Chapter 10: Digital Analytics Methodologies and Problem-Solving

Data quality audit checklist

An imperfect data point can still be useful, but only if its limitations are documented. Use this grid to audit the reliability of your collection according to the ISO/IEC 25012:2008 (SQuaRE Data Quality Model) standard.

Quality audit progress 0%
01. Accuracy (Accuracy)
Tolerable conversion gaps
Do transactions or leads in the analytical tool deviate by less than 10% from the CRM or billing system?
1 - Large discrepancies (+20%) 3 - 10% average discrepancies 5 - Écarts parfaits (<5%)
Level 1
Net Isolated Revenue
Are taxes and shipping fees excluded from net sales calculations to avoid distorting profitability?
1 - Everything confused 3 - Partial (excluding taxes) 5 - Perfect and automated exclusion
Level 1
Duplicate transaction filtering
Are transaction IDs verified to prevent double counting when refreshing a confirmation page?
1 - No filtering 3 - Fragile client-side JS filtering 5 - Robust server-side deduplication
Level 1
02. Completeness (Completeness)
Plausible engagement rate
Is the bounce rate or engagement rate realistic (no automated events triggered at 0s that skew engagement)?
1 - Chiffres faussés (rebond <5%) 3 - A few suspicious events 5 - Verified trigger rigor
Level 1
Complete form tracking
Do we not only measure successful submissions, but also validation errors (client journey frictions)?
1 - No error tracking 3 - Basic tracking (button click) 5 - Precise error events (field/type)
Level 1
Quantification of consent impact
Has the volume of lost data due to CMP cookie rejections been estimated and communicated to analysts?
1 - Totally ignored 3 - Vague estimates 5 - Rigorous loss modeling
Level 1
03. Consistency (Consistency)
UTM normalization
Do campaigns use a strict, documented, and normalized UTM naming convention in lowercase?
1 - Disorder (CamelCase, spaces) 3 - Unfollowed oral rules 5 - Strict GTM filters and nomenclature
Level 1
Inter-platform naming convention
Are event and parameter names (Data Layer) identical between the website, mobile app, and server-side?
1 - Different names everywhere 3 - Partially aligned 5 - Shared unified dictionary
Level 1
Common definition of conversions
Do teams use the same calculation method for conversion rates (e.g., session vs user)?
1 - Definition conflict 3 - Informal alignment 5 - Documented reference definition
Level 1
04. Traceability and privacy
Active data dictionary
Is there a living documentation listing each tag, its business purpose, trigger, and owner?
1 - No written record 3 - Outdated Excel document 5 - Continuously updated dictionary
Level 1
Minimization of personal data
Have we validated that no personal identifiable information (PII such as email, IP) is transmitted by URL or parameters to ad exchanges?
1 - IP address and email visible 3 - IP masked but URL risk 5 - Complete cleaning (GTM redaction)
Level 1
Strict Conditional Trigger
Are paid marketing tags blocked until the user has explicitly consented to the cookie banner?
1 - Tags launched by default 3 - A few pixel leaks 5 - Consent Mode v2 strict implemented
Level 1

Risk Level

Risk index: Very High

References and academic sources