Write an AI campaign brief your reviewer can actually use

A useful AI campaign brief tells the next person what to make, what they may claim and how to decide whether it is ready. Start with one audience situation, one real offer, one intended action and the evidence behind the message. Add a named reviewer and a short acceptance checklist. If an important fact is unknown, leave it visibly unresolved instead of asking AI to fill the gap.

This is a working document for a small team, a freelancer or a solo marketer reviewing their own work. It does not need to be a presentation. One page can be enough when the decisions are specific.

1. Brief the customer decision before the asset

“Write three posts” describes a quantity. “Help an operations manager decide whether our onboarding review fits the problem they are seeing” describes a job. The second instruction gives the writer something to explain and the reviewer something to evaluate.

Name the situation that makes the offer relevant. Then choose the smallest useful next step: read the scope, complete a worksheet or ask a scoping question. Do not make one asset responsible for awareness, education, signup and purchase at the same time. If the audience is still vague, work through the positioning guide before writing campaign copy.

2. Separate approved facts from the things you hope are true

Provide product or service facts, the current scope, relevant constraints and customer language you have permission to use. Remove personal or confidential information before putting material into a tool unless its use has been approved. Keep an evidence note beside each proposed claim so the reviewer can check its source.

“We will improve your conversion rate” is not an approved fact just because the team wants that outcome. Describe the work that is actually included and keep the desired result separate.

The one-page campaign brief

Copy these prompts into your working document or use the existing AI Campaign Brief worksheet. Write “unknown” where a decision is still needed.

Worked example: a fictional onboarding-review offer

This example is invented to explain the method; it is not a client result or an offer for sale. Imagine a consultant offering a review of one signup-to-first-task journey. The team wants a short article introduction that leads readers to a scope page.

• Audience: a small software team that sees new accounts created but does not know where users stop.
• Approved scope: a walkthrough of one journey and a written list of issues to investigate. Implementation and revenue forecasts are excluded.
• Message: make the first-use path visible before deciding what to change.
• Action: read the scope and send a question about fit. Do not imply that clicking books the work.
• Unknowns: current capacity and delivery timing. The owner must confirm both before the campaign describes them.
• Reviewer check: can a reader tell what is included, what is excluded and what happens after the inquiry?

A draft such as “Double activation in a week” fails this brief: neither the outcome nor the timing is supported. “See where your first-use journey needs investigation” stays within the illustrative scope. It still needs a useful explanation and a verified destination.

3. Ask AI to identify gaps before it writes

Using only the approved facts in this brief, identify missing decisions that would change the message. Separate facts, observations and hypotheses. Do not invent quotes, results, prices, availability or features. Then draft two distinct approaches for the same audience and action. For each, explain the customer question it addresses and list any statement a human must verify.

Ask for a small number of meaningfully different approaches, not dozens of cosmetic variations. Keep the approved brief alongside the output. A later revision should not quietly introduce a broader promise.

4. Give the reviewer a pass-or-revise test

Review the work in this order. A strong headline cannot compensate for a missing offer decision.

1. Facts: every material claim is supported, within scope and current enough for this use.
2. Audience: the example addresses the chosen situation without assuming every reader has it.
3. Handoff: open the actual destination on a phone. Check that the promise, instructions and next action agree.
4. Readability: check headings, image descriptions, link labels and anything that becomes hard to use on a narrow screen.
5. Ownership: record the decision, the person making it and any unresolved dependency.

Have an approved test process for forms and confirmations. A button that opens the right page does not prove that an inquiry reaches the right person.

5. Decide what you will count before release

Choose an observed action and write its counting rule. Keep a landing-page visit, a submitted inquiry, a qualified conversation and a paid engagement separate. Google Analytics lists distinct recommended events for lead submission and qualification; they require implementation rather than appearing automatically.

For links you control, consistent campaign parameters can help identify referring campaigns. Google's URL-builder guidance explains source, medium and campaign labels, plus content labels for creative variations. Those labels do not prove revenue or establish that one message caused a result.

Use the campaign experiment builder to capture the question, comparison and review date. When the sample is small, report counts and unanswered questions. Do not turn two inquiries into a reliable conversion benchmark.

Finish one brief before making more assets

Try this with one real campaign: complete the brief, ask a reviewer to identify the largest unresolved decision, then revise only what that decision changes. Note the time spent drafting and checking; speed is useful only if the output remains accurate and usable.

If you need to choose the marketing problem first, the free 30-Minute Marketing Reset provides a broader planning exercise. The worksheet and service options below are existing next steps; availability and any engagement scope still need confirmation.

Sources and editorial scope

• Google Analytics: collect campaign data with custom URLs. Publication date not stated; checked October 20, 2026.
• Google Analytics: recommended events. Publication date not stated; checked October 20, 2026.

The brief, review sequence and fictional example are original MarketingCoach editorial guidance. They are planning aids, not a validated score, tested conversion promise or permission to contact people.