Thinking · Newsletter
What a creative operation looks like when it actually works
Production, media and analysis usually sit in different tools, different meetings and different companies. Each handoff loses the context the next step needed.
A team can ship dozens of ads a month and learn almost nothing. Output is not the same as learning, and the difference is entirely operational.
Where context leaks
Three handoffs, each one lossy.
Brief to production. The hypothesis gets summarised into a deliverable. The reason the ad exists survives as a line in a doc that nobody opens on the shoot day.
Production to media. A batch of assets arrives with filenames. The media team cannot tell which ones are variations of the same bet, so they get treated as unrelated ads and compared as though they were.
Media to the next brief. A performance report arrives as a ranking. The next concepting session starts from a blank page because the ranking does not explain anything.
What closing the loop requires
Not another dashboard. Five things written down in a shared place:
- The question this batch exists to answer.
- What varies between assets, and what is held constant.
- Which asset belongs to which bet.
- The result read at the level of hook, format, device and message, not just ROAS.
- What the next brief does differently because of it.
Most teams have none of these written anywhere. The ones that do compound; the ones that do not restart every month.
The smallest version
You do not need a system to start. You need one shared document per batch with those five fields. If it exists and is filled in, the loop is closed. If it does not, no amount of tooling will close it.
Questions people ask
Do we need everything in one company? No, but every handoff between companies needs the five fields to travel with it, and they usually do not.
How do we read below ROAS? Tag assets on the dimensions you varied before they go live. Retrofitting that classification after the fact is possible and painful.