What this Skill helps with

Cross-Platform Attribution Auditor checks whether web, app, advertising-platform, CRM, consent, MMP, and server-side measurements describe the same conversion journey. It documents mismatches and the evidence needed to resolve them.

Who this Skill is for

It is intended for analytics teams, performance marketers, mobile-growth teams, agencies, and technical owners reviewing attribution across several systems.

What you provide

  • Event and conversion definitions for each platform
  • Attribution windows, models, time zones, and reporting periods
  • Browser, server, app, MMP, analytics, and CRM counts
  • Deduplication keys and event identifiers where available
  • Consent behavior, identity rules, and known tracking changes
  • Sample journeys or test-event evidence

What it produces

  • A source-by-source measurement map
  • Definition, window, identity, and time-zone mismatches
  • Duplicate and missing-event risks
  • Cross-device and app-to-web gaps
  • A reconciliation table with likely explanations
  • A ranked validation and repair plan

How the Skill works

The workflow maps how each system records and credits an event, then compares counts only where scope and definitions match. It separates expected differences from possible implementation faults and lists the evidence required for confirmation.

Installation and first use

  1. Download and add the Skill to Claude.
  2. State the conversion journey and systems in scope.
  3. Attach event dictionaries, settings, and matching-period exports.
  4. Ask for definition and scope checks before count comparison.
  5. Validate suspected faults with test events and technical owners.

Example workflow and expected output

Provide Meta, Google Ads, GA4, server-event, and CRM counts for the same lead journey. Ask for a table covering event definition, window, source, duplicates, missing stages, expected variance, and the next validation step.

Limits and review guidance

Different platforms may report different valid numbers because their attribution rules differ. The Skill cannot inspect network requests, consent state, or server logs unless that evidence is supplied. Do not treat reconciliation as causal measurement.

Frequently asked questions

Should all platform counts match exactly?

Often they should not. The audit explains expected differences before flagging possible faults.

Can it detect duplicate conversions?

It can identify duplicate risk from supplied identifiers and counts, but test-event or log evidence is needed for confirmation.

Does it change tracking code?

No. It produces a validation and repair plan for technical review.