Design a campaign system

Step 1 — Load the raw truth of the thing

I’m building a campaign for [thing]. Interview me, one question at a time: who it’s actually for; what’s genuinely true and provable about it; what action I want people to take; which channels I actually own and can publish to; my constraints and budget. Then summarize everything as CAMPAIGN FACTS — nothing invented, gaps marked UNKNOWN.

Same move as every lab: the quality of everything downstream is set by what you load, and marketing is where the UNKNOWN discipline earns its keep most — because this is the domain where the model will most happily invent a glowing customer who doesn’t exist.

Step 2 — Demand range, then choose

Propose THREE campaign concepts that are genuinely different from each other — different emotional angle, different visual world, different voice. For each: the core message in one line, why this audience cares, what the assets look and sound like, and one honest risk. Do not blend them.

Ask once and you get the average campaign — the one every competitor’s intern also got. Asking for three real territories is how you get a choice instead of a default. Pick one. Kill the other two out loud.

Step 3 — Write the system: the campaign brief

Turn concept [X] into a one-page CAMPAIGN BRIEF that any person — or any AI — could produce on-system assets from: THE AUDIENCE; THE SINGLE MESSAGE; VOICE, with three example lines and three “never like this” lines; VISUAL DIRECTION concrete enough to execute in Canva — colors, type feel, image style; THE CALL TO ACTION; CLAIMS WE MAY MAKE, drawn from CAMPAIGN FACTS only; and BANNED MOVES — clichés, invented claims, off-voice habits.

This page is the whole trick. It’s a persona prompt grown up: the standing context that pushes every future asset off the generic average and onto your campaign. Now make it permanent — create a project with the brief and CAMPAIGN FACTS in its files, so every asset conversation starts already steered.

Step 4 — Run the asset loop

Inside the project, produce the first batch, one ask at a time:

From the brief: draft [three social posts / the landing-page hero and sections / the announcement email / a one-page flyer]. Give me the copy exactly as it should ship, plus a visual spec I can execute in Canva — layout, imagery, where the text sits.

Then build the visuals in Canva from the specs. The division of labor is honest: the words arrive pre-steered by your brief; the visuals get made where visuals are made. If your tools generate images, the brief’s visual direction is the prompt.

Step 5 — The brand cop pass

Fresh window. Paste the brief, then every asset:

You are the guardian of this brief. Check every asset against it: voice violations, off-message lines, banned moves, and any claim that is not in CAMPAIGN FACTS. Return a table, worst first.

The fresh-window shape again — a checker living inside the window that wrote the assets goes easy on them. What the cop flags, you fix; what survives is a campaign that actually obeys its own system.

Step 6 — The production workflow

Build the production workflow: a four-week calendar — what ships, where, when; the checklist every asset passes before shipping, brand-cop pass included; and a weekly 15-minute ritual — what shipped, what performed, what the next batch is.

Put the calendar on the project page next to the brief. Next week, producing the batch takes minutes: open the project, ask for it — the window starts loaded, so the system does the remembering, not you.

---

Check your understanding: what does the brief do, mechanically, that makes week four’s assets match week one’s — and where do the campaign’s claims come from?

Use this process for other recurring communication. A standing brief + loaded project + brand-cop pass is any stream of communication that has to stay consistent: the podcast, the job search, the fundraise, the internal comms drumbeat. One page of steering, reused forever.