Building a Customer 360 — Strategy, MDM Foundation, and Data Model

A Customer 360 is not bought and switched on. It is built in a deliberate sequence on a master-data foundation, with governance underneath every step. Skip the foundation and you get a beautiful dashboard over inconsistent data — the exact problem the 360 was meant to solve.

The implementation strategy

The strategy is a pipeline, and master data is the first stage — master data should be implemented along with Customer 360, not after it:

Stage	What it does
MDM foundation	Scalable, flexible master data — the golden record the 360 is built on
1 · Customer match & merge	Resolve records across sources into one customer with a global ID
2 · Event-driven real-time integration	Stream changes so the profile is current, not batch-stale
3 · Customer 360 data model	A predefined model for quick implementation across subject areas
4 · AI / Gen AI integration	Recommendations, personalization, and chatbots on top of the trusted profile

Two inputs feed the front of the pipeline: existing data-source onboarding and data profiling & data quality. And one thing runs underneath the entire flow — master data & data governance. Value delivered: enhanced customer profiles, streamlined workflows, regulatory-risk reduction through standardized reporting, AI-based recommendations, and chatbot-driven interactions.

Design the model semantically, then source it

The model is designed meaning-first, then bound to sources — the same semantic-layer discipline applied to the customer:

1. Design a logical semantic data model — the business meaning of a customer, independent of any source.
2. Source from existing systems — map the logical model to what you already have.
3. Source from enterprise & other sources — extend coverage across the estate.
4. Source business use cases — take data to the users who need it.
5. Source scheduled & ad-hoc extracts — serve downstream consumers and reporting.

The six subject areas

A banking Customer 360 model organizes attributes into subject areas, each answering a different business question:

Subject area	Focus	Sample attributes
Customer master & product	Product relationship & segmentation	Enterprise keys, demographics, product affiliations, income/FICO segmentation, share-of-wallet
Customer portfolio	Portfolio health & loyalty	Multi-product balances, WTD/MTD/YTD metrics, retention score, lifetime value
Customer channel	Acquisition & interactions	Acquisition/servicing channels, multi-touch campaigns, digital interactions, preferences
Customer behavior & complaints	Sentiment & experience	Complaint type/volume, avg talk time, survey sentiment, churn analysis
Client transactions & events	Risk profiling & life events	Marital-status change, over-credit-limit, delinquency, bankruptcy, spend patterns
Prospect	Cross-sell & recommendation	Conversion ratio, qualified/unqualified leads, campaign offers, product recommendations

These map cleanly onto a dimensional model: a shared customer (member) subject area with conformed dimensions (customer, account, product, channel, date, business unit) and fact tables for transactions, risk exposure, suspicious activity, originations, complaints, campaigns, and daily balances.

-- Conformed dimensions shared across every fact → consistent joins, one grain
CREATE VIEW c360.customer_profile AS
SELECT c.enterprise_customer_id,           -- the golden, unique enterprise key
       c.household_id, c.segment, c.risk_band, c.lifetime_value,
       p.products, p.share_of_wallet,       -- portfolio subject area
       ch.channels, ch.preferences,         -- channel subject area
       b.sentiment_score, b.attrition_risk  -- behavior subject area
FROM   dim_customer c
JOIN   agg_portfolio p USING (enterprise_customer_id)
JOIN   agg_channel   ch USING (enterprise_customer_id)
JOIN   agg_behavior  b  USING (enterprise_customer_id);

The customer profile as an operational store

For real-time use cases (authentication, fraud, servicing), the model is materialized into an operational data store (ODS) that links data sources and attributes into one profile: customer profile, account profile, authentication profile, credential-use profile, device profile, and event history. That one profile then powers marketing (upsell/cross-sell), predictive modelling, and fraud detection from a single place.

The path to implement

1. Stand up the MDM golden record first — match & merge to a unique enterprise customer ID.
2. Add event-driven real-time integration so the profile reflects the latest state.
3. Adopt a predefined 360 data model with conformed dimensions across the six subject areas.
4. Materialize a real-time customer profile (ODS) for operational consumers.
5. Layer AI/Gen AI — recommendations, personalization, chatbots — on the trusted profile last.

Incipient's Data Product Factory supplies the predefined 360 model and governed products; the Data Marketplace is how consumers discover them; and it all rests on the MDM foundation. The analytics article shows what you can do once the model is live.

Talk to us about a Customer 360 build →