A campaign brief is a decision tool. It should tell a team who the campaign is for, what changes for that person, why the offer is credible and what happens next. AI can help organize evidence and explore alternatives, but it cannot supply missing product truth or grant budget authority.
Start with an approved offer description, audience research, prior results if available, brand guidance, channel constraints and the decision owner. For practice, use a fictional community workshop and label every invented detail as sample material. Avoid putting customer-level data into a tool unless your organization has approved that use.
What you will leave with
A one-page campaign brief, three differentiated message options and a test plan with an owner.
Step 1
Build an evidence pack
Give the agent the actual offer, relevant research and approved claims. Ask it to extract a short fact list with source references. Put unsupported audience beliefs in a separate hypotheses section. Include operational limits such as geography, eligibility, capacity and the next action a visitor can take.
The facts are traceable and the unknowns are visible.
Step 2
Choose one audience and one action
Ask for a few plausible audience/problem combinations, then select one based on evidence and business priorities. Write a sentence naming the audience, their problem, the offer and the action. This prevents the campaign from becoming a collection of unrelated copy. A human owns that strategic choice.
The brief names a specific audience and a single primary action.
Step 3
Draft alternatives with a reason
Have AI propose three message directions with different hypotheses, such as convenience, confidence or participation. Each should have a headline, supporting proof, call to action and the assumption being tested. Reject variations that merely swap adjectives or invent a testimonial. Keep the source claim beside each proposed claim during review.
Try this instruction
Using only this evidence pack, draft three campaign message directions. For each include audience problem, promise, evidence, headline, CTA and assumption to test. Mark unsupported claims instead of filling them in.
Every direction has a distinct hypothesis and only supportable claims.
Step 4
Turn the idea into a test
Define a primary outcome, a measurement source and the unit of comparison. Record what stays consistent between variations and who decides whether to continue. Include available budget and duration only when an authorized person supplies them. Do not let the agent turn a draft plan into a launched campaign.
The test has a measurable outcome, constraints and a named decision owner.
Step 5
Review the execution handoff
Ask a reviewer to compare the brief against the evidence pack. Check the offer, claims, destination, tracking and feasibility with the people who will execute it. The finished artifact should include owners and open questions so the next person can act without recreating the conversation.
Creative, measurement and launch responsibilities are explicit.
When you get stuck
The copy sounds good but could describe any company.
Add the actual offer, audience evidence and proof. Ask the reviewer to highlight sentences that have no supporting fact.
AI invents customer quotes or performance claims.
Remove them. Keep a claim-to-source table and require approval of material claims before publication.
Terms explained
- Message hypothesis
- A reason you expect one message to be more useful or persuasive, which still needs testing.
- Primary outcome
- The main observable result the campaign is intended to create.
Sources and further reading
The workflow and templates are teaching guidance. The sources below support the underlying tools and concepts; product setup and availability can change.
Take it into your next task
Your completion checklist
0 / 5 checked
Checks are saved on this browser. Mark them only after you inspect the evidence. These are your checks, not an automated certification.