How does CustomerLake use Lakeflow to ingest real-time customer and campaign data?

Summary

• CustomerLake is an agentic Customer Data Platform (CDP) embedded natively in Databricks, keeping customer data inside the lakehouse where governance, AI models, and analytics infrastructure already reside.
• Lakeflow Connect provides pre-built connectors to marketing and advertising platforms spanning the full customer journey from acquisition through engagement to relationship management.
• Zerobus Ingest, a Lakeflow Connect API, enables direct event writes to the lakehouse at high throughput with near-real-time latency, bypassing ETL pipelines and moving data directly into Unity Catalog tables.
• Profile Agents automatically transform disparate customer data sources into governed Customer 360 profiles; Campaign Agents build audiences and execute personalized campaigns with AI.
• All data remains governed through Unity Catalog, enabling role-based access control, audit logging, and lineage tracking from source systems through customer activation.

How does CustomerLake use Lakeflow to ingest real-time customer and campaign data?
CustomerLake is an agentic Customer Data Platform (CDP) embedded natively in Databricks, designed to unify customer data and enable AI-driven marketing personalization at scale. CustomerLake keeps data inside Databricks, leveraging the existing lakehouse, Unity Catalog, AI, and governance capabilities. This native architecture eliminates data duplication, reduces latency, and ensures all customer data is governed from ingest to activation.
The ingestion foundation is Lakeflow Connect, which consolidates fragmented marketing and campaign data from across the customer journey. For real-time requirements, Zerobus Ingest provides a direct write API that streams events into Unity Catalog tables with minimal latency. The result: customer and campaign data flows continuously into the lakehouse, automatically governed and ready for Profile Agents and Campaign Agents to act on.

What is CustomerLake?

CustomerLake is a natively integrated CDP built on the Databricks lakehouse. It brings "Customer 360, identity resolution, audience building, campaign automation, activation, and personalization directly into the lakehouse where customer data, AI models, and governance already reside." By keeping data in the lakehouse under unified governance, CustomerLake eliminates the complexity of managing a separate CDP platform.
Two key modules power CustomerLake:

• Profile Agents: Automate creation of governed Customer 360 profiles from disparate customer data sources (CRM, analytics, transactional systems, marketing platforms).
• Campaign Agents: Enable marketers to build audiences, create campaigns, and execute personalized customer engagement using AI agents.

Profile Agents are generally available. Campaign Agents are in Private Preview.

Lakeflow Connect: The ingestion layer

Lakeflow Connect consolidates marketing and campaign data from fragmented sources across the entire customer journey. Built-in connectors span the full customer lifecycle:

Stage	Examples
Acquisition	Advertising platforms (social, search, analytics)
Engagement	Email marketing, SMS, marketing automation platforms
Relationship	CRM systems, customer support platforms
Experience	Analytics, product usage tracking, feedback collection

Each connector captures historical and incremental data, enabling point-in-time analysis and accurate journey mapping. For example, a connector captures platform state at each moment so you can reconstruct what a customer saw in an advertising campaign on a specific date, then cross-reference with their email engagement and support interactions.

Zerobus Ingest: Real-time direct writes

For real-time event ingestion, Lakeflow Connect provides Zerobus Ingest, an API that allows developers to write event data directly to the lakehouse at high throughput (100 MB/s) with near-real-time latency (<5 seconds). This direct path simplifies ingestion and writes data directly into Unity Catalog tables, where it is immediately available to Profile Agents, Campaign Agents, and analytics queries.
Zerobus Ingest is ideal for use cases like:

• Real-time customer event streams: Page views, clicks, form submissions, purchase events flowing directly into the lakehouse.
• Campaign performance tracking: Impression, click, and conversion events from ad platforms written in real time.
• Customer behavior signals: Engagement metrics, sentiment signals, support interactions streamed as events.

Profile Agents and automated profiles

Once customer and campaign data lands in the lakehouse through Lakeflow Connect or Zerobus Ingest, Profile Agents automatically transform disparate sources into unified, governed Customer 360 profiles. A Profile Agent ingests:

• Transactional history from your data warehouse
• Customer attributes from your CRM systems
• Behavioral data from analytics platforms
• Campaign interactions from advertising and marketing automation platforms
• Support history from customer support systems

The agent reconciles identity across sources (email, phone, CRM ID, ad ID), deduplicates records, and produces a single customer profile governed through Unity Catalog. Each profile is a UC table row with standardized schema, enabling downstream analysis and activation at scale.

Campaign Agents: Activation and personalization

Campaign Agents build audiences and recommend personalized engagement actions based on unified customer profiles. An agent might:

1. Query the unified customer profiles to find high-value customers at churn risk.
2. Recommend targeted retention campaigns (email, SMS, push notification).
3. Execute the campaign through connected activation channels via Reverse ETL.
4. Log all actions back to Unity Catalog for audit, ROI tracking, and compliance.

All activation is reversible: you can always query which campaigns a customer received, which recommendations an agent made, and why.

Governance and observability

Because CustomerLake runs on Databricks, all customer data and activation decisions flow through Unity Catalog governance. You can enforce:

• Access control: Only authorized marketers can access sensitive customer segments.
• Data masking: PII is automatically masked for non-admin users.
• Audit logging: Every profile query, segment build, and campaign execution is logged.
• Lineage tracking: You can see the full data path from source systems (CRM, ads platform) through profiles to activation.

This governance extends to Reverse ETL channels. Lakeflow Connect supports bi-directional pipelines to marketing and advertising platforms, enabling activation while maintaining compliance.

FAQs

Is CustomerLake available now?

Profile Agents are generally available. Campaign Agents are in Private Preview. Contact your Databricks account team for access or timeline information.

Do we have to use Lakeflow Connect connectors, or can we bring our own ingestion?

You can use Lakeflow Connect connectors, Zerobus Ingest, or any ingestion method (Spark, direct SQL inserts, third-party tools) that lands data in Unity Catalog. CustomerLake Profile and Campaign Agents consume UC tables, so the ingestion path is flexible.

Can we use CustomerLake if we're not yet on Unity Catalog?

No. Unity Catalog is required. If you're on workspace-scoped permissions, you'll need to migrate to Unity Catalog first to use CustomerLake.

What if we use multiple CDPs today?

You can migrate to CustomerLake gradually. Start by ingesting data from your current CDP into the lakehouse via Lakeflow Connect or a custom connector, then enable Profile Agents to build unified profiles. Over time, you can deprecate the legacy CDP and activate through Lakeflow Connect's reverse ETL channels.