Customer 360 without customer surveillance

Completeness is the wrong goal

Customer 360 is often described as collecting every available signal into one profile. That framing rewards accumulation. A healthier question is: what is the minimum trustworthy context needed to serve this person, keep a promise, or make a fair decision?

The answer is usually smaller than the data warehouse. Contact history, consent, current work, open obligations, service context, and clearly labeled preferences often matter more than an ocean of behavioral exhaust.

Keep facts, observations, and predictions separate

A customer-provided address is different from an imported enrichment field. A meeting note is different from a model-generated sentiment score. Combining them without provenance makes the interface look confident while reducing trust.

• Show where a field came from and when it was observed.
• Attach retention and purpose to sensitive data classes.
• Label probabilistic inferences and allow correction.
• Restrict agent access to the context required for its current task.

Design the right to forget into the graph

Deletion and export become difficult when customer identity is copied into disconnected systems. A shared relationship model with explicit links makes it possible to find what must be retained, what may be removed, and what should be redacted from downstream context.

Customer 360 should increase the organization’s accountability to the customer—not merely its visibility into the customer.