Product Overview: Composable Customer Data Platform (CCDP)

Purpose

This document provides an overview of the Composable Customer Data Platform (CCDP) application, outlining the key features available in the current MVP release to help understand/navigate the platform, orchestrate campaigns, leverage AI agents, and monitor performance.

Executive Summary

The Composable Customer Data Platform (CCDP) serves as a marketer-centric product that bridges the gap between static enterprise data warehouses and real-time customer activation. Unlike traditional CDPs that function as "black boxes" requiring data duplication, the CCDP operates directly on top of your existing cloud infrastructure. It unifies customer identity, generates AI-driven insights, and orchestrates personalized experiences across Web, Email, and Paid Media—all without moving your data.

Functional Goals (MVP):

• Unified Activation Workflow: Provides a standardized interface to launch campaigns across multiple channels.

• Launching AI agents under Agentic AI solution: Utilizes AI Agents to assist with complex tasks like audience segmentation and insight generation.

Platform Navigation & Functional Scope

The platform is organized into 9 core areas. Some features are built natively within the new interface, while others seamlessly bridge to existing applications for data management.

Functional Modules: Deep Dive

Home (User Profile Dashboard)

The Homepage is the command center, designed to orient the user immediately upon login.

• Performance Overview: High-level metric cards track Active Campaigns, Influenced Revenue, Conversion Rate and total Audience Reach &

• Campaign Health Dashboard: A specialized view monitoring the operational status (eg. Active, Paused, Draft) of campaigns across channels:

  • Web

  • Email

  • Paid Media

• API Health: Monitoring key integration points, such as the Campaign API and Profile Attributes API.

• Recent Activity: A list or feed display the activity/log of the specific user

Customers

This module provides the data foundation for the platform.

• ID Resolution: Access to the ID Graph Syntasa App, which uses deterministic and probabilistic matching to stitch fragmented identifiers (cookies, emails, device IDs) into a single "Golden Record".

• Customer 360: Access the C360 view for each user by stitching fragmented identifiers into a single profile, serving as the central source for audience targeting, personalization, and profile enrichment across all campaigns

• Customer Insights: Redirects to specific dashboards (Superset) for deep dives into Customer 360

• Profile Attributes: A centralized, view-only list that displays all the global data fields/variables (e.g., loyalty_tier, last_purchase_date) that currently exist within the Profile Attributes API layer.

Campaigns

This is the core workflow for campaign orchestration.

A. All Campaigns [List View]

• Management: Users can Search, Sort by Audience Size, and Filter by Channel (Web, Email, Paid Media) or Status (Active, Draft, Paused, Scheduled, Activating, Failed).

• Quick Actions: A comprehensive action menu including Edit (pencil), Pause/Play (toggle status), Archive, and Delete & Duplicate, View Performance (graph icon)

• Campaign Performance Dashboard: Accessible via the "View Performance" action, every dashboard offers granular insights, such as performance by "Trigger Type," "Audience Segment," or "Product Category," enabling marketers to optimize active campaigns based on data-driven evidence.

  Key Metrics by Channel:

  • Web Campaigns: Tracks Recovered Revenue, Conversion Rate Uplift, Impressions, and Engagement Rate.

  • Email Campaigns: Focuses on Open Rate, Click-Through Rate (CTR), and Attributed Conversions.

  • Paid Media : Monitors Audience Sync Size, Match Rate, and Return on Ad Spend (ROAS).

B. The 3-Step Configuration Workflow

Every campaign follows a standardized flow to ensure consistency, with specific variations for each channel.

• Step 1: Define Campaign: Configuration of metadata including Campaign Name, Description, Labels (e.g., Banner-Affinity, Upsell), and the Primary Conversion Goal (e.g., Purchase, Form Submission).

• Step 2: Specify Personalization & Delivery: The core for defining eligibility conditions, additional data attributes to send downstream and activation method

  • Targeting Rules: A visual builder to define eligibility conditions via Profile Attributes, Behavioral Events, Audience Segments, and Predictive Traits, Add Existing Rule with logical conditions (AND/OR)
    - Built on an intuitive "When" (Condition) -> "Then" (Action) workflow, it allows users to define precise audience eligibility. To prevent ambiguous logic, the builder enforces strict evaluation using a 3-level structure:

    • Level 0 (Global): The overarching AND/OR operator connecting all primary rule component cards.

    • Level 1 (Primary Cards): The main condition blocks between the rule types (e.g., Behavioral Events, Profile Attributes).

    • Level 2 (Inner Groups): Nested group conditions within a rule type/card [All conditions within an inner group must share the same operator (all ANDs or all ORs).]

    • Intra-Category Nesting Constraint: Nested inner groups are strictly restricted by rule category. Cross-category logic—such as combining a Profile Attribute (e.g., "Gold Tier") with a Behavioral Event (e.g., "Completed Checkout")—cannot be nested within the same card.

    Rule Types:

    • Behavioral Events: Targeting based on user actions (e.g., "Purchased") with frequency and time refinements. Multiple events can be nested within a single block.

      • Inherited Time Windows: To streamline configuration, the primary time window defined at the top of a Behavioral Event rule automatically applies to all subsequent ‘then’ event conditions added within that specific card.

    • Predictive Traits (AI): Targeting based on AI/model scores like Churn Probability or LTV.

    • Profile Attributes: Static user properties like age, gender, location or loyalty tier.

