How Suning’s AI‑Powered Smart Replenishment Turns Retail from B2C to C2B

Smart retail uses Internet and IoT technologies to sense consumer habits, forecast trends, and guide production, offering diversified, personalized products and services. At the 2017 Shanghai CES Asia exhibition, Suning presented its intelligent replenishment system, marking a shift from B2C to C2B.

The core of future smart retail is big data, which transforms the traditional B2C supply‑chain into a retailer‑initiated C2B reverse‑driven model, using data to develop reverse logistics and precisely match supply and demand.

The showcased system collects historical data to build basic features, then creates cross‑statistical features across dimensions such as location, time, environment, and consumer profiles. These features feed a modeling pipeline that starts with a simple multivariate linear regression, followed by random‑forest and XGBoost regressions, finally fusing multiple model predictions into an optimal forecast that provides sales predictions and procurement recommendations, solving issues of uncertain purchase quantities, timing, and location scheduling.

In the demo, technical staff simulated three partner products using 4‑warehouse, 7‑warehouse, and 9‑warehouse distribution strategies, illustrating how the system maps supply‑chain radii and offers tailored procurement advice for different suppliers.

After determining replenishment locations, the system displays the impact of various macro‑ and micro‑level environmental factors on the selected product.

The replenishment algorithm analyzes both macro factors—overall market size, retail turnover, sales growth, competitor performance—and micro factors—monthly sales, inventory, growth trends, price trajectory, seasonality, consumer profiles—combined with promotional calendars and local usage habits to suggest optimal replenishment quantities. This enables merchants to confidently anticipate sales, embodying a C2B retail model.

Traditional brick‑and‑mortar retail can see customers but cannot discern their behavior or future trends. In the era of smart retail, algorithmic models driven by big data can predict consumption trends, integrating product analysis, user preferences, time, region, and quantity dimensions to create a data‑driven, intelligent retail ecosystem.