How utm_id works
Every other link tag is written for humans. Source, medium and campaign are meant to be read in a report. utm_id is written for machines: it carries a fixed code that identifies the campaign, and unlike a campaign name it is not supposed to change, be tidied up or be improved by whoever builds the next link.
Analytics stores it as a campaign identifier alongside the campaign name. On its own it does nothing visible — it does not affect channel grouping, it does not change how a session is classified, and a report built without it looks exactly the same. Its value appears only when something else needs to be matched to that campaign.
Why utm_id matters
Spend lives in one system and sessions live in another. To say what a campaign cost per enquiry, the two have to be joined, and a join needs a reliable key. Campaign names make poor keys because they are typed by people: one side abbreviates, the other side adds a month, and the rows fail to match for reasons nobody notices until the numbers look wrong.
An identifier removes that fragility. Import a cost file keyed on the same code the links carried, and every row lands where it belongs regardless of how the campaign was described in either system. The same argument applies when analytics data is pulled into a warehouse or a dashboard and joined to anything else.
It is most useful precisely where there is no automatic integration — a local ad network, a sponsorship, a printed poster with a code on it. Those channels never report their own spend into analytics, so a manual import is the only route, and an identifier is what makes that import trustworthy.
Common mistakes with utm_id
The first is adding it without ever importing anything. If no cost file, warehouse or dashboard is going to use the code, it makes URLs longer and buys nothing.
The second is treating it as a second campaign name and writing a description in it. The third is generating a fresh identifier per link or per creative, which turns a campaign key into a link key and breaks the very join it was meant to support.
The fourth is changing it midway. An identifier that changes is not an identifier, and the imported spend will match one half of the sessions and silently miss the other.
How to act on it
Use it only when you have a real reason to join data, and prefer the advertising platform’s own campaign identifier where one exists, so both sides of the join already agree without any translation. Where you are inventing codes yourself — for offline activity or a smaller network — keep them short, keep them unique, and record which code belongs to which campaign name in the same sheet where you register campaigns.
Then test the join with a small file before trusting a full month of imported spend. Getting the key right at the start is part of a sound GA4 setup, and it is far cheaper than reconciling two systems that were never designed to agree.
