User Portrait Development Process and Key Deliverables

As big‑data technologies mature, enterprises increasingly aim to leverage data for refined operations and precise marketing, which begins with constructing detailed user portraits.

During the planning phase of a portrait system, it is essential to define the development and launch process, schedule milestones, key deliverables, and potential difficulties.

Stage 1: Goal Interpretation – Clarify the target audience of the portrait (e.g., operations staff, data analysts) and the expected outcomes such as behavior analysis, personalized recommendation, churn prediction, or targeted marketing.

Stage 2: Task Decomposition & Requirement Research – Based on the identified audience, map business scenarios to data dictionary entities and tags, and determine analysis dimensions such as attribute, behavior, preference, and group‑level portraits.

Stage 3: Requirement Scenario Discussion & Confirmation – Produce a "Product User Portrait Requirement Document" that records application scenarios, tag definitions, and usage methods, and iterate with stakeholders until consensus is reached.

Stage 4: Application Scenario & Data Scope Confirmation – Align the defined tags with existing warehouse tables, produce a "Product User Portrait Development Document" describing the model, involved databases, and implementation workflow for internal data‑operations teams.

Stage 5: Feature Selection & Model Data Insertion – Data‑analysis engineers write HQL scripts to materialize model logic into temporary tables and validate that the extracted data meets business requirements.

Stage 6: Offline Model Data Acceptance & Testing – After data is loaded, schedule incremental updates, verify HQL logic, and provide feedback to the warehouse team for adjustments.

Stage 7: Program Launch & Effect Tracking – Use Git for version control, deploy the solution, and continuously monitor tag performance and business feedback to refine models and weight configurations.

Key Deliverables Across Stages

The project’s major outputs include:

Tag Development: Define tag metrics, confirm data sources, and implement tags, which constitute a large portion of the effort.

ETL Scheduling Development: Design task dependencies, develop scheduling and monitoring scripts, and launch the scheduling system.

Service‑Layer Integration: Build interfaces that expose warehouse data to downstream business systems.

Portrait Productization: Collaborate with product managers and engineers to prototype features, schedule development, and load data into target tables.

Optimization: Refactor tag calculation, scheduling, and data‑sync scripts for efficiency and stability.

Business‑Side Promotion: Create usage documentation and provide support to enable business units to apply portrait insights for increased engagement and revenue.

Source: Excerpt from "User Portrait: Methodology and Engineering Solutions" , published with permission.