Wumart cut inventory by making stores give up their back rooms

In July 2025, Wumart began piloting a daily-delivery model with Dmall's AI systems. The old model carried inventory in store back rooms, supplier buffers and warehouse safety stock, which made new O2O and group-buying demand harder to serve.

The clever move was organizational, not just algorithmic. Before daily delivery went live, stores converted or cleared storage space and returned or sold through goods so the model started from zero back-room inventory. AI selection then narrowed assortments toward higher-value SKUs.

The model also changed merchandising: lower shelves, fewer top-stock areas, fewer ineffective promo piles and AI-guided replenishment. More freight movements replaced larger static buffers.

CCFA's 2026 best-practice write-up said Wumart had scaled the model widely by the end of 2025, with inventory turnover days down about 15% and out-of-stock rates controlled within 3%.

Use moving cost to kill static cost

Removing the back room made stores obey the new cadence instead of quietly rebuilding inventory.

Narrower SKU choice made daily replenishment feasible and reduced dead stock.

Frequent supplier-to-store flow turned logistics cost into a substitute for tied-up capital.

AI ordering mattered because the operating system had been simplified first.

When inventory is the trap, do not only optimize the pile. Redesign the cadence so goods spend less time waiting anywhere.

The China Chain-Store & Franchise Association selected the Wumart-Dmall project for its 2026 retail supply-chain innovation cases. The association framed the method as a replicable way to exchange high-frequency logistics work for lower inventory risk and faster capital turnover.