Annex A · Written
Creative, marketing and media
The one worked annex, showing what the core disclosure set looks like in agencies, brands, studios and publishers, the sector where AI landed first and hardest.
Why this sector first
Marketing is the canary rather than the subject. Output is priced by volume, engagement is measurable to four decimal places, and the entry tier was fragile before any of this started. What happens here arrives in law, medicine, consulting, code and journalism a few years behind.
It is also the sector the articles are written from, which is the only reason it was the one to be written first. It illustrates the standard without being the standard itself.
The six conditions, here
- Hours: the time a person had for the brief, not the number of assets they shipped against it.
- Training: hours spent on craft (writing, art direction, strategy) rather than on learning this quarter’s generation tool.
- Junior headcount: whether studios and agencies are still hiring the juniors whose first jobs are the ones being automated.
- Human authorship: who made the asset, disclosed on the work and never assessed.
- Rework: the assets scrapped after the sprint, which are the true price of the speed.
- Exposure: position three is the audience, and the annex renders the condition as assets per reviewer per year (below).
Exposure in an agency: assets per reviewer per year
Assets per reviewer per year. In an agency, oversight authority is removed by throughput rather than by policy. A reviewer signing off tens of thousands of assets a year across multiple brands, against regulatory duties, brand safety rules and brand guidelines, cannot meaningfully discharge any of them. The metric traces directly to liability without authority in the core, which is what qualifies it as an annex metric under the annex rules.
Position three in an agency is the audience, and the harm class that populates is errors acted upon at low severity, such as a misleading claim or a product shown doing something it does not do. Adverse decisions sit near zero, and that is a legitimate result.
The table
Ten metrics this sector counts today, and what it would have to start counting instead. Two have been withdrawn for breaking Principle One; they stay on the table, struck through.
| Column AThe Growth LedgerWhat we count today | Column BThe Human LedgerWhat we’d have to start counting |
|---|---|
| 01Cost per asset | Copy linkSignal-to-volume ratioOutcome per asset, not assets per month |
| 02Time on site, scroll depth | Copy linkValue-per-minute“Was that a good use of your time?” asked of the audience |
| 03Content output volume | Copy link |
| 04Headcount reduction | Copy linkCognitive space ratio% of paid hours spent thinking and making, not QA-ing machine output |
The list
The ledger is a working spec, not a manifesto.
Get new principles, revisions and the arguments that killed them, roughly once a month and never more often.