A UTM taxonomy that survives contact with reality

What is a UTM taxonomy, and why does it break?

UTM parameters are query-string tags such as ?utm_source=linkedin&utm_medium=paid_social&utm_campaign=2026q4_demand_webinar-ai-search_emea that tell your analytics where a visit came from. A UTM taxonomy is the written set of rules for which values are allowed in each parameter, how campaign names are built, and who maintains the lists.

Taxonomies break for predictable reasons. Different people build links in different tools. Agencies bring their own conventions. Someone types LinkedIn once and linkedin the next time. A scheduling tool appends its own medium. Within a year, the same channel appears under a dozen spellings and the cross-channel dashboard shows a growing "Unassigned" bar. The fix is not a longer document; it is fewer free-text decisions.

Which UTM parameters should you use?

GA4 supports more than the classic five parameters. utm_source, utm_medium and utm_campaign populate session source, medium and campaign in the Traffic acquisition report; utm_id, utm_term, utm_content and utm_source_platform are also supported, while utm_creative_format and utm_marketing_tactic are listed but not currently reported in GA4 properties¹Source 1 · Google Analytics Help[GA4] URL builders: Collect campaign data with custom URLssupport.google.com.

Controlled vocabularies for source and medium

Medium is the most important parameter for channel reporting, because GA4’s default channel group rules mostly key off it. Paid Social requires a social source and a medium matching ^(.*cp.*|ppc|retargeting|paid.*)$; Organic Social accepts mediums such as social, social-network, social-media and sm; Email matches email, e-mail, e_mail or e mail in source or medium; Display matches display, banner, expandable, interstitial or cpm²Source 2 · Google Analytics Help[GA4] Default channel groupsupport.google.com. Choose medium values that land in the right group without custom rules.

Keep the source list just as tight: platform names in lowercase with no suffixes (linkedin, not linkedin.com or LinkedIn-Ads), newsletter names as newsletter-{name}, and partners by short domain. Store both lists in one place that every link tool reads from.

How should you name campaigns?

Use a fixed pattern with delimited slots, so a campaign name can be parsed back into its parts in SQL. Use underscores between slots and hyphens inside a slot. We recommend four slots:

Example: 2026q4_demand_webinar-ai-search_emea. Because the slots are fixed, a single SQL expression can split it, and your dashboard can group spend and outcomes by objective or market across every platform. Use the same string as the campaign name inside each ad platform; that is what lets spend join to sessions and revenue later.

12345678select
    utm_campaign,
    split_part(utm_campaign, '_', 1) as period,
    split_part(utm_campaign, '_', 2) as objective,
    split_part(utm_campaign, '_', 3) as initiative,
    split_part(utm_campaign, '_', 4) as market,
    utm_campaign ~ '^([0-9]{4}(q[1-4]|-[0-9]{2})|evergreen)_(awareness|demand|retention|launch|hiring)_[a-z0-9-]+_[a-z-]+$' as is_valid
from campaigns_seen

Platform-specific setups

The fewer links people tag by hand, the fewer errors you get. Each major platform has a way to automate tags.

1. 1
   Step 1:

   Google Ads: rely on auto-tagging

   Auto-tagging adds a gclid parameter to ad clicks. In GA4, auto-tagging is prioritized over manual tagging with no override option³Source 3 · Google Ads HelpAbout auto-taggingsupport.google.com, so you do not need UTMs on Google Ads links. Make sure auto-tagging is on and the accounts are linked.

2. 2
   Step 2:

   Meta, LinkedIn, TikTok: use URL parameter fields and dynamic values

   Put tags in each platform’s URL parameters (tracking) field rather than in the destination URL, and use dynamic values where offered, for example Meta’s {{campaign.name}} and {{ad.name}}. Then make the platform campaign name follow your pattern, and the tag follows automatically.

3. 3
   Step 3:

   Email platforms: set defaults once

   Mailchimp and Brevo can append UTM parameters to every link automatically. Set source to newsletter-{name}, medium to email, and the campaign to your pattern in the template, not per send.

4. 4
   Step 4:

   Social schedulers and organic posts

   Configure the scheduler’s UTM template with utm_medium=social and the platform as source. For profile and bio links, use a permanent campaign such as evergreen_awareness_profile-link_global.

5. 5
   Step 5:

   Offline and QR codes

   Use utm_medium=qr and a short redirect you control, so you can fix a tag after the brochure is printed.

