About this skill

Problem

Delivery store and brand data is split across Meituan Waimai, Taobao Flash Sale, and JD Waimai. Operators often need to answer questions such as how yesterday performed, which store underperformed in the last 7 days, or what the top 10 stores by revenue are on each platform. Manual analysis can miss comparison periods, platform boundaries, and TOP-N aggregation rules.

How It Works

dkd-brand-data-analysis parses a natural-language request into time, entity, platform, metric + TOP fields, then routes it to the right workflow. Standard reports use scripts such as generate_store_daily.py, generate_brand_weekly.py, and generate_platform_monthly.py; platform diagnosis uses generate_platform_diagnose.py and render_diagnosis.py; non-standard analysis uses extend_query.py or chain_business.py.

Standard reports: daily, weekly, monthly, near-N-day, and custom ranges, with comparison, trends, commercial-area rankings, and TOP lists.
Diagnosis loop: fetch data, generate HTML and Markdown, produce an AI diagnosis from diagnostic_data.json, then backfill the report.
Platform routing: normalize aliases such as Taobao, Flash Sale, and Ele.me into a single --platform value.

Boundaries

It fits operational analysis when a CHAIN_MCP_TOKEN and DKD data APIs are available. It does not replace business decisions, and it should not silently guess when store/brand, platform, or date scope is ambiguous. Cross-platform “top 10” should be calculated separately per platform rather than merged into one aggregate ranking.

Screenshots

Use Cases

When operations asks for store conversion on Meituan Waimai over the last 7 days, generate a platform weekly diagnosis report.
When a brand owner compares top 10 revenue across platforms, split the result into Meituan, Taobao Flash Sale, and JD Waimai lists.
Before a weekly review, a data analyst generates brand weekly HTML reports and a five-section AI analysis Markdown file.
When a store manager checks whether yesterday was abnormal, route to realtime first and fall back to yesterday's daily report.

Best For

Delivery brand operations: needs daily, weekly, and monthly brand performance with store rankings.
Store managers: needs to diagnose a store's recent performance on Meituan Waimai, Taobao Flash Sale, or other platforms.
Regional data analysts: needs to turn multi-platform TOP, period-over-period, and trend data into reviewable reports.
AI workflow engineers: needs to route natural-language requests to fixed diagnosis scripts and output standards.