UTM names that survive three agencies (a convention that sticks)

Open your Monday channel report and count the rows. If you see facebook, Facebook, FB, fb, facebook.com, meta, and Facebook_Ads, you do not have a traffic problem. You have seven split rows describing the exact same ad channel, and every ROAS calculation built on them is broken.

This does not happen because people are careless. It happens because three agencies, two contractors, and an email tool each built URLs without automated guardrails. A UTM parameter convention written in a slide deck decays within weeks. To keep your channel numbers reliable, you need a strict convention backed by an automated validator.

Why UTM conventions quietly fall apart

The documentation is rarely the problem. The failure is that nothing between the document and the report checks the tags. Every durable UTM system has an automated validator; every fragile one relies entirely on human memory.

The four rules of clean link naming

Everything else is optional detail. These four rules determine whether your tracking survives an agency transition.

Rule 1: source and medium are a closed set

utm_source and utm_medium are the two fields your analytics tool relies on to group channels. They must come from a fixed, short list that only changes through a deliberate team decision. Free-form values belong in campaign or content tags.

Medium is the traffic type (such as paid_social), while source is the platform where the visitor clicked (such as meta). Setting utm_medium=facebook reverses this logic and causes GA4 to quietly dump your visits into "Unassigned".

Rule 2: lowercase, underscores, and no spaces

UTM values are case-sensitive across Google Analytics, BigQuery, and SQL warehouses. Newsletter and newsletter remain two separate rows forever.

• Lowercase everything. Mobile keyboards automatically capitalise the first letter, which causes most case inconsistencies.
• Use underscores, never spaces. Spaces convert to %20 or + depending on the browser, splitting one campaign into multiple records.
• Avoid punctuation. Characters like ampersands break URL query parameters and truncate the remainder of your tracking string.
• Pick underscores or hyphens and stick to one. Mixing both causes summer_sale and summer-sale to coexist in your reports.

Rule 3: give utm_campaign a strict grammar

Campaign names combine multiple dimensions into a single string. Give them a predictable structure:

<period>_<initiative>_<funnel_stage>
2026q3_diwali_prospecting
2026q3_diwali_retargeting
2026q4_newyear_prospecting

A fixed order makes your campaign names parseable in SQL. With a consistent grammar, SPLIT(utm_campaign, '_')[OFFSET(2)] lets you isolate funnel stage across every campaign without writing brittle CASE WHEN logic.

Rule 4: never put UTM tags on internal links

Placing UTM tags on banners, headers, or buttons inside your own website starts a brand new session in Google Analytics. A customer who arrived from a paid Meta ad and then clicks an internal homepage banner has their conversion reassigned to that banner. The paid campaign loses credit, and internal links appear artificially profitable.

Use an internal tracking parameter like int_promo=homepage_hero instead, and ensure your analytics configuration ignores it for session attribution.

Catch bad tags with a weekly validator query

Do not rely on people reading a PDF manual. Provide a simple spreadsheet generator with dropdowns for link creation, and run an automated query each week to detect drift before bad tags contaminate historical reports.

Catching drift at 40 sessions allows an agency to fix a tag immediately. Catching it at 40,000 sessions leaves you with corrupted reporting tables that require manual data repairs.

Clean messy history with a normalisation view

You cannot change links that customers clicked last month. Instead, clean the data at read time using a database view so reports always see unified channels.

Point all dashboards and attribution models at this view. When a new tag variant appears, add one WHEN condition here rather than reprocessing millions of historical rows.

How to run an agency handover that lasts

Agency transitions are where naming conventions face their toughest test. Build these requirements directly into the onboarding process:

Setting up the third level takes less than a day of engineering, and it ensures your marketing data stays clean regardless of agency turnover.

Frequently asked questions

Are UTM parameter values case-sensitive?

Yes. While parameter keys like utm_source are treated case-insensitively by many systems, the values assigned to them (Meta vs meta) are strictly case-sensitive in GA4, BigQuery, and SQL databases. Always enforce lowercase characters.

Should I put manual UTM tags on Google Ads?

In most cases, no. Google Ads auto-tagging with gclid passes far richer data into GA4 without risk of manual typos. Adding conflicting manual UTM parameters can override auto-tagging and obscure auction-level signals. Reserve manual UTMs for channels without native IDs, like newsletters, affiliates, and influencer links.

How should we handle historical data that is already inconsistent?

Do not rewrite historical raw records. Build a database normalisation view using CASE statements to standardise old values when querying. This preserves raw audit logs while giving your analytics dashboards clean, unified metrics.

Can an agency introduce new traffic sources on their own?

They can request an addition, but it must be approved by your analytics lead. An uncontrolled source list degrades into separate rows for every minor variation within months.

Related: how channel CAC depends on clean taxonomy, and what to do when ad spend and orders share no join key.