Customer 360 & Real-Time Risk Intelligence Platform  ·  View 20 of 20

Customer 360 Data Model

The core curated entities, their keys, and the joins that produce the customer golden record and its risk and interaction history.

Editable source SVG draw.io All views
customer_xref
xref_id PK
customer_id FK
source_system
source_key
match_confidence
valid_from / valid_to
customer_xref...
dim_household
household_id PK
household_type
primary_customer_id
region_code
dim_household...
customer_360_profile
customer_id PK, FK
as_of_ts
tenure_months
relationship_value
channel_affinity
next_best_action
customer_360_profile...
dim_account
account_id PK
customer_id FK
product_code
open_date / close_date
status
valid_from / valid_to
dim_account...
dim_customer
customer_id PK
household_id FK
party_type
name_token PII
national_id_token PII
kyc_status
valid_from / valid_to
dim_customer...
dim_risk_rating
rating_id PK
customer_id FK
rating_scheme
rating_value
effective_ts
dim_risk_rating...
fact_transaction
txn_id PK
account_id FK
event_ts / ingest_ts
amount, currency
merchant_category
channel
dedup_key
fact_transaction...
fact_interaction
interaction_id PK
customer_id FK
event_ts
channel, device_id
session_id
event_type
fact_interaction...
fact_risk_event
risk_event_id PK
rating_id FK
customer_id FK
event_ts
signal_type
score, decision
fact_risk_event...
N : 1
N : 1
1 : N
1 : N
1 : 1
1 : 1
1 : N
1 : N
1 : N
1 : N
1 : N
1 : N
1 : N
1 : N
1 : N
1 : N
Customer 360 — Core Data Model
Customer 360 — Core Data Model
PII columns hold surrogate tokens; clear values resolve only through the vault under policy.
PII columns hold surrogate tokens; clear values resolve only through the vault under policy.
v 1.0 · owner Data Architecture · date 2026-08
v 1.0 · owner Data Architecture · date 2026-08Text is not SVG - cannot display

Identity resolution

  • customer_xref maps every source key to one enterprise customer_id with a match confidence
  • Deterministic matching on national identifier and account number first, probabilistic second
  • Low-confidence matches are held for steward review rather than auto-merged

Temporal correctness

  • Dimensions carry valid_from and valid_to so any past state is reconstructable
  • Facts carry both event_ts and ingest_ts so late arrival is distinguishable from late occurrence
  • Feature tables are built point-in-time correct to prevent training-time leakage

PII handling

  • Name and national identifier are stored as surrogate tokens, never in clear
  • Joins use hashed keys, so analytics never requires detokenisation
  • Clear values resolve only through the vault, per call, under a logged purpose
Solution Architecture v1.0 · Enterprise Data Architecture · 2026-0820-customer-360-data-model