Define the use case first.
Start with specific questions: has this customer purchased on Shopee but never on the website? Who has purchased twice and is now overdue for repurchase? Which existing customers should be suppressed from acquisition campaigns?
The use case determines which data needs to be connected first.
Identity resolution should be pragmatic.
Phone, email, customer ID, platform ID and device signals all have limitations. The business does not need perfect identity before creating value; it needs confidence levels and explicit matching rules.
Avoid false merges because an incorrect SCV can scale incorrect decisions very quickly.
The data model should support lifecycle decisions.
At minimum, distinguish profile, transaction, interaction, consent and derived attributes. From there the business can build lifecycle states, recency/frequency/value and next-best-action logic.
SCV is not a dashboard. It is a data layer that makes CRM decisions more consistent.
Governance cannot be postponed.
Consent, deduplication, access control, data freshness and source-of-truth need to be defined early. As volume grows, small governance gaps become large operating costs.