Setting UTM Conventions by Channel: Google, Naver, Meta, Kakao, Email
The most common UTM question is "so what values do I put for this channel?" There is no single right answer, but building a shared standard dictionary makes analysis comparable and reports clean. This guide lays out recommended source/medium combinations and practical examples by channel. Use it as a starting point and adapt it into your own team standard.
Why per-channel rules matter
If one person tags a Facebook ad as source=facebook&medium=cpc and another as source=fb&medium=paid, GA4 treats them as completely different traffic. You cannot combine performance and channel classification breaks. The goal of a rule is not to "be correct" but to make everyone use the same value. Use the table below as a starting point, and once set, do not change it.
Recommended combinations by channel
| Channel | utm_source | utm_medium | Notes |
|---|---|---|---|
| Google Search Ads | cpc | Top text ads | |
| Google Display (GDN) | display | Banner network | |
| YouTube Ads | youtube | video or cpc | Video ads |
| Naver Search Ads | naver | cpc | Search top |
| Naver GFA | naver | paid-social or display | Performance display |
| Meta (FB/Instagram) | facebook or meta | paid-social | Paid social |
| Kakao Moment | kakao | paid-social or display | Biz board etc. |
| Kakao channel message | kakao | message | Friend/alim talk |
| Email / newsletter | newsletter or send name | Newsletter | |
| SMS | sms | message | Text message |
| Influencer / affiliate | influencer/media name | referral or affiliate | Sponsored |
| Organic social (non-ad) | instagram etc. | social | Not an ad |
Split paid social (paid-social) from organic social (social). Even on the same Instagram, a paid ad and an ordinary post mean very different things. Splitting them in medium lets you view ad efficiency and content strength separately.
Keep the principle: source is the "who," medium is the "how." That is why a Facebook ad is source=facebook, medium=paid-social — facebook is the medium that sent the traffic, paid-social is the method. Repeating the same value in both loses analytical value.
Whether to use facebook or meta is your team's call. Since Instagram traffic also comes from Meta ads, decide up front: split source into facebook/instagram to see platforms separately, or standardize on meta to view them combined.
Campaign naming rules
If source/medium are "where and how," campaign is "what activity." Names matter more as campaigns pile up, so we recommend a convention that joins fixed components in order. For a date_purpose_product format:
2026q3_sale_running-shoes 2026-08_launch_new-jacket blackfriday_2026_all
Then the GA4 campaign report reads at a glance — timing, purpose, target — and related campaigns group naturally when sorted. The key is to fix the order of components and the separator.
Splitting creatives with content
To learn which creative worked within a campaign, split with utm_content. Mark a top banner, a bottom button, image A and B, or copy variants as banner_top, button_bottom, img_a, img_b, and you can compare clicks and conversions by creative within the same campaign. It is also a simple way to run A/B tests with UTMs alone.
Turn frequent combinations into templates
Once you have per-channel rules, saving frequent combinations as templates — instead of typing them each time — is the surest way to eliminate mistakes. Build "Google Search," "Meta Ads," and "Newsletter" templates, then just change the campaign name each time, and anyone on the team produces consistent values. Easy block UTM saves per-channel block setups as templates you load with one click, which fits this standardization especially well.
Summary
The goal of per-channel UTM rules is not a perfect answer but consistency. Use the table as a starting point for your standard, keep source as the medium and medium as the method, and split paid from organic in medium. Fix the order of campaign-name components and lock frequent combinations into templates, and your reports stay clean no matter how many campaigns you run.