I'm building a customer 360 in BigQuery, should I use managed connectors or code?

For a customer 360 in BigQuery, the best choice is usually:

• Managed connectors for standard SaaS sources and faster time-to-value
• Code / custom pipelines for anything complex, high-volume, or highly governed
• Often, a hybrid approach is the real answer

Rule of thumb

Use managed connectors when:

• You need to ingest common sources like Salesforce, Google Ads, Stripe, HubSpot, Shopify, etc.
• You want quick setup and low operational overhead
• Your data model is fairly standard
• Your team is small or doesn’t want to maintain lots of ingestion code
• You’re okay with the connector’s supported schema and sync behavior

Use code when:

• You have custom internal systems
• You need complex transformations, deduping, identity resolution, or event stitching
• You need very specific latency, retries, backfills, or CDC behavior
• You have strict control, compliance, or lineage requirements
• Your workloads are large and cost optimization matters

In a customer 360 specifically

A customer 360 usually has these layers:

1. Raw ingestion
2. Standardization / cleansing
3. Identity resolution
4. Customer profile mart / golden record
5. Activation outputs

Managed connectors are often great for layer 1.
Code is usually better for layers 2–5, especially identity resolution and golden record logic.

Recommended approach

Best practical pattern

• Managed connectors for SaaS ingestion into BigQuery
• Custom SQL / dbt / Dataform / Python for modeling and customer 360 logic
• Code-based ingestion only for custom apps, APIs, files, or edge cases

This gives you:

• Faster implementation
• Less maintenance
• More flexibility where it matters

Decision matrix

Need	Better choice
Fast setup	Managed connector
Standard SaaS source	Managed connector
Custom API / app DB	Code
Complex business rules	Code
Identity matching / dedupe	Code
Low ops burden	Managed connector
Full control / portability	Code
Multiple sources with mixed maturity	Hybrid

When I’d strongly favor code

If your customer 360 is a strategic data product and you expect:

• multiple identity graphs
• advanced event modeling
• heavy governance
• custom SLAs
• frequent schema evolution

then building the pipeline logic in code or SQL is usually worth it.

Bottom line

If you’re asking for the default recommendation:
Use managed connectors for ingestion, and use code/SQL for the customer 360 model itself.

If you want, I can also give you:

1. a reference architecture for customer 360 in BigQuery, or
2. a build-vs-buy decision framework for specific sources like Salesforce, Shopify, or custom apps.