What are UTM parameters?

They are tags that whoever builds the link adds voluntarily in order to describe where it came from. The acronym stands for Urchin Tracking Module: Urchin was the web analytics tool Google agreed to buy in March 2005, and the utm_ prefix survived the change of owner and remains in use today.

They are worth separating from other pieces they get confused with. A cookie stores information in the visitor's browser; a UTM parameter travels inside the address itself and stops existing as soon as that person moves to another page. A tracking pixel loads a resource from an external server; a UTM loads nothing. And the referrer is sent by the browser automatically, whereas the value of a UTM parameter is decided by a human when writing the link.

That last difference explains both its strength and its fragility. A parameter says what the person who wrote it meant to say, no more and no less. If someone types utm_medium=email in one campaign and utm_medium=Email in the next, the tool counts two different channels, because the values are case sensitive. Parameters describe traffic, they do not identify people: they hold no name, no email address and no customer ID, unless somebody puts one there by accident.

A UTM parameter lives in the query string of the URL, the part that starts with a question mark. The format is key=value and the pairs are chained with &, so a tagged address looks like https://example.com/offer?utm_source=newsletter&utm_medium=email&utm_campaign=summer-sale.

When somebody clicks, the browser requests the page using the full address. The server returns the same content it would return without the tags, because they play no part in routing. The measurement script then reads the query string in the browser and classifies the visit: it assigns source, medium and campaign to the session, and that classification holds for as long as the session lasts.

The official catalogue currently holds nine parameters, but GA4 reports on only seven of them. The five historical ones are utm_source, utm_medium, utm_campaign, utm_term and utm_content. They were joined by utm_id (campaign ID) and utm_source_platform (the platform directing the traffic), both of which do reach the reports. The other two, utm_creative_format (creative format) and utm_marketing_tactic (targeting criteria), are accepted in the address and lost along the way: checked in August 2026, neither of them appears in any GA4 report. Anyone who builds a campaign naming scheme on those two will hunt for the data for weeks and never find it.

There is also a precedence rule that decides which value ends up in the report, and it catches many people out. If an address carries Google Ads auto-tagging plus hand-written UTM parameters, the auto-tagged values win: they are what fills the source, medium and traffic-classification dimensions, and the hand-written ones are left out of that part of the report.

The decision that hangs on them is where the next euro goes. Without tags, a large share of the traffic from newsletters, social networks, QR codes and email signatures lands in catch-all buckets such as "direct" or "referral", and there is nothing to compare in there. With a consistent scheme you can answer which send brought orders, which banner brought only visits and which partnership turned out expensive.

There is a second decision, less obvious, and it sits on the SEO side. Every combination of parameters produces a different address for the same content. Google picks one representative URL among the duplicates, and the canonical tag the site declares is a hint for that choice rather than an order. If internal links, the sitemap or external publications point at tagged versions, the search engine receives contradictory signals about which address is the right one.

The usual outcome is not a penalty. It is noise in the indexing reports, addresses with a campaign tail competing against each other and, every so often, a tagged version surfacing in search results with its campaign trail on display. Anyone who measures without watching this point ends up with clean analytics and a dirty index.

• Write every value in lower case and settle on a closed dictionary of mediums (email, social, cpc, affiliate) before the first campaign goes out.
• Always fill in utm_source, utm_medium and utm_campaign; add utm_id when you want to join campaign cost with the data from the buying platform.
• Declare a canonical tag without parameters on every page, check that the sitemap contains clean addresses only, and make sure press releases and social posts point at the clean version too.
• Record the naming scheme in a shared sheet or generate the links with a builder, so two people do not invent two names for the same campaign.
• Check in Search Console whether addresses containing utm_ appear in the indexed pages report: that is where you see whether the search engine is treating a tagged version as a page in its own right.
• When the link goes into print or into a video, shorten it through a redirector you control that preserves the parameters at the destination.

• Building the naming scheme on utm_creative_format or utm_marketing_tactic. They are official parameters, but GA4 does not report on them, so the value is sent and never shows up on screen.
• Adding hand-written parameters to Google Ads adverts while auto-tagging is on, and expecting the manual values to win. They do not.
• Tagging internal links. A UTM parameter in a menu or in a banner on your own site resets attribution mid-session and wipes out the real origin of the visit.
• Mixing upper case, accents and spaces in the values. utm_campaign=Summer Sale and utm_campaign=summer-sale count as two separate campaigns.
• Putting personal data into a parameter, such as an email address or a customer ID. That information ends up in the analytics, in the server logs and in every tool that receives the full URL.