventory management system, drive intelligent inventory adjustment, goods allocation, and customized promotion strategies, and maximize profit margins.

The architecture design diagram is as follows:

Use the product

• Dataproc
• Bigquery
• Vertex AI
• Compute Engine
• Cloud Load Balancing

Customer benefits

Note: [All the following statistical data are from the statistical results of the customer background]

As a cross – border e – commerce enterprise with a global business presence, Hong Kong Huali Technology Co., Ltd. has been committed to optimizing inventory management and enhancing operational efficiency. Through cooperation with Google Cloud, Huali Technology has successfully established an intelligent sales forecasting system, achieving the following remarkable results:

• Data processing efficiency increased by 20%: The powerful computing power of the Dataproc cluster helps Holley Technology efficiently process transaction data, increasing the speed of data cleaning and preprocessing by 15% and laying a solid foundation for model training.
• The prediction accuracy has increased by 25%: By leveraging Vertex AI’s automated machine learning framework, Holley Technology has built precise prediction models for different product categories and regions. The sales volume prediction accuracy has increased to 95%, effectively reducing the risks of inventory backlogs and stockouts.
• Inventory turnover increased by 15%: The intelligent inventory management system automatically adjusts inventory strategies based on real – time forecasting results. As a result, the inventory turnover increased by 15%, releasing a large amount of working capital and reducing warehousing costs.
• Automated Model Building: Vertex AI’s automated machine learning platform helps Holley Technology quickly build and deploy high-precision sales forecasting models. There is no need for professional machine learning knowledge, simplifying the model building process and the iteration of new versions.

Through Google Cloud’s solutions, Holley Technology has achieved an intelligent upgrade of inventory management, significantly improving operational efficiency and profitability, and injecting new impetus into the continuous development of its global business.