Conversion attribution audit
A conversion attribution audit explains why your ad platforms, GA4, call tracking, and CRM report different numbers, and establishes which number to use for which decision. Kodelytics quantifies each gap, identifies the ones caused by defects rather than by design, and delivers a reconciliation methodology your team can run monthly.
Four systems counting the same business event will produce four numbers. Some of that variance is structural — different attribution windows, different models, click-date versus conversion-date — and is not a bug. Some of it is a defect: an event firing twice, a location parameter that never populated, a call tracking platform whose conversions never reach the ad platform. Until the two categories are separated, nobody can trust any of the four, and budget decisions get made on whichever number is closest to hand.
The organizational cost is worse than the analytical one. When four numbers exist and none is authoritative, reporting becomes a negotiation, and the person who explains the discrepancy most confidently wins the budget argument. That is not a measurement culture; it is rhetoric with charts.
An attribution gap is the difference between conversions a system records and conversions it can attribute to a source — for example, conversion events that fire without the location or campaign parameter needed to assign them. The conversions are real; the ability to credit them is missing.
What's actually wrong
These are the symptoms buyers of this service recognize before they can name the problem.
- Monthly reporting requires a conversation about which conversion number is being used.
- Google Ads reports materially more conversions than the CRM records leads.
- Per-location performance can't be reconstructed because location isn't a parameter on the conversion event.
- Call conversions are counted twice: once by the call platform, once by a click-to-call event.
- The reported cost per acquisition and the finance team's cost per customer differ by a multiple.
- Nobody can say what attribution window or model any of the reports use.
What the engagement includes
- A full inventory of every system that counts a conversion, and what it counts.
- Gap quantification: the numeric difference between each pair of systems, over a defined period.
- Separation of structural variance from defects, with the cause named per line.
- Event-level tracing of the highest-value conversion paths, end to end.
- Multi-location attribution analysis — whether location, service line, and campaign can be resolved on each conversion.
- A documented reconciliation methodology: which source is authoritative for which decision, and how the monthly comparison is run.
- A remediation plan for the defects, with effort estimates.
- A definitions register so a metric name means one thing across every report.
Structural variance versus defects
Structural variance is designed-in and permanent: different attribution models, click-date versus conversion-date, session-scoped versus click-scoped counting, platform-specific lookback windows. It should be quantified, written down, and then ignored.
Defects are fixable and usually few: duplicate event firing, broken cross-domain linkage, unverified primary conversion actions, missing parameters, and call tracking counted twice. The audit's core output is the line between the two, because everything downstream — reporting, bidding, budget allocation — depends on knowing which gap is worth engineering effort.
The reconciliation methodology
The higher-value half of the deliverable is not the defect list; it's the monthly routine that follows. One page: the four counts, the expected variance between each pair, the threshold at which a variance becomes an investigation, and who runs the check.
Run monthly, it converts an argument into a known quantity. A gap you predicted is a footnote in a client report. A gap discovered live on a client call is a credibility problem that takes two quarters to repair.
Which number to use for which decision
| Decision | Authoritative source | Why |
|---|---|---|
| In-platform bidding | Ad platform conversions | Smart Bidding optimizes on what it can see |
| Channel comparison | GA4 key events | Only source that sees all channels |
| Revenue and CPA to finance | CRM or booking system | The only source tied to money |
| Call volume | Call platform, deduplicated | Platform events double count |
| Per-location budget | Warehouse, on one definition | Requires joined, parameterized data |
How it's scoped and priced
Delivered as a fixed-fee engagement, one to three weeks depending on the number of systems and locations. The audit is diagnostic; implementation of the fixes is scoped separately against the plan, and frequently handed to the client's own team.
For organizations that keep re-litigating the same discrepancy every quarter, the reconciliation methodology is the higher-value half of the deliverable — it makes the gap a known quantity rather than a recurring argument.
Kodelytics does not publish rates. Every engagement is quoted after a discovery call, because the same service name covers materially different amounts of work.Ask for a quote.
What you get at the end
- A reconciliation model showing each system's count and the explained variance between them.
- A conversion path trace for each primary conversion action.
- A defect register with causes and effort estimates.
- A written reconciliation methodology for monthly use.
- A metric definitions register.
Questions
Why don't my Google Ads and GA4 conversion numbers match?
Because they are designed to count differently. Google Ads credits the conversion to the date and campaign of the click and applies its own attribution model; GA4 credits the session in which it happened, on its own model and window. A stable gap of a few percent is normal. A large, moving, or directionally odd gap usually indicates duplicate firing, broken cross-domain linkage, or an unverified conversion action.
What is an attribution gap?
An attribution gap is the difference between the conversions a system records and the conversions it can attribute to a source — for example, conversion events that fire without the location or campaign parameter needed to assign them. The conversions are real; the ability to credit them is missing, which makes per-location or per-campaign performance unmeasurable.
Which number should we report to the client?
One authoritative source per decision, stated in the report. Platform numbers for in-platform optimization, CRM or booking-system numbers for revenue reporting, and a documented expected variance between them. The failure mode is not the variance; it's an unexplained variance.
Do you need access to our CRM?
Read access, or an export covering the period under review. Without the CRM side there's no way to test whether platform conversions correspond to real leads.
How is this different from a Google Ads audit?
The Ads audit looks at how one account is built and where spend is wasted. The attribution audit looks across systems at whether the conversion data any of them report can be trusted.