Pixel Auditor assigns every tag (and every site overall) an A–F grade on a 100-point scale. The grade is the single most useful number for "is my tracking healthy?"
Grade reference
| Grade | Range | Meaning |
|---|---|---|
| A | 90–100 | Healthy. No major issues. |
| A− | 85–89 | Minor issues — likely a couple of missing optional params. |
| B | 70–84 | Functional but worth fixing — slow tags, occasional duplicates. |
| C | 55–69 | Real problems — pre-consent fires or missing required params. |
| D | 40–54 | Multiple violations stacked. Tracking is lossy or non-compliant. |
| F | < 40 | Broken. Don't trust this site's analytics until fixed. |
Deductions
| Issue | Deduction |
|---|---|
| Pre-consent fires | −20 (capped) |
| Duplicate installations | −15 per duplicated platform |
| Retired tags (e.g. Universal Analytics) | −10 |
| Dormant or redundant GA4 IDs (configured but never received a hit) | −10 |
| Missing measurement IDs | −5 each, max −15 |
Missing required event params (e.g. purchase without transaction_id) | −10 each, max −20 |
| PII leaks (raw email/phone in payload) | −15 per affected platform |
| Inconsistent Consent Mode state (gcs=G111 but observed denied) | −10 |
Health reasons list
The Audit tab's "why this grade" panel lists every active deduction with the specific tag and event causing it. Example:
Site grade: B (76 / 100)
−20 Pre-consent fires
2 fires before consent_update granted ad_storage:
· gtag page_view (G-9X4K2MZ7QP) at +0.04s
· fbq PageView (247831049502) at +0.12s
−4 Missing optional params
1× GA4 purchase missing 'tax'
How to recover points
- Open the Audit tab.
- Read the "why this grade" list top to bottom — issues are sorted by impact.
- For each deduction, click the linked tag/event row to inspect the specific failure.
- Fix the highest-impact issue first (pre-consent and duplicates usually win).
- Reload the page and re-audit. The grade updates live.
tip
Use Saved Runs to capture the current state before fixing, then load the saved run for a side-by-side diff to confirm the fix worked.