Chapter 24 — Maintaining UTM Data Quality (Auditing & Clean-up)

How often to audit?

Run light checks often and deeper audits on a schedule. This keeps small errors from turning into budget-level mistakes.

Pull the data you actually need

Collect every unique UTM value used in the period. Start small; expand if needed.

• Google Analytics 4: Export an Exploration with session_source, session_medium, session_campaign, and any campaign dimensions you rely on (e.g., utm_term).
• Warehouse/BigQuery: If GA4 is linked, query event- or session-scoped campaign dimensions for the timeframe you’re auditing.
• Marketing automation/email: Export click URLs to confirm email UTMs match your standards.
• UTM registry/spreadsheet: Compare “intended” values with what you observe in analytics and the warehouse.

What to scan for and how to fix it

List distinct values for utm_source, utm_medium, utm_campaign, plus any extras (e.g., utm_id). Compare them to your canonical list.

GA4 won’t rewrite history. Do retro fixes in your BI/warehouse. Handle forward fixes in link builders, ad templates, GTM, and team checklists.

KPIs that prove your clean-up is working

UTM compliance rate: Percent of sessions with valid source, medium, and campaign.

Normalization reduction: Percent drop in unique values after mapping.

Variant drift rate: Count of new, non-standard values since last audit.

Untagged share: Percent of traffic showing as (direct)/(none) but likely attributable.

Time-to-fix: Days from detection to deployed correction.

Error rate by origin: Issue frequency by team/source (email, social, paid, partners).

During launches with PR, paid social, and nurture emails, target an untagged share under 5% and time-to-fix under two days.

The TrackFunnels Clean‑Map‑Patch Loop

A simple framework to keep UTM data tight.

1) Extract unique values - Pull distinct session_source, session_medium, session_campaign for the last 6–12 months.

2) Classify quickly - Bucket into Errors (typos), Variants (case/synonyms), Legacy (retire), Orphans (unrecognized).

3) Quantify impact - Attach sessions, leads, and pipeline to each issue. Fix the biggest leaks first.

4) Clean history in reporting - Build a canonical mapping table in BI/warehouse. Join it to raw fields to standardize history.

5) Patch live sources - Update link templates, ad platform templates, and GTM transforms to stop recurrence.

6) Close the loop - Share a short readout and store before/after screenshots for audit history.

Monitor (not set) and (direct)/(none) before they snowball

Set guardrails and investigate quickly.

• Create alerts when:
• Untagged share rises above a threshold (for example, >8% week over week)
• (not set) sessions jump >20% week over week
• Usual causes to check first:
• Sales- or partner-shared links without UTMs
• Email or chat apps stripping parameters
• New ad placements launched without templates

Archiving keeps names from colliding later

• Retire old campaign names and IDs in your registry to prevent reuse.
• Mark deprecated rows as “Archived” and hide them from picklists.
• Keep an alias map so historical reports still resolve correctly.

This avoids a 2026 field event inheriting a 2024 expo name and polluting trend lines.

Lightweight tools that get the job done

• Excel/Sheets: Pivot each UTM field. Use LOWER(), TRIM(), and find/replace to spot drift.
• Looker Studio/BI: Create calculated fields with LOWER() and CASE mappings.
• Scripts/Warehouse: Schedule a query that flags non-canonical values.
• UTM audit utilities: Tools like DumbData’s UTM Audit surface anomalies fast.

Use any tool that helps you list unique values, normalize, and monitor.

Run a 30‑minute quarterly audit sprint

• 0–5 min: Export last 6 months of session_source/medium/campaign from GA4.
• 5–15 min: In Sheets, pivot distinct values; highlight anything not on your canonical list.
• 15–20 min: Tally sessions and leads tied to each anomaly to size the impact.
• 20–25 min: Update your BI mapping for retro clean-up.
• 25–30 min: Create two tickets: fix the offending template; add a partner/sales kit with pre-tagged links.

What a 30‑minute quarterly sprint usually finds

SQL pattern for a quick scan (warehouse)

Adjust field names to your schema.

SELECT
  LOWER(session_source) AS src,
  LOWER(session_medium) AS med,
  LOWER(session_campaign) AS cmp,
  COUNT(*) AS sessions
  FROM analytics.sessions
  WHERE event_date BETWEEN '2026-01-01' AND '2026-06-30'
  GROUP BY 1,2,3
  ORDER BY sessions DESC;

Follow up with a LEFT JOIN to your canonical mapping to flag non-matches.

Your reusable audit checklist