Experience Architecture and Technical Design of JD Fashion Digital Store

With rapid technological development and rising living standards, traditional retail models no longer satisfy consumer demands; JD introduced the concept of "borderless retail" and uses its fashion digital store as a case study to illustrate both digital and experience upgrades.

Experience Architecture – The store redesign addresses customer attraction, browsing, and fitting experiences while confronting challenges such as network bandwidth, lighting, hardware costs, and sensor maturity. The overall experience framework is shown in Figure 1.

Interactive Marketing – Face‑recognition (gender, age, attractiveness) and personalized "thousand‑person‑thousand‑face" ads are combined with deep‑image matting, skeleton‑keypoint detection, and AI‑driven games. Coupon, red‑packet, and discount mechanisms increase engagement, while brand‑centric, invisible ads boost foot traffic (Figure 2).

Interactive Guidance & Product Information Enhancement – When a customer approaches a garment rack, perception sensors trigger an information‑enhancement terminal that displays product videos, reviews, and promotional data. Pick‑up sensors capture when an item is lifted, feeding real‑time data to the cloud for analytics and personalized recommendations.

Virtual Shelf & Out‑of‑Stock Solution – Interactive shelves allow customers to scan missing items and complete online purchases with home delivery, improving conversion rates and store space utilization.

Virtual Fitting & Immersive Try‑On – Near‑field virtual fitting combines deep‑camera capture, facial fusion, and 3‑D body modeling to render a personalized avatar wearing the selected outfit, supporting scene‑based and 360° previews, and enabling one‑click assistance from sales staff.

Perception Layer – Edge sensors (proximity, pick‑up, try‑on, scan) collect user behavior and demographic data, which is transmitted to the cloud for user understanding and personalized experiences.

Technical Architecture – The system is divided into three layers: perception, display & shallow‑compute, and service. The perception layer feeds data to an edge gateway; the display layer handles interactive terminals and lightweight calculations; the service layer (IoT platform, data platform, business platforms) performs deep analytics. The overall module diagram is shown in Figure 3 and the detailed stack in Figure 4.

Data Asset Accumulation – By instrumenting shelves and garments with sensors, the store can capture traffic metrics (exposures, dwell time), pick‑up counts, and conversion data (order value, items per transaction). These metrics enable fine‑grained, data‑driven store operations as illustrated in Figure 5.

Technical Principles

Scenario‑driven rapid front‑end iteration

Unified language and integrated architecture

Visible testing and real‑time quality assessment

Algorithm adaptability and hardware stability in complex scenes

Componentized, configurable, and minimalistic design

Edge gateway with strong local cache, push‑instead‑of‑pull, asynchronous data back‑haul

Low‑power chip with noise filtering

Engineering SLA based on billion‑scale web architecture

Product Principles

Human‑goods‑scene perception and intelligent interaction

Smart, scene‑based greeting and personalized recommendation

What‑you‑see‑is‑what‑you‑get experience

Continuous, coherent user journey from entry to exit

Fun, immersive, and novel interaction

Proactive, immersive marketing with invisible content embedding

Value‑added content, data tags, and emotional resonance

Usability improvements: cost reduction, screen brightness, touch response

Template‑driven multi‑store, multi‑scenario marketing

Product‑driven rather than demand‑driven development

Solution delivery: capability, product, or scenario‑based

Conclusion

In the fashion digital store, JD has transformed customer attraction, browsing, and fitting through edge‑perception, AI, and IoT technologies, built a low‑cost, scalable solution, and leveraged the accumulated data assets for cloud‑side multi‑dimensional analysis, enabling continuous operational improvement and true digital retail.