Getting Started with UTM Drift Guard: Your First Audit

New to UTM Drift Guard? Run your first UTM audit in about 60 seconds: paste a GA4 export, read your A–F grade, and accept a clean taxonomy.

Updated Jun 30, 2026

You have a year of UTM data in GA4 and a quiet suspicion the reports do not add up. The same channel is spelled four ways, a chunk of traffic sits in Unassigned, and you are not sure where to start. UTM Drift Guard is built for exactly that moment: paste your data, get a grade, and see the few clusters that are splitting your real channels — before you spend a weekend on a spreadsheet.

What UTM Drift Guard does#

Most UTM tools are builders: they help you write new links and assume the past is fine. UTM Drift Guard starts from the other end. It audits the utm_source, utm_medium, and utm_campaign values that already reached your property, finds where they drifted apart, and gives you a single grade for how clean the data is. GA4 is case-sensitive — email, Email, and EMAIL are three different mediums — so one channel quietly splits into three rows, and 10–20% of sessions commonly land in Unassigned. The audit surfaces that fragmentation in plain terms, then helps you lock a canonical taxonomy so it stops happening.

CSV is the day-1 path

Direct GA4 connection is on the roadmap, behind a waitlist. For now the fastest route is a CSV or pasted export — see exporting your GA4 source/medium report for the exact columns.

Run your first audit in about 60 seconds#

The whole flow is four steps. The first public scan needs no account, so you can grade your data before deciding anything.

1Paste or uploadA CSV or GA4 source/medium export
2Read your gradeAn A–F health score in ~60s
3Review clustersIssues ranked by traffic impact
4Accept taxonomyOne canonical token per value

The first-audit flow: from raw export to an approved taxonomy, left to right.

1
Paste or upload your data
Open the audit page and drop in a CSV, or paste a GA4 source/medium/campaign export. The scan runs synchronously for up to roughly 500 links, and the first public scan has no signup wall. If you are unsure which report to pull, the GA4 CSV export guide lists the exact dimensions to include.
2
Read your A–F grade
You get a single letter grade with a one-sentence reason — the headline number for how clean your tracking is. A is consistent; an F means source and medium values are largely fragmented. The full scoring logic is in how the audit works.
3
Review the issue clusters
Below the grade, issues are grouped into clusters ranked by traffic impact: case duplicates, source/medium swaps, typos, and un-tagged links sitting in Unassigned. The Anthropic Claude API does the clustering — fb, Facebook, facebook, and meta collapse into one group. It is bounded and reviewable: it proposes, you decide.
4
Accept a canonical taxonomy
Accept the suggested canonical values in one click — one lowercase token per source, medium, and campaign — or override anything that does not match how your team talks. Nothing is pushed to GA4 or your ad platforms; you are approving a map. From there you can route new links through the governed link builder so the drift stops at the source.

Reading your A–F grade#

The grade is deliberately blunt. It is meant to be the one number you can put in front of a teammate or a QBR slide without a 40-tab workbook. A first audit of a year-old account often lands around a C — readable, but leaking sessions to duplicates and Unassigned.

C71 / 100C — a handful of clusters are splitting real channels

A typical first grade: a C means a handful of clusters are splitting real channels — fixable in a few clicks.

What each grade usually means, and the next move it points to.
Grade	What it usually means	Typical next move
A	Values are consistent; almost nothing is fragmenting	Lock the taxonomy and schedule a recurring audit
B	A few small clusters split real channels	Approve the canonical map, re-check next month
C	Several clusters; sessions leaking into Unassigned	Fix the top clusters, route new links through the builder
D	Most channels are forked across spellings	Do a full cleanup pass and document the convention
F	Source / medium values are largely inconsistent	Start from the canonical taxonomy and rebuild link creation

The grade matters most as a trend. Re-run the audit after you apply the canonical map and you can show you moved from a C to a B — and that the channel totals can finally be trusted. If your grade is being dragged down by traffic with no UTM at all, the deeper read on GA4 UTM tracking that is not working covers the root causes.

What the audit reports back#

An A–F health grade with a one-sentence reason you can quote directly.
Issue clusters ranked by traffic impact, so you fix the few that move channel totals first.
A dimension breakdown across source, medium, campaign, and the Unassigned bucket.
A suggested canonical taxonomy — one approved token per value — you accept in one click.
A shareable public report and a branded PDF you can send to channel owners before a review.

You keep the judgment

The audit ranks and suggests; it never renames a channel on its own. Approve only the canonical values your team will actually use, and the map stays yours.

Where to go next#

Once you have a grade and an approved taxonomy, these are the next docs to read, in roughly the order you will need them:

Export GA4 to CSV — the exact report and columns to pull for an accurate audit.
How the audit works — how the grade and the issue clusters are scored.
Taxonomy and canonicalization — how variants collapse into one canonical token.
Governed link builder — lock new links to approved values so drift stops at the source.
Recurring audits — schedule drift detection so the next review starts clean.

Do I need an account to run my first UTM audit?

No. The first public scan has no signup wall — paste or upload a CSV or GA4 source/medium export and you get an A–F grade and the ranked issue clusters in about a minute. You only create an account when you want to save a report, accept a taxonomy, use the governed link builder, or schedule recurring audits.

What file or export do I paste to get started?

A CSV or pasted GA4 source/medium/campaign export works. In GA4, open Reports → Acquisition → Traffic acquisition, add session source, medium, and campaign, set your date range, and export to CSV. The scan runs synchronously for up to roughly 500 links. The GA4 CSV export doc walks through the exact columns.

Will UTM Drift Guard change my UTM data in GA4?

No. The audit never writes back to GA4 or your ad platforms. It reads your data, proposes a variant-to-canonical map, and lets you approve it in one click. Because past GA4 hits are immutable, you apply the approved map where you report — Looker Studio, BigQuery, or a spreadsheet — and stop the drift going forward with the governed link builder.

How does the AI clustering decide which variants are the same channel?

The Anthropic Claude API groups values that mean the same thing — case duplicates like Facebook and facebook, nicknames like fb and meta, and common typos — into a single canonical token. The clustering is bounded and reviewable: it surfaces a suggested grouping ranked by traffic impact, and a human approves each one. It does not act on its own.

What is a good UTM health grade to aim for?

There is no universal pass mark, but a first audit of a year-old account often lands around a C: readable, with a few clusters splitting real channels and 10–20% of sessions in Unassigned. An A means your source, medium, and campaign values are consistent and almost nothing is fragmenting. Treat the grade as a trend — moving from a C to a B is the signal your channel totals can be trusted.

Run your first UTM audit now

Paste a GA4 export or CSV and get an A–F health grade with ranked issue clusters in about a minute. No signup for the first scan.

Run a free audit

Next Export GA4 UTM Data to CSV for an Audit (Exact Steps)