Briefing creative that survives contact with reality

Most creative briefs fail in one of two directions: they're so vague the work comes back generic, or so prescriptive the work comes back lifeless. A brief that survives contact with reality does three things. First, it ties the asset to a real plan task (an audience, an objective, a channel), so "good" has a definition before anyone opens a design tool. Second, it restricts the claims: only evidence with sources goes in, and everything else is decoration pretending to be persuasion. Third, it names how success will be measured (the event, the URL, the window) before the asset exists, because measurement designed after the fact is just storytelling.

There's a subtler discipline in the review step. Creative review works when it's a preference with a reason ("passed, because the headline buries the offer"), not a verdict from authority. Reasons teach; verdicts just rank. Over time, a team that records why it rejected directions builds a shared taste that compounds faster than any style guide. And the brand constraints (voice, fonts, colors, logos) aren't the enemy of good work; they're the rails that let a team move fast without drifting off-brand.

Finally, respect the industry you're in. Persuasion tactics have ethical boundaries that aren't optional: scarcity and loss-aversion framing stays out of mental health; scarcity, extreme-situation, and competition framing stays out of healthcare; extreme-situation framing stays out of financial services, fitness/wellness, and legal; competition framing stays out of real estate. A brief that ignores these isn't edgy. It's a liability.

Plan tasks route into Create from the task itself: the run starts from the task's title and channel and stays linked to the task. The Brand Style Guide link and the Brand Kit (voice, fonts, colors, logos, Style Notes) keep assets on-brand by default. Idea review uses Liked / Passed / Not rated with reasons on rejection, so taste accumulates instead of evaporating. The Content brief template encodes the whole discipline: approved evidence only, compliance constraints up front, and the measurement event attached before generation.