Every performance team says it has a naming convention. Almost none of them can run a report on it a year later. The convention existed, it was documented, and it quietly stopped being true somewhere around the third freelancer.
This is worth fixing, because the ad name is where most teams try to store the one thing the platform cannot report on: what the creative actually is. Platforms give you spend by campaign for free. They do not give you results by angle, hook or format unless you put that information somewhere. The name field is where people put it.
What a good convention looks like
A convention is a small schema written into a text field. The shape that survives is boring:
- A fixed order of fields separated by one delimiter. An underscore is easiest because it splits cleanly in a spreadsheet.
- A controlled vocabulary for each field. "UGC" and "ugc" and "creator" must not all mean the same thing.
- A concept ID, not a description. Descriptions get reworded. IDs can be joined back to a brief.
- A version suffix, so that a re-cut opening is v2 of the same concept and the lineage is not lost.
A workable example:
| Field | Example values | Why it earns its place |
|---|---|---|
| Concept ID | C0142 | Joins the ad to the brief and the shoot |
| Angle | Problem, Proof, Offer | The thing you can brief again |
| Format | Video, Static, Carousel | Separates format effects from message effects |
| Hook | Question, Claim, POV | Lets you test openings across concepts |
| Source | Owned, Partner | Shows how much spend rests on borrowed credibility |
| Version | v1, v2 | Keeps iterations attached to their parent |
That gives names such as C0142_Proof_Video_Question_Owned_v2. Keep the field count small. Each extra field is another chance for someone to skip it.
Why conventions decay
None of these failures is carelessness. They are what a free text field does when many hands touch it.
| What happens | What it breaks |
|---|---|
| A freelancer joins and uses their own labels | The same angle appears under three spellings and splits into three small samples |
| An agency hands the account over | Two eras of naming sit in the same table and cannot be grouped together |
| Twelve variants ship on a Friday | Fields get skipped, copied from the wrong ad, or left as "new ad copy 2" |
| An ad is duplicated for a new audience | The duplicate keeps the old name even though something changed |
| The schema grows a field and old ads are not updated | Half the history is missing a column |
The deeper problem is that a name records what the person who made it intended. It cannot record what is on screen. If someone labels an ad "Founder" and the video opens on a product shot, the name is wrong and the report will be confidently wrong with it.
Creative is where the largest share of the outcome sits. NCSolutions' analysis of nearly 450 CPG campaigns, reported by MarketingCharts, put it at 49%, ahead of brand, reach, targeting and recency. It is CPG and mostly offline, so treat it as direction. But it is why a reporting stack that cannot group by creative attribute is missing the biggest lever, and why messy names cost more than they look. See Creative is 49% of the outcome.
Keep it alive
Audit it, do not trust it
Export the last 90 days of ad names. In a spreadsheet, split each name on the underscore and check how many produce the right number of fields with allowed values. In Google Sheets, =SPLIT(A2,"_") does the splitting and a count of blanks shows the damage. The share of ads that parse cleanly is your real compliance rate, and it is usually lower than anyone expects.
Make the right name the easy name
Give people a template to copy rather than a document to read. A shared sheet that builds the name from drop downs removes most spelling drift.
Put one person in charge of the vocabulary
Not of the names, only the allowed values. New angles get added deliberately instead of invented on the fly.
Decide what you will do when it fails
It will fail. A convention is a bet that people will be consistent under deadline pressure. The question is what your reporting does on the day the bet loses.
Knack reads attributes from the creative itself, the actual frames, rather than from the ad name, so the analysis survives an inherited account, a rushed Friday and a freelancer with their own labels. Each ad gets structured fields for who it speaks to, what motivates them, format, angle, hook, offer, talent, production, pacing, tone and call to action, plus media and whether it is owned or a partnership from the ad object. Owned versus partnership comes from the account the ad was published from, so it exists before anyone tags anything. A good naming convention still helps you, because it holds the intent and the concept ID, and Knack reads it too: label the parts of your names once and each becomes a dimension you can slice patterns by, next to the tags. It stops being the only thing standing between you and a usable report. How Knack reads a creative covers the detail.
Book a demo to see this run on your own account.
The owned versus partner field is worth the trouble
If you can only afford one attribute in the name, make it the source of the asset. The debate about whether UGC beats brand made video goes nowhere without it, and so does any question about how much of your spend depends on someone else's credibility.
Sources
- NCSolutions sales driver analysis of nearly 450 CPG campaigns: creative 49%, brand 21%, reach 14%, targeting 11%, recency 5%. Reported by MarketingCharts.
- The schema and audit method are working practice rather than published research.