Governance: making the right link the easy link

Governance fails when it depends on memory. It works when the compliant path is faster than the non-compliant one.

Agencies and freelancers get the vocabulary file and link builder on day one, and their contract includes the convention. Reports from external partners use your campaign names, not theirs.

Cleanup rules for the mistakes that still happen

Even with good governance, some links will be wrong, and historical data predates your convention. Fix it in a cleanup layer in your data model rather than editing raw data, so every fix is visible and reversible.

123456789101112131415161718with normalized as (
    select
        session_id,
        date,
        lower(trim(utm_source))   as src,
        lower(trim(utm_medium))   as med,
        lower(trim(utm_campaign)) as campaign
    from web_sessions_raw
)
select
    n.session_id,
    n.date,
    coalesce(sa.canonical, n.src) as utm_source,
    coalesce(ma.canonical, n.med) as utm_medium,
    n.campaign                    as utm_campaign
from normalized n
left join source_aliases sa on sa.alias = n.src   -- e.g. 'linkedin.com', 'lnkd.in' -> 'linkedin'
left join medium_aliases ma on ma.alias = n.med   -- e.g. 'paidsocial', 'paid-social', 'cpc_social' -> 'paid_social'

Keep the alias tables small and reviewed. Each new alias is a signal that the link builder or a platform template needs fixing at the source. If a large share of sessions shows (not set) for source or medium, check the tagging basics first: Google documents causes such as a missing session_start event when the tag is not triggered on initialization, unlinked Google Ads accounts, disabled auto-tagging and incomplete UTM parameters⁴Source 4 · Google Analytics Help[GA4] What the value (not set) means in your reportssupport.google.com.

How do you audit UTMs every month?

Run one query on the first business day of each month and send the result to campaign owners. It lists every value outside the vocabulary, ranked by sessions, so people fix what matters.

1234567891011121314151617select
    utm_source,
    utm_medium,
    utm_campaign,
    count(*) as sessions,
    case
        when utm_medium not in (select value from allowed_mediums) then 'medium not allowed'
        when utm_source not in (select value from allowed_sources) then 'source not allowed'
        when utm_campaign !~ '^([0-9]{4}(q[1-4]|-[0-9]{2})|evergreen)_[a-z]+_[a-z0-9-]+_[a-z-]+$' then 'campaign pattern'
        else 'ok'
    end as issue
from sessions_clean
where date >= date_trunc('month', current_date) - interval '1 month'
  and date < date_trunc('month', current_date)
group by 1, 2, 3
having count(*) >= 20
order by issue <> 'ok' desc, sessions desc

In Kimo you can save this as a model and attach an alert that fires when non-conforming sessions exceed 5% of the month. The same model feeds the Unmapped row on the Marketing command center, so the problem is visible where people already look.

Worked example: one campaign across five channels

Suppose your team launches a webinar on AI search in Europe during the fourth quarter. The campaign name is decided once, in the campaign registry: 2026q4_demand_webinar-ai-search_emea. Every channel then reuses that exact string, and only source, medium and content change.

A month later, the dashboard can answer questions that are impossible with ad hoc tags: how many registrations the webinar produced per channel, what each registration cost on paid channels, how organic posts compared with paid ones, and whether EMEA demand campaigns as a group are improving quarter over quarter. Because the objective and market are parseable slots, the same query works for every future campaign without editing.

How the tags connect spend to revenue

Three joins make cross-channel reporting work. Spend joins to sessions on the campaign name and date. Sessions join to sign-ups or leads on the session or user identifier your product captures at conversion. Leads join to opportunities or orders in your CRM or shop. If any link in that chain uses a different spelling, the chain breaks and the revenue falls into Unassigned. This is why a taxonomy is a revenue-reporting prerequisite, not a tidiness project. Our post on one marketing dashboard shows the finished view, and the command center in the Kimo Marketing app renders it on simulated data.

Common mistakes and how to avoid them

One more subtle point: GA4 also keeps the traffic source that first acquired a user, and those values do not change after acquisition, even if the user later arrives through new campaigns⁵Source 5 · Google Analytics Help[GA4] BigQuery Export schemasupport.google.com. Decide whether a report should use first-user or session attribution and label it clearly. A clean taxonomy is what makes both views trustworthy, and it is the backbone of a unified marketing dashboard and of any attribution model you adopt later.