Customer 360 build: integrating fragmented customer data

Context

Universal commercial bank UZ. 800k retail customers, 12k SME, 800 corporate. Customer data fragmented across 7 product systems (ABS, cards, lending, deposits, FX, mortgage, investment). Cross-sell ratio 1.6 products average.

Marketing campaign hit rate 1-2% (target 4-5%). Contact centre handle time high — operator switches between systems.

Diagnostic (5 weeks)

Identity resolution audit:

• 800k customer records → 580k unique persons (28% duplicates).
• Cross-product visibility: <40% of customers visible across product systems.
• Consent state: dispersed, partial, partly invalid.

Approach (12 months)

Months 1-3. Architecture design. Greenfield CDP, identity resolution engine, event ingestion. Not replacement of product systems, a layer above.

Months 3-6. Identity resolution build. Probabilistic matching, golden record builder, manual stewardship workflow for conflicts.

Months 6-9. Profile store + event ingestion. Real-time updates via event bus.

Months 9-12. Channel activation. Pilot — contact centre first (high impact), then mobile app, marketing.

Results

After 12 months:

• Identity resolution: 580k unique persons identified (vs 800k duplicate-laden raw).
• Match rate in profile store: 91%.
• Cross-sell ratio in pilot channels: 1.6 → 2.4 products average.
• Marketing campaign hit rate: 1.5% → 4.2%.
• Contact centre handle time: −22%.
• Customer NPS in pilot channel users: +18.

What is critical

Greenfield architecture decision — do not try in legacy DWH.

Strong CDO sponsorship — cross-team alignment essential.

Stewardship workflow for conflicts — automation does not resolve all cases.

Pilot first channel (contact centre) — measurable impact justifies broader rollout.

What would not have worked

Fold CDP into existing DWH — latency and schema constraints.

Rules-only identity resolution — too brittle.

Big-bang channel activation — operations risk.

Related

• /en/solutions/banking-customer-360-platform/ — customer 360
• /en/architecture/banking-cdp-architecture/ — CDP
• /en/architecture/banking-master-data-management/ — MDM
• /en/decisions/banking-build-vs-buy-cdp/ — CDP decision