UTM tags look trivial until you try to report on six months of them and find nine spellings of the same campaign. The fix is boring and has to be agreed once.
Every term used above is defined in our experimentation glossary, and three experiments taken apart step by step sit in the case studies. Longer pieces with the workings attached are in the field notes.
Most analytics tools treat Facebook and facebook as two sources. One person typing a capital letter splits a channel in half and nobody notices for a quarter.
Do this: Write the rule down: lowercase, no spaces, hyphens between words. Then enforce it with a builder rather than trust.
Source is where it came from, medium is the kind of traffic. Campaign, content and term describe this particular effort. Mixing those two levels is why reports stop rolling up.
Do this: Keep a fixed list of allowed sources and mediums. Let campaign be free text.
Send traffic and baseline rate. We reply with the smallest effect your test can actually detect. One email, no call.
A UTM on a link inside your own site starts a new session and reattributes the visitor to yourself. It is the single most common way attribution data gets destroyed.
Do this: Search your own site for internal links containing utm_. Remove every one you find.
Any change splits your reporting at the date it happened. That is survivable if you record when and why; it is chaos if you do not.
Do this: Keep a one line log of scheme changes with dates. It costs nothing and saves an argument every year.
This is teaching material and our own reading of standard practice, not advice for your specific site. Check anything important with your own specialist before you act on it.
Traffic, baseline rate and what you changed. We reply with the minimum detectable effect and whether the result meant anything. No cost, no call.