An attribution audit is a structured review of conversion tracking, campaign data, attribution settings, and revenue reporting. Its purpose is to explain why platforms such as Google Ads, Meta, GA4, CRM systems, and attribution tools report different performance numbers.
The objective is not to make every system show identical results. Different tools use different attribution models, conversion windows, identity rules, and reporting dates. A useful audit identifies which differences are expected, which indicate a technical problem, and which data source should support each business decision.
1. Define What Each Conversion Means
Start by documenting the conversion events used across every reporting system. A platform may count form submissions, while the CRM reports qualified leads or closed deals. Comparing those totals directly creates an apparent discrepancy even when both systems are working correctly.
Create a simple conversion map:
| Conversion stage | Definition | Primary system |
| Lead | Valid form submission or enquiry | Website analytics |
| Qualified lead | Lead accepted using agreed criteria | CRM |
| Opportunity | Qualified lead with pipeline value | CRM |
| Customer | Completed purchase or closed deal | CRM or ecommerce platform |
| Revenue | Confirmed transaction value | Finance, CRM, or ecommerce platform |
Marketing, sales, and finance should agree on these definitions before the audit continues.
2. Test the Conversion Tracking Setup
Review whether each event fires at the correct moment. A lead event should fire after a successful form submission, not when someone clicks the submit button or opens the form.
Check for duplicate events caused by thank-you page refreshes, repeated form submissions, browser history, duplicate tag installations, or multiple CRM synchronizations. Missing events can be caused by consent settings, script errors, redirect behavior, or tags that do not load on every required page.
The audit should record the expected trigger, the actual trigger, the value passed, and whether the event can fire more than once. The basic principles of conversion tracking provide the foundation for this part of the review.
3. Review Campaign and UTM Data
Inconsistent campaign parameters can divide one campaign across several reporting rows. For example, paid_social, paid-social, and social_paid may all represent the same channel but appear separately in analytics.
Review source, medium, campaign, content, and term values across active campaigns. Check for missing parameters, inconsistent capitalization, unclear abbreviations, and links copied from older campaigns.
A documented UTM governance system helps prevent these discrepancies by establishing consistent naming rules before campaigns launch.
4. Compare Attribution Models and Windows
Platforms may report different conversion totals because they assign credit differently.
Google Ads may credit a conversion after an advertisement interaction, while GA4 may use a different attribution model across the wider journey. A CRM may record only the source attached to the lead or opportunity. The systems are not necessarily reporting the same question.
During the audit, document:
| Setting | What to compare |
| Attribution model | First-click, last-click, multi-touch, or data-driven |
| Click window | Number of days after a click that can receive credit |
| View window | Whether advertisement views receive credit |
| Conversion date | Interaction date or actual conversion date |
| Cross-device rules | Whether activity can be matched between devices |
| Direct traffic treatment | Whether direct visits replace or preserve earlier sources |
A difference caused by attribution logic usually requires explanation rather than a technical fix.
5. Check CRM and Offline Conversion Matching
For B2B, SaaS, local services, and other sales-led businesses, website conversions may occur weeks before the final revenue outcome. CRM delays, incomplete source fields, and failed lead matching can create large reporting gaps.
Check whether campaign information is preserved when leads enter the CRM and whether it remains attached as the lead progresses into an opportunity or customer. Review how duplicates, merged records, reopened opportunities, and delayed sales updates are handled.
Offline conversions should also use stable identifiers and clearly defined lifecycle events. If marketing data cannot be matched with CRM outcomes, reporting may stop at lead volume rather than pipeline and revenue.
6. Investigate Traffic and Identity Gaps
Not every customer journey can be reconstructed completely. Consent choices, cookie restrictions, cross-device behavior, private browsing, offline interactions, and shared devices can all break the connection between touchpoints.
Invalid traffic can also add clicks or sessions that do not represent genuine customer interest. These limitations should be documented so stakeholders understand why attribution totals may not equal actual customer totals.
The audit should distinguish between data that can be repaired and data that is fundamentally unavailable.
7. Classify Every Discrepancy
Each issue should be assigned to one of three categories:
| Category | Example | Required action |
| Expected difference | Different attribution windows | Document and explain |
| Tracking problem | Duplicate form events | Fix and retest |
| Data limitation | Unidentified cross-device journey | Record the limitation |
This classification prevents teams from spending time trying to eliminate normal platform differences while serious tracking errors remain unresolved.
8. Create a Reporting Source of Truth
The final audit output should state which system is trusted for each metric. Advertising platforms may remain useful for campaign delivery and bidding, analytics for website behavior, CRM for pipeline, and finance or ecommerce systems for confirmed revenue.
Reliable attribution reporting brings these sources together without pretending they measure the same thing.
A successful attribution audit does not create perfect data. It creates documented definitions, tested events, understood discrepancies, and a reporting structure that marketing, sales, and finance can use consistently.