Why "Facebook" and "facebook" are two rows in GA4 — and how to fix it

The problem, in one screenshot's worth of pain

You open GA4's Traffic acquisition report and your single biggest channel is scattered:

That's one channel wearing four costumes. No single row tells you what Facebook actually drove, and any chart that groups by source is quietly wrong.

Why GA4 does this

It's not a bug. UTM values are just strings passed in the URL, and analytics tools store them verbatim. Two things make casing especially dangerous:

• Values are case-sensitive. F and f are different characters, so Facebook ≠ facebook. GA4 groups on the exact string.
• Every entry point is a chance to diverge. An ad platform auto-tag, a teammate's manual link, and a link from an email tool can each capitalize differently — and nothing reconciles them.

Case sensitivity is the single most common cause of fragmented channel data. The good news: it's also the easiest to eliminate, because the fix is one rule — lowercase everything.

How to fix the data you already have

1. Find every variant

In GA4, look at Source and Medium dimensions and note every casing variant of the same real channel (Facebook, facebook, FB, etc.). Do the same for mediums (Paid-Social vs paid-social).

2. Consolidate for reporting

You generally can't rewrite historical hits in GA4, but you can consolidate going forward and in analysis:

• Use Explorations and group variants into a single segment or custom channel group for clean reporting.
• If you route data to BigQuery, normalize with LOWER(source) in your queries so variants collapse into one.
• Fix the links still in circulation — emails, bios, ads — so new sessions stop adding to the mess.

3. Audit your live links

Export or paste the URLs currently running and check them against a single lowercase standard. An audit tells you how many are off-spec and which are one edit away from correct. (This is exactly what Tagsmith's audit does — it scores your back-catalog and suggests fixes.)

How to stop it happening again

Cleaning up once is satisfying; doing it every quarter is not. The permanent fix is to remove human casing decisions from the process entirely:

• Force lowercase automatically on every UTM value at build time. If people can't type a capital letter into a saved link, the whole class of problem disappears.
• Constrain sources and mediums to an allow-list so FB can't sneak in alongside facebook.
• Enforce it where links are made — in the builder and the browser — not in a style guide people have to remember.

As one analytics lead we talked to put it: "Lowercase-forcing alone kills a whole class of my problems." It's a small rule with an outsized payoff — and it's the foundation of a proper UTM naming convention and UTM governance more broadly.