Attribution Auditor

Attribution Audit

by AgriciDaniel · MIT license · ★ 9,466 on GitHub · updated Sep 18

Paste the sales and spend numbers each ad platform claims, and get back a plain explanation of why they disagree and which figure to trust.

---
name: ads-attribution
description: "Audit cross-platform attribution, conversion definitions, reporting windows, GA4, AdServices and AdAttributionKit, MMPs, browser and server events, offline conversions, and platform reconciliation. Use for attribution audit, attribution models, conversion windows, requests to add or total Meta and Google conversions, incompatible reporting-window aggregation, GA4 attribution, MMP review, AppsFlyer, Adjust, Branch, Singular, or cross-platform discrepancies."
---

# Attribution Audit

1. Read the main `ads` contract and normalized account snapshots.
2. Declare the business conversion, value, data window, timezone, currency, and
   decision the attribution analysis must support.
3. Inventory every browser, server, platform, analytics, MMP, offline, and app
   attribution source with its identity, counting, deduplication, and privacy rules.
4. Reconcile comparable events and explain differences caused by eligibility,
   view-through rules, consent, modeled data, conversion lag, thresholds, or scope.
5. Separate measurement quality from platform-reported performance.
6. Return findings, contradictions, confidence, missing evidence, and a measurement
   improvement plan through the common JSON contract.

Overview

Do not assume one platform is ground truth, add incompatible reports together, or recommend an attribution model without the operator's decision context.

Manual install (npx, ZIP)

Claude Codenpx skills add AgriciDaniel/claude-ads --skill ads-attribution -g -a claude-code

Claude Code (manual)

mkdir -p ~/.claude/skills && curl -sL https://codeload.github.com/AgriciDaniel/claude-ads/tar.gz/ac2164493391 \
  | tar -xz -C ~/.claude/skills --strip-components=2 claude-ads-ac2164493391/skills/ads-attribution

Windows (PowerShell)

New-Item -ItemType Directory -Force "$HOME\.claude\skills" | Out-Null; Invoke-WebRequest "https://m.agentalley.io/zip/e7dee2898cf0a595941b/ads-attribution.zip" -OutFile "$env:TEMP\ads-attribution.zip"; Expand-Archive "$env:TEMP\ads-attribution.zip" -DestinationPath "$HOME\.claude\skills" -Force

Claude.ai / desktopDownload the ZIP and upload it under Settings → Capabilities → Skills.

Codex, Cursor, other agentsPoint them at SKILL.md. The other files are plain text.

Connects with Agent Alley MCP · agentalley.io/mcp

Files included

SKILL.md1.9 KB

Read the files here

AgriciDaniel/ac216441 file

SKILL.md1.9 KB

SKILL.md

Show the full text33 lines

ads-attribution/SKILL.md33 lines · 1.9 KB · ac21644

PreviewCode

Outline

RawView on GitHub

Attribution Audit

Read the main ads contract and normalized account snapshots.
Declare the business conversion, value, data window, timezone, currency, and decision the attribution analysis must support.
Inventory every browser, server, platform, analytics, MMP, offline, and app attribution source with its identity, counting, deduplication, and privacy rules.
Reconcile comparable events and explain differences caused by eligibility, view-through rules, consent, modeled data, conversion lag, thresholds, or scope.
Separate measurement quality from platform-reported performance.
Return findings, contradictions, confidence, missing evidence, and a measurement improvement plan through the common JSON contract.

Do not assume one platform is ground truth, add incompatible reports together, or recommend an attribution model without the operator's decision context.

Comparability gate

Reject aggregation until the sources share, or are explicitly normalized to, the same conversion event and value definition, attribution window, click/view scope, counting method, deduplication identity, timezone, currency, attribution model, and modeled-data treatment. Until then, report the values side by side with their definitions; do not compute a total.

Example: Meta seven-day conversions and Google thirty-day conversions are incompatible. Refuse to add them, reconcile windows and definitions first, and only aggregate a newly comparable dataset.

1	---
2	name: ads-attribution
3	description: "Audit cross-platform attribution, conversion definitions, reporting windows, GA4, AdServices and AdAttributionKit, MMPs, browser and server events, offline conversions, and platform reconciliation. Use for attribution audit, attribution models, conversion windows, requests to add or total Meta and Google conversions, incompatible reporting-window aggregation, GA4 attribution, MMP review, AppsFlyer, Adjust, Branch, Singular, or cross-platform discrepancies."
4	---
5
6	# Attribution Audit
7
8	1. Read the main `ads` contract and normalized account snapshots.
9	2. Declare the business conversion, value, data window, timezone, currency, and
10	decision the attribution analysis must support.
11	3. Inventory every browser, server, platform, analytics, MMP, offline, and app
12	attribution source with its identity, counting, deduplication, and privacy rules.
13	4. Reconcile comparable events and explain differences caused by eligibility,
14	view-through rules, consent, modeled data, conversion lag, thresholds, or scope.
15	5. Separate measurement quality from platform-reported performance.
16	6. Return findings, contradictions, confidence, missing evidence, and a measurement
17	improvement plan through the common JSON contract.
18
19	Do not assume one platform is ground truth, add incompatible reports together, or
20	recommend an attribution model without the operator's decision context.
21
22	## Comparability gate
23
24	Reject aggregation until the sources share, or are explicitly normalized to, the
25	same conversion event and value definition, attribution window, click/view scope,
26	counting method, deduplication identity, timezone, currency, attribution model,
27	and modeled-data treatment. Until then, report the values side by side with their
28	definitions; do not compute a total.
29
30	Example: Meta seven-day conversions and Google thirty-day conversions are
31	incompatible. Refuse to add them, reconcile windows and definitions first, and
32	only aggregate a newly comparable dataset.
33