# Reconcile Paid Conversion Attribution

Platform reports can overstate conversions when several channels claim the same order. This skill matches conversion exports to an order-ID truth set and produces a normalized reconciliation view.

## Install

```bash
npx skillstore add aaron-he-zhu/attribution-reconciler
```

## Metadata

- Status: approved
- Slug: aaron-he-zhu-attribution-reconciler
- Version: 19.0.0
- Author version: 19.0.0
- Skillstore revision: r2
- Version status: valid
- Tree hash: 425911264f6bc82f06b8d0d59adb4bd22d4612c9e0ec9e1775fdc1d566c491d7
- Author: aaron-he-zhu
- GitHub username: aaron-he-zhu
- License: Apache-2.0
- Repository: https://github.com/aaron-he-zhu/aaron-marketing-skills/tree/main/ad/scale/attribution-reconciler
- Ref: adcb3549b15782055d0beb6d29f113d20de08f92
- Supported tools: Claude, Codex, Claude Code
- Audit status: complete
- Agent install advisory: allowed
- Manual install advisory: allowed
- Artifact signature: available
- Audit attestation: unavailable
- Human verification: not\_verified
- Risk factors: external\_commands, network, filesystem
- Quality score: 78
- Quality tier: bronze
- Public page: https://skillstore.pages.dev/skills/aaron-he-zhu-attribution-reconciler
- Manifest: https://skillstore.pages.dev/api/skills/aaron-he-zhu-attribution-reconciler/manifest

## Capabilities

- Matches platform-reported conversions to GA4 or ecommerce order IDs.
- Separates matched, double-counted, and unmatched conversion claims.
- Normalizes attribution windows and monetary values to a stated common basis.
- Compares credit allocation across at least two attribution models.
- Reads geo or holdout results when a suitable export is supplied.
- Prepares de-duplicated counts for a separate ROI and ROAS calculation.

## Use Cases

- Audit overlapping platform claims: Compare Meta and Google conversion exports with a GA4 order export to identify sales claimed by multiple platforms.
- Build a monthly reconciliation: Create a recurring workbook that tracks matched, double-counted, and unmatched conversions across paid channels.
- Interpret a holdout test: Compare test and control order exports with platform spend to distinguish measured lift from last-click credit.

## Prompt Templates

### Start a basic reconciliation

```
Reconcile these Meta and Google conversion exports against this GA4 order-ID export for May. Show matched, double-counted, and unmatched orders.
```

### Normalize platform reporting

```
Build a monthly attribution reconciliation. Normalize Meta 7-day-click and Google 30-day windows to the order dates, then state the currency basis used.
```

### Compare credit allocation

```
Using this reconciled order set, compare last-click and linear attribution by platform. Keep the total number of real orders unchanged.
```

### Read a geo holdout

```
Use these test-region and control-region order exports and spend data to assess incrementality. Compare measured lift with last-click claims and mark unsupported conclusions.
```

## Limitations

- Requires a GA4 or ecommerce order-ID export as the truth set.
- Uses only the accuracy and completeness of the exports supplied.
- Cannot infer incrementality without a geo or holdout test export.
- Does not calculate ROAS, CPA, ROI, or other ratio metrics.

## Best Practices

- Provide complete order IDs, timestamps, values, currencies, and attribution windows for every export.
- Use the ecommerce or GA4 order export as the truth set, not platform-reported totals.
- Review unmatched claims before using reconciled counts in budget decisions.

## Anti Patterns

- Do not add platform conversion totals together without checking overlapping order IDs.
- Do not compare counts from different attribution windows or currencies without normalization.
- Do not claim incrementality when no valid geo or holdout evidence exists.

## Security Audit

- Audited at: 2026-07-26T09:55:42.326\+00:00
- Summary: All 35 static findings are false positives caused by Markdown code formatting, relative documentation links, or a public homepage URL; no command execution, network request, path traversal, or system reconnaissance is instructed. The skill handles user-provided marketing exports and can persist order-level reconciliation results after confirmation, so data retention controls should be documented.

## Stats

- Views: 1
- Downloads: 7
- Favorites: 0
- Popularity score: 0