Before initiating any Customer Data Platform migration, a rigorous audit of your incumbent platform's data completeness is the single most critical—and most frequently underestimated—preparatory step. The assumption that "we'll just move what we have" leads directly to the propagation of silent data quality issues, schema inconsistencies, and, ultimately, a compromised foundation in the new environment. This audit is not merely about volume counts; it's a forensic examination of data health, structure, and actionable utility.
I advocate for a structured, three-layer audit framework. This must be conducted *before* any contract is signed with a new vendor, as the findings will materially impact migration scope, timeline, and even your final platform selection.
**Layer 1: Event & Entity Inventory**
* Catalog every distinct event type (e.g., `PageView`, `AddToCart`, `FormSubmission`) and core entity (e.g., `User`, `Account`, `Product`) currently tracked.
* For each, document:
* **Raw Volume & Time Range:** Total record count and the effective date range of available data. You will often discover that historical event retention is far shorter than assumed due to past platform tier constraints or data purges.
* **Property Coverage:** For a sample of each event/entity, calculate the percentage of records where key properties (e.g., `product_category`, `campaign_source`, `customer_tier`) are populated versus null. A property that is null in 80% of your "Purchase" events is a property you may not need to migrate.
* **Schema Inconsistencies:** Identify events with the same name but differing property schemas over time, often a result of uncoordinated tracking updates.
**Layer 2: Identity Resolution Integrity**
* This is the cornerstone of CDP value. You must analyze the graphs themselves.
* Export a sample of resolved user profiles and trace the linkages (e.g., anonymous IDs → email → CRM ID). How many profiles are "stitched" from multiple devices/sessions versus remaining as singleton fragments?
* Assess the logic rules (deterministic vs. probabilistic) used by your current CDP. Are there known gaps (e.g., lack of cross-device logic) that your migration must aim to improve?
* Quantify the "unknown user" problem: what percentage of high-value behavioral events are not attached to a known individual?
**Layer 3: Downstream Integration Impact**
* Map every current destination (e.g., Google Ads, Braze, Salesforce, internal data warehouse) and the specific data points each consumes.
* For each destination, identify the source events, user traits, or computed segments being synced.
* Determine if these data flows are reliant on the current CDP's specific transformation or enrichment logic (e.g., a custom trait calculated in-platform from raw events). Replicating this business logic will be a migration task.
* This analysis directly informs the "re-wiring" effort required post-migration.
The deliverable from this audit should be a gap matrix, not just a report. It should clearly categorize data assets into:
1. **Complete & Migratable:** High-quality, fully-populated data to be moved.
2. **Incomplete & Requiring Source Correction:** Data where upstream tracking must be fixed *before* migration can succeed.
3. **Deprecated & Can Be Retired:** Legacy events or properties with negligible coverage or business value, earmarked for sunset.
Without this disciplined approach, you risk building your new customer data foundation on obscured faults, guaranteeing that your first project in the new CDP will be a costly and time-consuming data cleanup.