Of the five UTM parameters, medium is the one that decides which channel group a visit falls into. Analytics tools match it against a fixed list of expected values — email, cpc, social, referral, affiliate, display — and anything outside that list lands in unassigned, where it is invisible in every default report.
Source and campaign are labels you choose. Medium is a value you match. Treating it like a free text field is the most common and most expensive tagging error.
What does each parameter do?
Three are load-bearing and two are optional.
| Parameter | Required | What it answers |
|---|---|---|
| utm_source | Yes | Where the click came from: newsletter, google, linkedin |
| utm_medium | Yes | What kind of traffic it is: email, cpc, social |
| utm_campaign | Yes | Which effort it belongs to: spring-sale-2026 |
| utm_content | No | Which variant: header-link, footer-cta |
| utm_term | No | The paid keyword, mostly a legacy field |
Source and medium answer different questions and are frequently confused. LinkedIn is a source; social is the medium. A newsletter is a source; email is the medium. Getting them the wrong way round produces reports that look plausible and group nothing correctly.
Why does case matter?
Because the parameters are case-sensitive and analytics treats Email and email as two separate values. Half your newsletter traffic tagged one way and half the other produces two rows that have to be manually added together forever.
The same applies to spacing and separators. "spring sale" and "spring-sale" are different campaigns; so are "spring-sale" and "spring_sale". None of this is an error at the point of tagging, which is exactly why it accumulates.
Lower-casing everything and picking one separator is the entire fix, and it needs to be a rule rather than an intention.
What does a convention look like?
Four rules, written down where whoever builds the next link will see them.
- Lower case, always. No exceptions, including for brand names.
- Hyphens between words, never spaces or underscores.
- Medium from the fixed list only — email, cpc, social, referral, affiliate, display.
- Campaign names include the period, as in spring-sale-2026, so they stay distinguishable a year later.
The fourth rule is the one people skip and regret. A campaign called "newsletter" is unusable after the second newsletter, and renaming it later does not merge the old data.
Should you tag internal links?
No. Tagging a link between two pages of your own site restarts the session attribution, so a visitor who arrived from an organic search and then clicked a tagged internal banner is recorded as arriving from that banner instead.
The effect is to erase your actual acquisition data and replace it with a record of your own navigation. If you want to measure internal banners, that is what event tracking is for.
The same caution applies to redirects and email footers that point back into the site — anything a visitor can click after they have already arrived.
What about links you do not control?
Anything you hand to a partner, an affiliate or a publisher will be copied, edited and occasionally mangled. Give them the finished URL rather than the parameters to assemble, and keep the campaign name short enough that a typo is obvious rather than plausible.
It is also worth deciding in advance who owns the source value for a partner. If three publishers all tag themselves as "partner", the report tells you that partnerships work and not which partner does — and the data cannot be split retrospectively.
Are the parameters visible to the visitor?
Yes. They sit in the address bar and get copied whenever someone shares the link, which has two consequences worth planning for.
The first is cosmetic and real: a long tagged URL looks like tracking, and some people strip it or decline to share it. The second is that a shared tagged link attributes the new visitor to the original campaign, which inflates it — a phenomenon most visible when a tagged link is posted somewhere public.
Neither is a reason to stop tagging. Both are reasons to keep the parameters short and to treat unusually large campaign numbers with suspicion.
Questions people ask
Does it work with anything other than Google Analytics? UTM parameters are a de facto standard and most analytics platforms read them, though channel-grouping rules differ. The three required parameters are safe everywhere.
What if I only set source? Most tools will record it and leave the medium empty, which usually means the visit is grouped as referral or unassigned depending on the platform.
Can I tag a QR code? Yes, and it is one of the few reliable ways to attribute offline material. Remember that the tagged URL makes the code denser, so the printed size may need to grow.
Is utm_term still worth using? Rarely. Paid platforms usually pass keyword data through their own integrations, and the field survives mainly for manual tagging of search campaigns.
How long do the parameters last? Until the session attribution expires, which is a setting rather than a constant. A visitor who arrives tagged, leaves and returns directly weeks later is usually attributed to the campaign again if the window has not closed — which is worth knowing before reading too much into a long tail.
Match the medium, invent the rest, and write the convention down. The UTM link builder lower-cases by default and warns on non-standard mediums, the QR code generator turns the result into something printable, and the word counter is there for the campaign names that grew too long.