Name creatives so a report makes sense

A name typed into a plan item travels onto the change request and then onto the campaign. A name given to a creative travels with it into every report that groups anything.

That makes naming a measurement decision taken at production time, which is why it gets made badly: the person naming the file is thinking about the file, and the person who needs the name is six weeks away and does not exist yet.

The consequence is specific. A report can only group by what is in the name. A dimension you did not encode is a dimension you cannot filter on without opening every asset individually, and by then nobody will.

The concept. What this asset is about, in a token you would recognise in a year. Not the file's working title, not the client's nickname — the thing a stranger reading a report would need in order to know what they are looking at.

The variable under test. If this asset exists as one side of a comparison, the variable is the most valuable segment in the name. It is what lets a report answer which kind of opening works rather than which asset won.

The format. Because formats are not comparable with each other and a report that mixes them silently is a report producing averages across incomparable things.

The date it entered. Not the date the file was made — the date it started running. This is what makes a series readable when the same concept is refreshed, and it is what lets you answer how long the last asset lasted the next time something fatigues.

Four segments, one separator, chosen once. Pick a separator that never appears inside a segment, because most report tooling splits on a character and a separator that also appears in a value turns a clean split into a fragile one.

The naming generator applies a shape consistently, which is most of the value — the scheme matters less than the fact that everybody uses the same one.

Who made it. Initials and team names are production metadata: useful during production and noise forever after. They also change, which means a name encoding them stops being true.

The version number of the file. Third revisions of an asset are a production concern. What matters downstream is whether this is a different asset, and if it is, it gets its own name and its own date.

Anything time-relative. New, current, latest, Q1. All of them are false within months and all of them fragment a report — three assets called new is three assets you cannot tell apart.

Anything you would not filter on. The honest test for every segment. If it would never appear in a filter, it is taking space a segment you will use could have had.

A campaign, the groups inside it and the assets inside those each carry a name, and they should agree with each other. A campaign named for a job and a group named for a date is a report where two levels cannot be joined by anything except by hand.

The simplest rule that works: each level encodes what varies at that level and inherits what does not. The campaign carries the job and the flight, the group carries the audience or the split, the asset carries the concept, the variable and the format. Repeating the campaign's name inside every asset name is how a name gets too long to read and still tells you nothing new.

Alignment with the plan matters too. A plan item's name is carried straight onto the draft, so the name typed into a plan is the campaign name in the report — which makes the plan the right place to make the decision rather than a place it gets improvised.

The naming convention template is the scheme as a document, which matters mostly because a convention held in one person's head has three synonyms in it by the end of the quarter.

A name change is a change request like anything else: drafted, checked against the guardrails, and applied when an authority clears it.

The arrangement, stated once: your connections read your accounts, a change is written as a draft on the record with the state it would change from and the checks it passed, applying it is gated on approval — by you, or by a rule you wrote — and no platform write reaches a customer account until that platform has granted write access. Applying a cleared change to a platform is in private beta.

The mechanical cost is small. The real cost is that renaming does not retrofit the reports you already ran, so a rename splits a series: the same asset appears under two names either side of a date, and anything grouping by name treats them as two things.

So the practical rule is the same as for tagging conventions. Do not rewrite history. Pick a date, apply the scheme from that date, and write the date down wherever the numbers are quoted. A series with a documented break is readable; one that silently changed meaning is not.

Hand a report to somebody who has never seen the account and ask them to find every asset testing a particular idea, and then to tell you which of them is the most recent.

If they can do both from the names alone, the scheme works. If they need the account open beside them, the scheme is a filing system for whoever built it rather than a convention.

Run a creative test that answers something is why the variable segment earns its place, and Brief a creative for paid social is where the concept and the variable get decided in the first place.