Customer Data Platform CDP

⬢ 层级 2技术

高

薪资影响

2 个月

学习时长

中等

难度

职业

一览

A Customer Data Platform (CDP) consolidates customer data from 20+ sources (web analytics, email, CRM, ads, transactions) into single profiles. Engineers build: data pipelines (collect from sources), identity resolution (which events belong to same user?), segmentation (group users), and activation (send to destinations). Mastery takes 3-4 weeks. Senior CDP engineers earn $150-250k because they enable $10M+ in marketing revenue. Becoming one of the 8% of engineers who can design CDPs is valuable.

什么是 Customer Data Platform CDP

A Customer Data Platform (CDP) consolidates customer data from 20+ sources (website analytics, email, CRM, ads, transactions) into unified customer profiles. It enables: personalization (show products based on past behavior), segmentation (group similar customers), analytics (understand customer journeys), and activation (send to marketing tools). Core components: data collection (SDKs, APIs), identity resolution (link devices to users), data modeling (transform raw events into profiles), segmentation (define customer groups), and reverse ETL (push segments to destinations).

🔧 工具与生态

CDP platforms (Segment, mParticle, Tealium)Data warehouse (Snowflake, BigQuery)ETL tools (dbt, Airflow)Identity resolution enginesEvent tracking librariesSegmentation buildersReverse ETL toolsAnalytics databasesCustomer journey mappingTesting tools

💰 各地区薪资

地区	初级	中级	高级
USA	$85k	$145k	$240k
UK	$53k	$90k	$150k
EU	$58k	$98k	$160k
CANADA	$90k	$155k	$255k

🎓 认证

Segment Academy dbt Fundamentals Google Analytics Academy

🎯 使用 Customer Data Platform CDP 的职业

Blockchain Auditor

Email Marketing Specialist

Retail Analyst

⚖ 对比

Marketing Analytics

❓ 常见问题

What's the difference between a CDP and a data warehouse?

Data warehouse (Snowflake, BigQuery): stores all raw data, queries fast. CDP: unified customer profiles + ready-to-activate segments. DW = tool for analysis. CDP = tool for marketing/personalization. Many companies use both (DW for analysis, CDP for activation).

What's identity resolution and why is it hard?

Identity resolution: connecting events across devices/sessions to single user. Hard because: users have 5+ identities (email, phone, device ID, anonymous cookie). CDP must: match emails across devices, suppress duplicates, resolve conflicts.

How do you prevent bad data in CDP?

Implement event validation (schema enforcement): every event must have required fields (user_id, timestamp, event_type). Failed events go to dead-letter queue (not silently dropped). Monitor data quality metrics (% invalid, latency).

What's reverse ETL and how does CDP use it?

Reverse ETL: push data FROM warehouse/CDP TO destinations (email, ads, CRM). Example: CDP creates segment 'churning customers', reverse ETL sends list to email tool (automated win-back campaign). Closes the loop.

How do you segment customers effectively?

Behavior-based (users who did X), demographic (users in location Y), RFM (Recency, Frequency, Monetary value). Complex segments: users who visited product page AND opened email AND spent >$100. SQL/dbt required.

What's the privacy impact of CDPs?

GDPR/CCPA: users can request data deletion. CDP must: identify all records for that user (identity resolution important), delete them all. Consent management: must track which users opted in to tracking. Non-compliance = fines.

How do you measure CDP success?

Business metrics: uplift from personalization (email CTR +10%, conversion +5%), cost of customer acquisition (targeting more efficient). Technical metrics: data freshness (how stale is profile?), segment accuracy (are the right users in segment?), API latency.