A lot of marketers — myself included, at first — tag their Meta (Facebook / Instagram) ad URLs with utm_source=meta. It feels natural: the company is called Meta now, right?

But if you open Google's officially distributed GA4 Default Channel Group Source Categories spreadsheet and search through all 819 entries, you'll find something strange:

• facebook is there
• fb is there
• facebook.com is there
• instagram is there
• meta is not

That one missing line isn't a trivia item. It's the difference between Meta ad revenue showing up in your Paid Social channel versus quietly falling into Referral or (other).

This post is a practical, receipts-first walkthrough of what GA4 actually checks, why meta breaks classification, and the URL format I'd recommend fixing on going forward. No opinion — just the file Google itself publishes.

TL;DR

1. Set utm_source=facebook (lowercase, fixed). Not meta, not Facebook, not fb.
2. Set utm_medium=cpc. GA4's Paid Social regex matches on .*cp.*|ppc|retargeting|paid.*. cpc is the safest pick.
3. The 80% of bad-data incidents I see are casing and variant drift: facebook / Facebook / fb / meta splitting one campaign across four rows.

1. What GA4 actually checks when deciding "Paid Social"

Per Google's public documentation on default channel groups[1], the Paid Social classification requires both of these to be true (for manual UTM tagging):

• Source matches a regex list of social sites
• Medium matches the regex ^(.*cp.*|ppc|retargeting|paid.*)$

That "regex list of social sites" isn't a mystery. Google actually publishes it as a downloadable spreadsheet[2]. It's an 819-row file mapping source strings to categories like SOURCE_CATEGORY_SOCIAL, SOURCE_CATEGORY_SEARCH, and so on.

Here's what it contains for the major Meta-related entries:

meta as a string does not appear anywhere in Google's classification file. Traffic tagged utm_source=meta is evaluated against the social list, misses every entry, and falls through to Referral or (other).

"But the company rebranded to Meta" — true, and completely irrelevant to the classification engine. Google Analytics did not update its source file when the corporate rebrand happened. The file still references the long-standing industry-standard string facebook.

Brand names and classification tokens are independent layers. Confusing them costs money.

2. Four failure modes I keep seeing

Failure 1: utm_source=meta silently erases Paid Social

Since meta doesn't match the social list, these sessions land in Referral with source/medium meta / cpc. From the monthly report you'll see "Paid Social revenue dropped." Nothing dropped. The classification box moved.

Failure 2: Case sensitivity splits one campaign into two rows

GA4 stores utm_source values case-sensitively. facebook and Facebook are two distinct dimension values and two distinct rows in your report. One campaign, two boxes, each showing half the revenue.

Failure 3: utm_medium=social promotes paid traffic to Organic Social

GA4's Organic Social rule[1] is: Source matches social list OR Medium is one of social | social-network | social-media | sm. That's an OR. If you use utm_source=facebook with utm_medium=social, the medium matches Organic Social before the logic ever reaches "but wait, this is paid." Your paid budget gets counted as organic, and ROAS looks worse than reality.

Failure 4: Dynamic parameter macros confirmed late

Meta's own [URL Dynamic Parameter feature][3] lets you embed macros like {{campaign.name}} / {{adset.name}} / {{ad.name}} in destination URLs, resolved at delivery time. Useful, but: at insertion time the preview still shows the raw macros, so skipping the post-launch GA4 check means you only discover empty utm_campaign values or unexpected characters (spaces, Japanese text, uppercase) once real money has shipped.

3. The URL format I'd fix on

Given the constraints above, this is what I tag Meta ads with:

utm_source=facebook
utm_medium=cpc
utm_campaign={{campaign.name}}
utm_content={{ad.name}}
utm_term={{adset.name}}
utm_id={{campaign.id}}

Why each value

• utm_source=facebook — matches SOURCE_CATEGORY_SOCIAL in GA4's file, lowercase for casing stability, the industry-canonical value. fb works too (same category), but pick one and stop the drift
• utm_medium=cpc — matches Paid Social's regex, does not collide with Organic Social's social* list. Safest minimum overlap
• utm_campaign={{campaign.name}} — delivered-time substitution; verify values populate correctly once campaigns are live
• utm_content={{ad.name}} / utm_term={{adset.name}} — gives you ad-level and ad-set-level drill-down inside GA4's campaign dimension, without needing custom dimensions

"What about Instagram-only?"

If you want Instagram campaigns to appear as their own row, switch utm_source from facebook to instagram. This is a reporting trade-off: you gain Instagram visibility, you lose the "all Meta ad revenue in one row" view. The cleaner pattern: keep utm_source=facebook (platform-level) and put placement detail in a separate placement={{placement}} custom-dimension — your Meta vs Instagram split survives even when campaign naming changes.

4. Verifying after you ship

Two checks that catch 90% of breakage:

1. Pre-launch URL preview — in Meta Ads Manager, check the URL preview panel on the ad edit screen. Confirm macros resolve to expected values and no unexpected whitespace / CJK / uppercase slipped in
2. Post-launch GA4 realtime — within 30 minutes of first impression, check Realtime → Traffic source (or classic Acquisition). You should see facebook / cpc. If you see meta / cpc or (direct) / (none), the URL didn't deliver what you expected

Skipping these two is the single biggest reason "our attribution looks weird this month" reports exist.

5. One-time cleanup check (10 minutes)

If you inherited a GA4 property and don't know what's been tagged, this is a fast audit:

1. GA4 → Reports → Acquisition → Traffic acquisition
2. Switch dimension to Session source / medium
3. Search for facebook
4. If you see multiple rows among facebook / Facebook / fb / meta — your campaigns are fragmented
5. Standardize going forward, and (if your stack allows) normalize historical data downstream

If meta / cpc exists as an independent row with meaningful volume, your Paid Social number is currently understated by exactly that amount.

Closing

The "correct" utm_source for Meta ads isn't a matter of opinion, agency convention, or vendor recommendation. It's whatever Google's classification file says it is — and Google publishes that file.

Opening an 819-row spreadsheet and searching for meta is maybe three minutes of work. It's strange how often the answer to a recurring "which is right?" argument is sitting in a publicly downloadable CSV.

I'm building RevenueScope, a revenue-first analytics layer that sits next to GA4 and normalizes UTM drift (e.g. auto-merges facebook / Facebook / fb / meta into one Paid Social row) so you can read channel-level revenue, AOV, and RPS from a single dashboard.

This post originally appeared in Japanese on the RevenueScope blog. Canonical source link is set accordingly.

References

1. Google Analytics Help — Default channel group
2. Google — GA4 Default Channel Group Source Categories (official downloadable spreadsheet, linked from [1])
3. Meta Business Help Center — URL Dynamic Parameters specification