Measurement Health Framework
The Measurement Health Framework is our methodology for verifying and monitoring analytics setups: a structured system of 40+ checks, refined through years of production measurement work for brands like Watsons, Teknosa and Vakko, and built on 15+ years of experience in digital. It answers one question continuously: can these numbers be trusted?
Why a framework
Measurement does not fail loudly. It fails as a duplicate purchase event, a payment gateway hijacking attribution, a consent update silently dropping a third of your events. Each looks small; together they decide whether your ROAS, LTV and budget decisions are grounded in reality. Ad-hoc checking catches these late or never. A framework catches them systematically.
The five layers
Every audit and every day of ongoing monitoring runs the same five layers:
Tracking implementation. GA4 and GTM configuration, dataLayer integrity, event and parameter coverage against the tracking plan, web and app.
Data quality. Duplicate transactions, revenue mismatch against backend truth, silent drops on key events like purchase, add_to_cart and begin_checkout, volume anomalies.
Attribution integrity. Payment gateway referral pollution, direct and (not set) inflation, cross-domain breaks, UTM hygiene, channel definition consistency.
Consent and compliance. Consent Mode configuration, estimated data loss from privacy choices, cookie behavior.
Infrastructure. BigQuery export completeness and schema health, MMP configuration, SKAN readiness where apps are involved.
Checklist plus automation
The framework has two halves. A structured expert checklist covers what needs human judgment: configuration choices, naming logic, tracking plan fit. Automated BigQuery checks cover what humans miss: they query your raw event data daily for duplicates, gaps, mismatches and anomalies, at a depth interface reports cannot reach. In audits, both halves run once and produce a severity-ranked findings report. In measurement partnerships, the automated half runs every day and we investigate whatever it flags.
What it catches, concretely
Real patterns this framework has caught in production: purchase events firing twice and inflating revenue by low single digits for months. Payment gateway redirects claiming credit for conversions that paid channels earned. A consent banner update cutting event volume by a third with no one noticing for weeks. A GA4 BigQuery export silently missing a day of data. App and web revenue definitions drifting apart until reports disagreed by double digits.
How to get it
The framework runs in two forms. A Measurement Audit applies it once to your entire stack and hands you the findings. A Measurement Partnership keeps it running continuously, with our team investigating and fixing what it flags. Most clients start with the audit.