skills.rest
mattwg
mattwgCOMMUNITY
Explore all skills by this author →
GitHub

promo-campaign-recap

Computes standardized promo campaign performance cuts from Databricks transaction and Amplitude data.

1|1|Updated Jul 31, 2026
Search Tags:#sql#attribution#databricks#amplitude#coursera#promo analysis#campaign recap
One-click install
npx skills add https://github.com/mattwg/eureka-consumer-ds --skill promo-campaign-recap-mattwg
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: promo-campaign-recap
Source: https://github.com/mattwg/eureka-consumer-ds/tree/main/skills/data-science/consumer_and_degree_strategy_ds/consumer_ds/consumer_promo_analysis/promo-campaign-recap
Command: npx skills add https://github.com/mattwg/eureka-consumer-ds --skill promo-campaign-recap-mattwg

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes references (resource) components.

What problem does it solve? Rebuilding promo campaign post-mortem analyses from scratch each quarter leads to inconsistent numbers and repeated mistakes. This Skill computes the standard recap cuts (cash, redemptions, NPL/Non-NPL, region, channel, page-level attribution) the same way every time, with documented corrections and a reusable campaign registry. ## Core Features & Use Cases - Standardized recap cuts: Produces cash, redemptions, payer-type, regional, traffic-source, and page-level attribution breakdowns for B2C promo campaigns via saved Databricks SQL queries. - Extensible campaign registry: New campaigns are added to one reference file with confirmed promotion_ids and date windows, without touching the methodology. - Documented corrections: Standing fixes (payment_order filtering, C+ Annual SKU dual detection, late-arrival trims, timezone padding) prevent known undercounting and inflation bugs. - Use Case: A stakeholder asks for a Q3 tentpole post-mortem broken out by region and channel; the Skill reuses saved query patterns and the campaign registry to produce consistent, cross-checked numbers. ## Quick Start Ask the assistant to build a post-mortem recap for a named promo campaign broken out by region and traffic channel using the Databricks MCP read-only SQL tool.

Frequently Asked Questions about promo-campaign-recap

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I build a promo campaign post-mortem report in Databricks?▼

Start from the saved queries in the queries/ directory, which implement base metrics, region, channel, and page-level cuts. Apply the standing corrections from known-corrections.md and confirm the campaign's promotion_ids and date window with the requester before running.

How do I add a new promo campaign to the analysis?▼

Discover its promotion_ids by profiling completed_carts with the discovery query in query-patterns.md, then confirm the scope and official date range with the requester. Add the confirmed entry to references/campaign-registry.md and embed its campaign_map in the relevant saved queries.

Why are my C+ Monthly redemption counts roughly double the expected number?▼

Recurring subscription renewals are being counted as new redemptions. Apply the payment_order = 1 OR payment_order IS NULL filter so only a subscription's first payment counts as a redemption.

Why does my Amplitude attribution query time out in Databricks?▼

Databricks cannot use a runtime-computed bound for partition pruning on event_date. Resolve the campaign window with a small standalone query first, then paste the literal dates into the Amplitude-joined query and run one campaign at a time.

When should I not use this promo recap skill?▼

Do not use it for a single ad-hoc metric pull (use promo-metrics-lookup) or for diagnosing why a metric moved (use promo-metric-rca). It also never treats past LTV or revenue results as authoritative; always recompute live.

Related Skills

Explore More Data & Analytics Skills→
daniel-dihardjadaniel-dihardja

brand_brief

Assemble a data-grounded brand brief for a venue from its profile and operating signals.

Community
Intermediate
ryanallenryanallen

analyst-diagnostics

Analyze project documentation to diagnose root causes using Ishikawa diagrams and Five Whys.

Community
Advanced
ai-analyst-labai-analyst-lab
16

data-inspect

Present cached schema details for tables, columns, row counts, and relationships from active.yaml and schema.md files.

Official
Intermediate
ReadyToGo-EducationReadyToGo-Education
15

trafego-insights

Pull Meta Ads performance metrics via Graph API and compute derived KPIs.

Official
Advanced