User Persona & Crowd Insight Skill

The Core Problem: The Last Mile of Personas and Insights

Many teams accumulate user data and attempt segmentation and persona creation, but frequently encounter two core pain points:
1. Persona assets are "built but never used": Outputs remain as tag lists or fictional personas, lacking binding to quantifiable segments, making them unusable directly by algorithms or strategies.
2. Insights fail to loop into actions: Teams know who users are, but not "whom to target and how." Insights stop at report conclusions, not translated into specific operational actions, channels, and expected metrics, preventing validation of effectiveness.

This often leads to user persona projects becoming "self-congratulatory outputs" that fail to drive precise marketing, recommendation optimization, or product decisions.

How It Works: A Structured Loop from Data to Actionable Decisions

This skill aims to build a cognitive and execution line from "data readiness" to "action implementation." Its workflow and core capabilities are:

1. Diagnosis and Construction (Profile Layer): Establishing a Quantifiable Foundation
* Inventory and Assessment: First, evaluate data foundations (behavior, transaction, attributes, etc.) and identity linkage status with a 6-dimension readiness score (1-5), clarifying gaps and priority supplementary items (P0/P1).
* Design Standard Tagging System: Design tags following the industry-standard CDP Six-Category Taxonomy (Basic Attributes, Behavioral, Consumption Value, Life Cycle, Channel Source, Predictive), ensuring coverage of at least 4 categories. Tag definitions include explicit calculation logic and data sources.
* Build Segmentation Model: Employ standard paradigms like RFM + K-means for segmentation. All segmentation rules must be quantified (e.g., spent ≥3 times in the last 90 days) and assigned a value positioning (high/medium/low).
* Generate Representative Personas: Produce one Persona card (3-5 total) for key segments. These are not fictional "Xiao Mings" but are bound to specific segments (linked_segment), consolidating behavioral characteristics, needs, and pain points to serve team alignment and design.

2. Insight and Closure (Insight Layer): From Cognition to Action
This is the skill's core increment, solving the "unused" problem:
* Crowd Opportunity Matrix: Based on segments, quantify opportunities for each segment along two dimensions: value positioning and growth potential, identifying opportunity types like "core growth" or "value preservation."
* Insight → Action Mapping: Transform each insight into specific, executable operational actions, defining action content, reach channels, and quantifiable expected metrics (e.g., increase redemption rate by 15%, repurchase rate +5%), with target segments bound to previous segments.
* Opportunity Priority Scoring: Score each opportunity (from the matrix and mapping) on a 1-5 priority scale, providing reasoning and execution sequence.

3. Output and Validation
The final output is structured JSON data containing results from all steps, validated via local scripts for schema verification, Markdown rendering, and desensitization checks, ensuring deliverables can be directly used as tagging platform requirements, marketing audience definitions, or product analysis inputs.

Scope and Key Considerations

Clearly Define What This Skill "Does" and "Does Not Do":
* Does:
* Provide a complete framework of persona assets + insight closure, including readiness scoring, six-category tags, quantified segmentation, personas, and mapping from insights to actions.
* Does Not:
* Diagnose overall membership operation maturity (focuses on "persona readiness").
* Define user lifecycle stages and transition rules (refer to skills like User Lifecycle Journey).
* Design specific benefits, activities, or reach strategies (the insight loop provides direction; specific strategies require other skills).
* Write code for algorithms (predictive tags only provide calculation logic and validation requirements).

Core Principles:
* Explicitly Annotate Data Gaps: When data is insufficient, score conservatively and clearly list "missing information and assumptions" and "supplementary checklist" in the output; never fabricate data.
* Quantification is Mandatory: Segmentation rules and expected metrics must have quantification markers (time windows, thresholds); avoid vague terms like "better" or "moderate."
* Default Desensitization: Output must not contain any real customer names, brand names, internal data sources, or identifiable information.