[Interview] Wekeep CEO Jang Bo-young: “Platform-Specific Delivery Rules Need to Be Absorbed by the System”

Changes in logistics center operations are emerging amid the race for faster delivery. As multi-platform selling by merchants has expanded to Naver, Cafe24, Gmarket, and others, logistics centers have been faced with the challenge of simultaneously responding to different order cutoff times across platforms. At the same time, they are also expected to meet different fulfillment standards by channel.

Operational complexity is rising further as prepacking based on demand forecasting is added to the mix. As a result, competition in fulfillment is shifting from warehouse size to data and operating systems.

To respond to these changes, Wekeep operates its in-house system, "FBW(Fulfillment By Wekeep)." FBW is designed to link order and inventory data from more than 30 shopping malls with logistics center operations. It also uses AI to predict products with a high likelihood of shipment and applies a pre-shipment packaging technology called "PrePack."

Wekeep’s "PrePack" is a function that analyzes sales data to prepackage products likely to be shipped before an order is placed. When an actual order comes in, a shipping label is attached afterward and the item is dispatched immediately.

Wekeep presented the effect of "PrePack" by comparing annual operating results at its Incheon Hub Center in 2023 and 2024. As a result, average fulfillment processing time fell from 36 hours to 7 hours.

Work efficiency indicators also improved. Picking throughput per worker per hour rose from 60 orders to 100 orders.

The basis for these operational results was presented as an AI-based technology stack. The stack consists of prepacking, automatic inbound requests, and automatic inventory replenishment and movement, and the technology was designated last year as the 8th Excellent New Logistics Technology by the Ministry of Land, Infrastructure and Transport.

The person in the photo is Jang Bo-young, CEO of Wekeep. Jang said the core competitive edge lies in an operating system that connects AI prediction results to actual warehouse work.

Jang said that simply improving the performance of AI models does not automatically raise logistics center productivity, and that prediction results must be reflected in on-site procedures. He added that only when such procedures are implemented on the ground can technology be connected to actual operating results.

In the Q&A session, Jang explained that as platform-centered operations have intensified in fulfillment sites, logistics processing standards have become different for each platform. Regarding recent changes, he said customer demand is increasing and that logistics processing tailored to each platform’s operating rules has become important.

The platforms Wekeep participates in are Naver N Delivery, Cafe24 Daily Delivery, and Gmarket Star Delivery. Jang said each of these platforms has different order cutoff times and different data integration methods.

For sellers on multiple platforms, even the same product must be shipped according to the standards of each channel. Accordingly, Jang said that to process orders from multiple platforms simultaneously at a single center, the system must be able to absorb each platform’s operating rules.

Jang pointed to real-time integration as the key to handling platform-specific delivery rules. He added that platform delivery has the characteristic of promising a consumer-facing arrival date, so meeting the set order cutoff time and processing time becomes an operational requirement.

He also said that when an order is placed, it must immediately be converted into a work instruction for the logistics center. He explained that a method in which a person downloads and transfers files cannot meet such demand for speed.

Wekeep develops and operates FBW in-house and is linked with more than 30 shopping malls. Accordingly, when platform specifications change, its own system can respond. However, to secure delivery speed, on-site work speed is also important, and even if order information is received quickly, delays in product picking and packaging can lead to delivery delays, making simultaneous design of the system and on-site operations necessary.

Under this structure, PrePack works by predicting products with a high likelihood of shipment based on past sales data and prepacking them before an order is received. Once an order comes in, the item can be shipped immediately after the shipping label is attached.

PrePack does not apply to all products. It is used for fast-moving items and products for which sufficient forecast reliability has been secured. At the Incheon Hub Center, PrePack is applied to up to about 60% of total shipments during peak periods, and 99.5% of products processed through PrePack are shipped within 5 days.

Wekeep continuously checks actual sales and shipment flows while adjusting the volume of PrePack items. Products that are not shipped as expected can be converted into general inventory if necessary. It also does not follow AI predictions blindly, instead operating a structure that adjusts work volume by reflecting actual shipment data.

The interviewee identified securing frontline workers’ trust in AI predictions as the biggest challenge in introducing AI into a logistics center. He explained that frontline workers make judgments based on their existing experience and the situation on the day, so it is difficult for them to immediately start prepacking simply because the system predicts that a certain item will ship in 3 days.

He said that is why continuous confirmation of actual shipment results is necessary. He explained that the purpose of checking actual shipment results is to verify prediction reliability.

He also said that procedures are needed to connect AI predictions to work instructions, along with adjustment procedures based on on-site needs. He stated that improving AI model performance alone does not automatically increase logistics center productivity, and that prediction results must be reflected in on-site procedures.

He further explained that the condition for technology to lead to operating results also lies in reflecting it in on-site procedures. The gist of his remarks was that the key tasks in adopting AI are building trust and linking it to on-site procedures.

He said whether to operate logistics directly should be judged based on volatility in volume and the possibility of recovering investment. Given the nature of businesses where seasonal volume can change severalfold, facilities and labor need to be secured for peak season, but doing so increases fixed costs during the off-season, he explained. Accordingly, he suggested a model in which stable-volume areas are handled in-house while highly volatile areas are handled together with outside specialists. The gist of his remarks was that logistics operating methods should be decided by volume volatility and investment recovery potential.

This is not a decision based solely on company size. Even large companies that operate their own centers may choose outsourcing, and factors in that decision include fixed-cost burden and operating efficiency. In this process, logistics companies are required to absorb volume fluctuations.

A question was then raised about the dividing line for future fulfillment competitiveness. The gist of the answer was that delivery speed and operational stability are important. The trend that the discussion of competitiveness is shifting from facility size to operating quality was presented.

In particular, if SKU growth and continued changes in order patterns persist, organic integration of inventory, fulfillment, returns, and CS is necessary. This is because even a problem in a single process can affect the overall quality of service. It was emphasized that as order and inventory complexity increases, the ability to manage multiple operational stages in an integrated and stable way becomes more important.

Accordingly, the direction for fulfillment development was presented as integrated management of demand forecasting, inventory allocation, and shipment priority. It was summarized that future competitiveness will shift from warehouse size to the ability to process complex orders and inventory precisely and stably. The gist of his remarks was that precise and stable processing capability, rather than owning a larger warehouse, will determine competitiveness.

Source: IT DAILY · Kim Byung-ju
Original: https://www.itdaily.kr/news/articleView.html?idxno=241004

References

• https://www.itdaily.kr/news/articleView.html?idxno=241004