Customer Data
One customer record the operating teams can actually share.
CRM, transactions, product usage, support, and marketing data rarely disagree because the business has three customers. They disagree because those systems were never designed as one data product.
Customer data is abundant. A customer definition is not.
Sales, marketing, finance, and support each hold a partial view. Identity resolution is informal. Lifecycle metrics cannot be reconciled. AI and personalization inherit the fragmentation.
Identity is local to each system
The same person or account exists under different keys in CRM, billing, product, and support.
Lifecycle metrics disagree
Acquisition, activation, retention, and expansion are calculated on different grains.
Activation is stuck in the CRM
Downstream analytics and AI cannot use customer context without another extract.
Outcomes
A governed customer data product
Resolved identities, shared attributes, and a grain the business has agreed to.
Consistent lifecycle measures
Funnel, retention, and value metrics that match across dashboards and applications.
A foundation for service and AI
Support, product, and agentic systems read the same customer context under Unity Catalog.
How the architecture works
01
Source mapping
CRM, commerce, product events, billing, and support—each with an owner and an identity key.
02
Resolution and history
Silver models that preserve source fidelity while publishing a usable customer entity.
03
Gold consumption
Segments, 360 views, and features for analytics, applications, and governed AI.
What we implement
- Identity resolution design
- CRM and product event integration
- Customer data products on Databricks
- Unity Catalog permissions by domain
- Lifecycle and value metrics
- Activation into applications and AI
Where this shows up
B2B account 360
Account, contact, opportunity, product, and support history as one operating view.
B2C customer 360
Profile, orders, behavior, and service interactions with a defensible identity model.
What should be measured
The business case is built on a baseline, not a promise. These are the numbers this solution is accountable to.
- Match rate across systems
- Time to answer a customer question
- Duplicate-record rate
- Campaign and retention lift on unified segments
The first sensible pilot
One resolved customer entity, two consuming teams
Resolve identity across CRM and billing for one business line, publish a governed customer data product, and put it in front of sales and service simultaneously.
Questions
Is this a CDP implementation?
Not by default. Many organizations need a governed customer data product on the lakehouse first. A packaged CDP is a later choice, not a prerequisite.
Do we have to replace the CRM?
No. The CRM remains an operational system. Databricks becomes the place those operational records are combined with product, finance, and service data under shared definitions.
Related
Databricks SQL & AI/BI
Building Trusted Executive Analytics on Databricks
Executives do not need more dashboards. They need a small number of numbers that survive contact with finance, operations, and the next question.
9 min
Start with the business case
Find the first data or AI opportunity worth proving.
We evaluate the business problem, systems, data, architecture, and economics behind it—then identify the smallest production engagement capable of proving whether the opportunity is real.
Business case first · Architecture-led · Production-focused