Skip to main content

Shinkong Mitsukoshi Luxury Retail PoC

Published 2026-05-14Updated 2026-07-024 min read

"A department store innovation PoC that ties together VIP members, foreign members, luxury tenant brands, duty-free, and store data with external trend, FX, and tourism signals through Ontology + Agentic AI"

VIP membership across 19 Shinkong Mitsukoshi stores, 700+ luxury tenant brands, foreign tourist behavior (Japan, Hong Kong, Southeast Asia), and duty-free data integrated into a single KG. Uses public information only.


🚀 Interactive Demo

👉 Open demo page (new window)

Switch between 11 scenarios from the sidebar to experience the persona switcher, foreign tourist analysis, luxury SOV, and VIP concierge UIs hands-on.


1. PoC One-Line Summary

AxisDescription
Target audience5 departments: Marketing, Customer Insights, Digital, VIP Membership, MD
SubjectVIP members, foreigners, 700+ tenant brands, duty-free, stores + 4 external signal types
MethodOntology (KG with 25 classes) + Agentic AI (Bedrock + AgentCore)
DemonstrationLive execution of 11 scenarios + 5-department persona switcher
DifferentiationForeign duty-free recommendation + luxury SOV + VIP concierge — 3 specialized capabilities

2. Business Structure (Public Information)

AxisDescription
Stores19 (Taipei Xinyi/A11/A8/A9/A4 + Taichung/Tainan/Kaohsiung/Taoyuan/Chiayi)
Tenant brands700+ (LV, Hermes, Chanel, Gucci, Dior, Prada, Tiffany, etc.)
Membershipskm-member (VIP/Black/Platinum/Gold/Standard)
Duty-freeForeigner TaxRefund + duty-free shop operations
Key seasonsAnniversary Sale (週年慶, major fall sale), Chinese New Year (春節), SS/FW

3. 5-Department Personas

CodeDepartmentKPI
P1Marketing & CampaignsAnniversary Sale ROAS · Foreigner acquisition · New product launches
P2Customer InsightsVIP LTV · Foreigner share · SOV
P3Data & DigitalModel accuracy · Duty-free data quality · Mobile SLA
P4VIP & MembershipTier upgrades · Lounge utilization · Concierge NPS
P5MD & Store OperationsStore and floor revenue · Tenant brand turnover · Counter occupancy

4. 11 Scenarios (One-Line Summary)

#ScenarioData Mix
S1Natural-language semantic searchTenant brand SKU + members + external reviews (Xiaohongshu, Dcard)
S25-department persona chatbotAutonomous invocation of all tools
S3Category / BU insight cardsFirst-party GMV + FX + tourism + competitors
S4Lifestyle persona matching + clusteringRFM + luxury affinity + social
S5Omnichannel campaign ROAS simulatorAnniversary Sale, DM, SNS, influencer channel mix
S6External signal fusion (4 types)First-party GMV × 4 external signals (including tourism metrics)
S7Omnichannel member journeyFirst-party online → store → duty-free → repeat visit
S8Marketing consent / PII guardrailsBedrock Guardrails + duty-free regulations
S9-MForeign tourist behavior analysis + duty-free recommendationPassport, nationality, FX, tourism
S10-MLuxury brand SOV + counter occupancy700+ tenant brands × stores × seasons
S11-MVIP tier lounge conciergeBlack/Platinum/Gold recommendations and pre-sale

5. AWS Stack (Common)

Same as LG H&H + multilingual (Traditional Chinese, Japanese, English), live FX added, ~550K edges (enriched tenant brand catalog).


6. Top-5 Demo Flow (30 minutes)

OrderScenarioDemo HighlightTime
1S1 Semantic search"Female in her 30s + LV/Chanel + Anniversary Sale 50% off"4 min
2S9-M ⭐ Foreigner duty-free"Japanese tourist in her 30s → duty-free recommendation + FX-aware"5 min
3S10-M ⭐ Luxury SOV"Hermes vs Chanel counter occupancy by store"4 min
4S11-M ⭐ VIP concierge"Pre-sale recommendation for a Black-tier member"5 min
5S2 Persona chatbot"Anniversary Sale ROAS" P1 vs P5 comparison4 min

Details: 99. Demo Storytelling


7. Deliverable Index