Mission: Principal Data Analyst - High-Risk SKU Quality Crisis

[Bibiana-B-Padua]

Mission/open goal Description

Investigated a hidden product-quality crisis where sales, support metrics, customer satisfaction, and reviews collide. Specifically targeted the performance and reliability of Zava's Premium smart-apparel line.

Harness and model

GitHub Copilot with Agent Mode (GPT-4o)

Turn-by-turn journey

1. Prompt: "Help me identify hidden product-quality risks by checking customer support tickets and reviews."
   Agent action: Scanned support logs and text reviews to isolate common failure patterns.
   Result: Detected a high concentration of smart-fabric connectivity complaints after washing for SKU ZCPTM-SS-M-BW.

2. Prompt: "Confirm the exact sales volume and total revenue impact for SKU 'ZCPTM-SS-M-BW'."
   Agent action: Executed aggregate_records tool calls.
   Result:

   • Total Order Lines: 108
   • Total Revenue: 19,825.05

Completion

• Yes, the agent completed the mission or goal.
• No, the agent did not complete the mission or goal.

Bonus work

Executive Summary

Product at risk

• ZCPTM-SS-M-BW — Premium Short Sleeve Men’s Top
• Category: Premium
• Product Type: ShortSleeveTop
• Channel: ZavaCore Field

Quality crisis signal

• The strongest defect pattern is a smart-fabric connectivity failure after washing.
• Supporting evidence:
• Critical support ticket TKT-20240110-00001 reports the product losing Bluetooth/app connection and reconnecting only after pairing troubleshooting.
• Multiple review documents explicitly describe the top “stopping connecting after one wash,” “sensor offline after laundry,” and “app pairing never works again.”

Defect nature

• The failure appears to be in the smart fabric’s wash/dry cycle resilience or its embedded sensor/connection module.
• Customers still find the garment wearable, but the connected functionality — the core value proposition — degrades or fails after laundering.

Volume and revenue impact

• SKU sales lines count: 108 order line occurrences
• Total revenue for ZCPTM-SS-M-BW in the data: 19,825.05

Implication for leadership

• This is a real product-quality risk for a higher-volume premium SKU.
• The defect is not just a single ticket: it is supported by both critical support incidents and multiple customer reviews describing the same wash/connectivity failure.
• Leadership should prioritize a review of the smart fabric durability and return/repair policy for ZCPTM-SS-M-BW, and consider whether additional QA or product recall action is warranted for the Premium smart-top line.

💡 Reflections & Limitations

• What worked: The agent was highly effective at translating high-level requests into targeted aggregate filters and pinpointing specific defect keywords across unstructured review data.
• Where the agent struggled/Corrections: Initially, the agent required clear steering to narrow down from general feedback to calculating hard business metrics (revenue and exact line-item volume) for a single high-risk SKU.

[Bibiana-B-Padua]

📊 Data Flow & Analysis Architecture

Below is the conceptual flow of how the Copilot agent correlated the qualitative feedback with quantitative business metrics to isolate the high-risk SKU:

graph TD
    %% Data Sources
    subgraph Core SQL Database
        A[(Customer Reviews)] -->|Text Mining / Parsing| C{Copilot Agent Engine}
        B[(Support Tickets)] -->|Ticket Filtering| C
        D[(SalesOrderLines)] -->|Aggregation Queries| C
    end

    %% Agent Actions
    subgraph Agent Analysis Phase
        C -->|Isolates Defect Signal| E[Pattern: Wash/Connectivity Failure]
        C -->|Executed aggregate_records| F[Metrics Calculation]
    end

    %% Insights & Outputs
    subgraph "Business Impact (SKU: ZCPTM-SS-M-BW)"
        E --> G[Premium Smart-Top Defect Discovered]
        F --> H[Volume: 108 Order Lines]
        F --> I[Revenue: $19,825.05]
    end

    G & H & I --> J[Executive Summary & Leadership Recommendation]

    %% Styling
    style C fill:#0078d4,stroke:#005a9e,stroke-width:2px,color:#fff
    style J fill:#107c41,stroke:#0b5930,stroke-width:2px,color:#fff

Loading