A UTM naming convention for music campaigns

Campaign parameters on a URL are the only thing that tells you which of your channels delivered a click. Without them, an analytics tool lumps everything it cannot identify into direct traffic, and direct traffic is a synonym for do not know.

Adding them takes seconds. Agreeing what they mean takes one conversation and saves a year of ambiguity.

The five parameters

utm_source is where the click came from. A named place: instagram, tiktok, newsletter, google, meta, spotify, bandsintown, the specific publication.

utm_medium is what kind of placement it was. Not the platform again: paid-social, organic-social, email, display, search, bio, qr, press.

utm_campaign is the thing you are promoting. The release, the tour, the announcement. This should be identical across every link for that campaign, or it will not group.

utm_content distinguishes variants within a campaign: which creative, which post, which button.

utm_term is for keywords in search buying. Most music campaigns leave it empty.

The rules that stop it degrading

Lower case, always. Analytics tools treat Instagram and instagram as two sources, and once both exist in a report the numbers are split forever.

Hyphens, not spaces or underscores. Spaces get encoded and turn into unreadable rows.

A fixed vocabulary for source and medium. Write down the allowed values. If someone invents facebook-ads as a source when meta already exists as source and paid-social as medium, you now have two overlapping ways to count the same thing.

Campaign names that a stranger can read in a year. A convention like artist-release-2026-03 beats newsingle, and beats campaign2.

Never reuse a campaign name. Two releases sharing a name merges two datasets permanently.

No personal data, ever. No names, no email addresses, no user identifiers. Parameters end up in analytics stores, server logs and shared reports, and they are effectively impossible to remove afterwards. This is a rule with legal weight, not a style preference.

A template that works

Keep one document, with the vocabulary and a worked example per channel.

https://yourlink.example/track
  ?utm_source=instagram
  &utm_medium=organic-social
  &utm_campaign=aster-bloom-single-2026-08
  &utm_content=story-swipe-a

Then generate links from it rather than typing them. Typed links are where the casing errors and the invented sources come from, and a spreadsheet with a formula is enough tooling.

What tagging makes possible

Comparing channels on the same outcome. Newsletter against paid social against bio link, measured on arrivals and on click-through to a service, not on impressions.

Separating owned from paid. The most common surprise is how much of the response was already yours. A campaign that looks like it worked sometimes turns out to be mostly newsletter traffic, which changes the budget conversation.

Reading press properly. Tag the link you give each outlet and you learn which coverage actually delivered, rather than which coverage felt biggest.

Post-campaign forensics. Six months later, when somebody asks what worked on the last release, the answer is a query rather than an argument.

Where tagging stops

Parameters survive as far as your own page. They do not follow the visitor onward into a music service, and no service reports back what happened. So the tagged data ends at the click, and everything after that is co-movement rather than attribution.

That is still a lot. Knowing which channel produced the intent is most of the marketing question, and it is the part you can actually prove.

Retrofitting

If you have a year of untagged links, do not try to fix history. Start the convention on the next campaign, write it down, and accept that the previous year is one bucket.

The second campaign under a convention is where it starts paying, because that is the first time you can compare like with like.