Attribution &
UTM Taxonomy

How Attribution Works

Attribution determines which touchpoints in a user's journey receive credit for a conversion. Getting it wrong is one of the most common reasons analytics data misleads business decisions.

GA4 uses an event-based data model and assigns credit across the touchpoints a user encounters before completing a conversion event. The model you select in Admin → Attribution Settings applies globally to all conversion reporting.

The Attribution Chain

Attribution Models

GA4 supports two categories of attribution model. The choice affects which channels appear to perform well — and therefore where budget gets allocated.

Rule-Based Models

Data-Driven Attribution

Key Variables to Watch

Lookback Windows

GA4 defaults to 30 days for most conversions and 90 days for purchase events. If your sales cycle exceeds these defaults, you will systematically undercount early-funnel touchpoints.

Set lookback windows in Admin → Attribution Settings. B2B funnels commonly need 60–90 days for non-purchase conversions.

Channel Groupings Are Fragile

GA4's default channel definitions rely entirely on UTM parameters being applied correctly. A paid campaign without utm_medium=cpc will be misclassified as organic or direct. This is the single most common source of attribution errors.

Cross-Domain Tracking

When users traverse multiple domains (main site → subdomain checkout → third-party cart), sessions fragment and attribution breaks unless:

Direct Traffic Inflation

GA4 has a "last non-direct click" fallback — if the last touchpoint is direct, it looks back for a prior campaign touchpoint. But when UTMs are missing or broken, direct absorbs credit it doesn't deserve.

Model Switching Creates Discontinuities

You can compare models in Advertising → Attribution without changing anything. But switching the active model in Attribution Settings retroactively changes how conversions are counted across all reports. Set it once, early, and document the date if you ever change it.

Engaged Sessions vs. Sessions

GA4 attributes using engaged sessions as the unit, not raw sessions. An engaged session requires one of: 10+ seconds of engagement, a conversion event, or 2+ page views. This changes how multi-touch paths surface in funnel analysis compared to Universal Analytics.

Getting It Right

Before Launch

At Implementation

Ongoing

UTM Taxonomy Overview

A UTM taxonomy is the naming convention system governing how you tag URLs across every campaign and channel. It is the foundation everything else rests on.

If it's inconsistent, channel groupings break, attribution lies, and reports become noise. The taxonomy is not a technical concern — it is a governance concern.

What a Taxonomy Controls

The Five Parameters

utm_source and utm_medium are the two parameters GA4 uses to classify channels. Getting those wrong misfires your entire channel report. The other three are for campaign and creative reporting.

Channel Grouping Rules

GA4 maps source/medium combinations to channels using built-in logic. These are the exact values that trigger each channel classification. Any deviation pushes traffic to Unassigned or Direct.

Building a Naming Convention

A naming convention has three non-negotiable properties:

Recommended Structure

utm_source   = [platform]
               google | meta | linkedin | klaviyo | newsletter

utm_medium   = [channel type]
               cpc | display | paid_social | email | organic_social

utm_campaign = [objective]_[audience]_[date]
               awareness_cold-audience_2026q1
               retarget_cart-abandoners_2026q2
               nurture_trial-users_ongoing

utm_content  = [creative]_[variant]
               hero-v1 | carousel-blue | cta-learn-more

utm_term     = [keyword or segment]
               enterprise-analytics | dmo-retarget

Campaign Naming Pattern

The campaign name pattern is where teams diverge the most. A consistent structure of [objective]_[audience]_[date] lets you filter and aggregate in GA4 and BigQuery without regex gymnastics.

-- Good examples
awareness_cold-audience_2026q1
retarget_cart-abandoners_2026q1
nurture_trial-users_ongoing
brand_all_2026q2

-- Bad examples (do not use)
Q1 Brand Campaign       -- spaces, mixed case
Meta_Retargeting         -- no audience or date
AWARENESS-2026           -- uppercase

Enforcement Strategies

The naming convention means nothing without enforcement. The convention and the tool are one artifact — not two separate things.

Auditing a Broken Setup

When inheriting a messy UTM setup, use this sequence to assess damage quickly before attempting any remediation.

BigQuery Diagnostic Query

SELECT
  (SELECT value.string_value
   FROM UNNEST(event_params)
   WHERE key = 'utm_source') AS utm_source,

  (SELECT value.string_value
   FROM UNNEST(event_params)
   WHERE key = 'utm_medium') AS utm_medium,

  COUNT(*) AS sessions

FROM `your_project.analytics_XXXXXXXX.events_*`
WHERE _TABLE_SUFFIX BETWEEN '20260101' AND '20260131'
  AND event_name = 'session_start'
GROUP BY 1, 2
ORDER BY 3 DESC
LIMIT 100

Run this monthly. The source/medium distribution in raw events is your ground truth — it shows you what GA4 is actually receiving before any channel classification logic is applied.

Approved Values

Mediums — GA4 Channel Mapped

Do NOT Use as utm_medium

These values break GA4 channel grouping. Traffic tagged with them will land in Unassigned.

Approved Sources by Category

Do & Don't Rules

Always Do

Never Do