    • Audience Segments: Membership in pre-existing lists built in the Audience Builder.

    • Add Existing Rule: Provides the ability to search and import previously saved campaign rules into a new configuration, enabling logic reuse and consistency across similar campaigns. [Note: Rules must have a unique name. This name is combined with the Campaign ID to generate a Qualified Segment Key used for API responses and reporting.]

  • Select Data to Send:

    • Mandatory Data: The MCVisID (Visitor ID) is automatically included for every qualifying user to identify them across activation channels.

    • Additional Attributes: Users select data attributes (e.g., purchase_status, product_id) to be passed alongside the segment result.

    • Data Retention: Defines how long this specific user data remains available in the Profile API layer before it expires and is automatically removed.

  • Trigger Mechanism [Optional]: Selection of Event-Based (e.g., add_to_cart + 60s delay), Session-Based, or Exit-Intent triggers.

  • Frequency Capping [Optional]: Define a limit/cap on how frequently users experience a campaign

  • Priority [Optional]: Assign a priority (1-100) to prioritize campaigns if users qualify for multiple campaigns

  • Activation Method: 
    - Profile Attributes API (Default): The default activation method and underlying data layer that powers the Profile Attributes list. When a user qualifies for a campaign, this API automatically merges their segment eligibility flag and any selected additional attributes directly into their central Customer 360 (C360) profile.
    - Dedicated Campaign API: Creates a standalone, separate API endpoint exclusively for this campaign's data, used for isolated data delivery

• Step 3: Activate Campaign: Users set the Campaign Schedule (Start/End dates or "Always On"), set Scheduling Frequency (Streaming/Batch) and a Summary of the configuration details

C. Campaign Rules [List View]

This section serves as a searchable list of all campaign rules created, allowing users to view and reuse eligibility conditions across different campaigns without rebuilding logic from scratch

Audiences

The Audience module enables the creation of precise segments using user defined attributes.
NOTE: Selecting a sub-menu redirects the user to the relevant existing platform UI

• Data Layer: Operates as the backbone, defining the schema for raw data ingested from external sources (S3, Google Cloud, databases). It structures this raw data into tables accessible for segmentation.

• Attributes Builder: Allows users to define individual traits or behaviors derived from the Data Layer.

  • Examples: "Purchased in last 30 days" (Behavioral), "LTV > $500" (Computed), or "Churn Risk High" (ML Score).

  • Function: These attributes serve as the reusable variables for building broader audiences.

• Audience Builder: A no-code, drag-and-drop interface where marketers combine Attributes into actionable segments.

  • Capabilities: Supports complex inclusion/exclusion logic (e.g., "Users who viewed 'Laptops' BUT did not 'Purchase'").

  • Activation: Once built, these audiences can be synced to Paid Media (Google/Facebook), CRM, or Website personalization tools.

• Audience Sync: Manages the activation of defined audiences by configuring and monitoring synchronization jobs to external destinations

Insights

This module provides the analytics foundation for the platform. 
NOTE: Selecting a sub-menu redirects the user to a dedicated landing page (folder view) displaying a collection of relevant Superset dashboards.

• Customer Insights: A repository of dashboards for analyzing Customer 360 profiles, retention rates, and behavioral trends.

• Audience Insights: Dashboards focused on segment composition, audience growth, and overlap analysis.

• Model Insights: Visualization of model performance metrics and drift evaluation

Models

Exposes the platform's AI/ML capabilities. 
NOTE: To streamline navigation, models are conceptually organized into dedicated folders, with sub-menus redirecting users directly to these specific workspaces

• Pre-Built Models: Redirects directly to the pre-built models folder.

  • Product AI: Redirects directly to the pre-built models folder for the Review Summarizer and Recommendations models.

  • Customer AI: Redirects directly to the pre-built models folder for the OTF (On-the-Fence) and Cart Abandonment models.

Integrations

Centralized interface for configuring and monitoring connections with external systems.
NOTE: Selecting a sub-menu redirects the user to the relevant app folder view

• Inbound Integrations: Manages the configuration and execution of data ingestion pipelines. It handles the intake of raw data from various sources (e.g., Data Warehouses, CRM systems like Salesforce/HubSpot, Web Analytics tools like Google Analytics, or direct API connections) into the CCDP's foundational data layer.

• Outbound Integrations: Manages the configuration of data delivery pipelines. It controls how data is pushed from the CCDP to external activation destinations (e.g., Paid Media platforms like Facebook/Google Ads, Email Marketing platforms, or external databases).

AI Agents

These are autonomous, Gen AI Agents designed to guide the user from analysis to action.

• Customer Insight Agent: Acts as a conversational analytics partner. Users ask natural language questions (e.g., "Show me the churn rate for high-value customers in NY"), and the agent converts this to SQL to generate insights/visualizations.

• Customer Audience Agent: Accelerates segmentation. Users describe a target group in plain English (e.g., "Users who abandoned cart twice in the last week"), and the agent automatically builds the corresponding Attribute rules in the Audience Builder.

Settings

Centralized control for system apps and data pipelines

• System Apps: View and manage system level apps (eg. Rules Parser, Schema Parser)

• System Configurations: Manage the URLs and access points for your core user-facing interfaces, including operational dashboards, application folders, AI agents & essential API endpoints

• Streaming Configurations: Define the operational rules for data pipelines, including Spark runtime settings, event sources, data sinks, and schema mappings.