← All projects Discuss the project
Project ·
Greenfield Real-Time CDP implementation in insurance
Sector: Insurance
Context: a large Spanish insurance company
- Adobe Experience Platform
- Real-Time CDP
- Adobe Journey Optimizer
- Query Service
- Data Prep
- Confluent Kafka
- Web SDK
The problem
The insurer had no CDP. Historical data and real-time events needed to feed app personalisation through AEP and AJO. The first step was an MVP that could move into production and support further use cases.
What I did
- I defined the architecture and led implementation from MVP to production readiness, deciding which data belonged in Profile and which in the data lake.
- I resolved identities arriving from legacy systems with Data Prep and Profile. I scheduled Query Service queries to reconcile them.
- I integrated consent from Kafka through to Profile and activation. I prepared a separate load for historical consent records.
- I built app personalisation in AJO. Testing showed that journey updates to Profile did not persist in the data lake, so I designed nightly reconciliation. The handover documented that dependency, monitoring and recovery.
- I defined the target architecture for Decisioning and new channels, separately from the deployed capabilities. Decisioning was not in production at handover.
Result
- Duplicate profiles created by events and loads from legacy systems dropped by around 30%.
- Emails that arrived unmapped fell to less than half of their initial volume.
- The instance moved into daily operation, with over a million profiles.
Where to go next
How I can help
First production use case →
For taking a first use case from its sources through activation and operational handover.
Try the decision
Profile or data lake? →
Try choosing which events need Profile and which can stay in the data lake, a central decision in this project.
Other related patterns
Do you have a similar problem?
Tell me what you need to solve and which platforms you use.