The Tracking Audit Checklist: 12 Checks for Data You Can Trust
A tracking audit verifies that reported marketing data matches reality: that every counted conversion happened, every real conversion is counted, and credit lands on the true cause. This twelve-point checklist runs in an afternoon, groups checks by failure type, and gives a pass condition for each.
- Audit in order: inflation, blindness, distortion.
- The reload test on thank-you pages often exposes the biggest inflation source in two minutes.
- A channel census catches uncounted phone leads, which were 52% of one firm's leads.
- Reconcile platform totals against banked outcomes; explain any gap over 10 to 15%.
Why this matters
Tracking failures in 2026 tend to be invisible. A thank-you page double-fires and results look better than they are; an enhanced-conversions setup loses match quality after a checkout change and still appears to work; phone leads are never recorded. Across the accounts we audit, the average is 8.4 critical issues, and most had been present for months because nothing looked broken. A regular audit is how these problems get caught while they are still cheap to fix.
The three failure types
The checklist is organised around three failure types, audited in order.
| Type | Definition | Why It Ranks Here |
|---|---|---|
| Inflation | Counting conversions that did not happen | Corrupts every downstream decision, so fix first |
| Blindness | Missing conversions that did happen | Hides the campaigns doing real work |
| Distortion | Crediting the wrong cause | Only matters once counts are honest |
The twelve checks
Each check has a pass condition. Budget an afternoon for the first pass; the monthly repeat takes about half an hour.
Inflation
- Reload test. Thank-you pages, several reloads with a debugger open. Pass: one conversion per genuine completion.
- Duplicate-tag sweep. List every tag firing a conversion event. Pass: one owner per action, no legacy tags.
- Test-traffic filter. Internal IPs and staging excluded. Pass: your own team cannot inflate the numbers.
- Bank reconciliation. Platform conversions versus orders or booked work. Pass: gaps under 10 to 15%, each larger gap explained. This exposed the roughly 40% phantom inflation in our electrical contractor rebuild.
Blindness
- Channel census. List every way a customer can convert. Pass: each channel tracked or consciously excluded. The most expensive blindness is the phone; one firm found 52% of leads uncounted, in the call tracking case study.
- Consent gate check. Defaults fire before tags, all four parameters update, grant rates known. Pass: you can state your grant rate. See the Consent Mode fix guide.
- Recovery-stack check. Enhanced conversions healthy; Meta running Pixel plus Conversions API, deduplicated, EMQ above 7. Pass: both platforms' health metrics green and baselined.
- Journey walk. Complete one real conversion per path, desktop and mobile. Pass: every step exists in the data, including cross-domain hops.
Distortion
- Definition inventory. Every action mapped to a real outcome. Pass: nothing the bidding optimises to is unexplained.
- Attribution audit. Model and windows match the sales cycle and are re-checked after updates. Pass: the setting is a decision, not a default.
- Outcome loop. Off-site outcomes flow back as imported conversions. Pass: bidding optimises to revenue-shaped events.
- Single source of truth. One document names the arbitrating number. Pass: everyone answers "what did marketing produce" from the same place.
After the audit
Expect findings; clean accounts are rare. The species guide the response. Inflation fixes make performance look worse and be better, so warn stakeholders before the chart dips. Blindness fixes reveal performance that was always there, and budget should follow it. Distortion fixes change allocation, which is where the money moves. Phone-led businesses should weight the channel census and outcome loop most heavily, the focus of our home services work, where an uncounted call is uncounted revenue. For the full system this checklist maintains, see the measurement guide.
Frequently Asked Questions
A full audit twice a year, with a light monthly reconciliation between: platform conversions against banked outcomes, plus the diagnostics platforms expose. The cadence matters more than the depth, because silent failures are cheap to catch monthly and expensive to find annually.
Reload your thank-you page several times with a tag debugger open. If the conversion fires on each load, every refresh and revisit is being counted. In one account this accounted for about 40% of reported results.
Record your baseline when the implementation is fresh, then watch for decay rather than an absolute number. A match rate drifting down over months usually means the user data being hashed has degraded, often after a form or checkout change.
Most of it, yes. Tag Assistant, the Google Ads conversions summary, GA4 DebugView, and Meta Events Manager are point-and-click, and reconciliation against orders is a spreadsheet task. Engineering help is mainly needed for the fixes.
Ready to fix what’s costing you conversions?
We’ll review your acquisition funnel, show you exactly what’s underperforming, and hand you a clear, prioritised plan, whether or not you choose to work with us.
Get a clear diagnosis of your acquisition performance.
- 30-minute review
- Free, with no obligation
- Clear findings and priorities
- You keep the findings
30 minutes · Free · No obligation · You keep the findings