Students who have experience in cross-border e-commerce are familiar with the scenario of purchasing on behalf of 1688. Users place orders on independent websites, and then purchasing agents procure the goods from 1688, inspect and store them, and finally package and ship them overseas.

The most troublesome part of this process is not the procurement itself, but the excessive manual operations involved in the procurement process. From receiving an order to completing the procurement, there are various steps including checking goods, comparing prices, placing orders, urging shipment, and entering logistics information - all of which are done manually.

Can this process be automated? The answer is yes, but there are several key points that need to be addressed. Recently, I have studied several purchasing systems, and I will take Taocarts as an example to illustrate how its automated workflow is set up.

Step 1: Automatically matching the purchasing source. When the user places an order for a product on the independent website, the SKU listed on the website needs to be mapped to the actual supplier on 1688.

This mapping relationship cannot be manually matched one by one, as it would collapse when the volume becomes large. A good practice is to store the "upstream source" as metadata during the product listing stage.

Use the collection tool to link the supplier_id, product_id, and sku_id of 1688 products to the independent station product records. During subsequent purchases, directly pull based on this mapping without manual intervention.

Step 2: Automatic price comparison. The same SKU may be available from multiple 1688 suppliers, with varying prices and shipping costs.

The automated procurement system should be able to simultaneously request quotations from three to five suppliers within one second, and then sort and push them to the procurement operator based on the total cost (including shipping). This process typically employs a parallel request + timeout control scheme.

Send out five requests, and use the one that comes back first within two seconds. Discard and skip any that are overdue.

Step 3: Automatic order placement. This step is sensitive as it involves real money transactions.

Currently, most schemes utilize a semi-automatic mode - the system presents the price comparison results, and after the buyer clicks confirm, the system automatically completes the order on 1688. Why not make it fully automatic? There are mainly two reasons: Firstly, there is a verification code on the 1688 order page, and secondly, there is a risk of fraud control when placing orders in bulk.

Therefore, the most stable approach is "system preparation of orders + manual confirmation + automatic execution", combining human decision-making with machine efficiency.

Step 4: Logistics feedback. After the supplier ships the goods, 1688 will generate a logistics tracking number.

At this point, a scheduled task is needed to retrieve the logistics information of shipped orders and synchronize it to the backend of the independent website. The logistics status displayed to the user should not deviate from the actual situation by more than two hours.

Using webhook callbacks is better than polling in this case, but the logistics API capabilities of 1688 are limited. Most of the time, we still have to rely on scheduled pulls and exception retries.

Running the entire assembly line, a skilled buyer can handle a maximum of 30 to 50 manual purchases a day. However, with this automated process, it can easily handle over 200 orders in the same amount of time.

Moreover, by reducing manual operation steps, the error rate has actually decreased.

If you are using Taocarts or a similar system, it is recommended to take some time to configure this automated procurement process. The initial investment of time in configuration will save you several hours every day in the future.