Best Prompts for Marketing

Customer research often arrives as a pile of interview notes, survey responses, call transcripts, and sales observations. The marketing team still has to turn that material into a campaign someone can approve and execute. That translation is where useful detail gets lost. A striking customer quote becomes a vague theme. A recurring problem becomes a broad claim. Assumptions begin to look like findings.

A strong prompt can help produce a first campaign brief without hiding those gaps. The goal is not to ask AI for a campaign idea. It is to give it a disciplined method for moving from research evidence to a focused recommendation.

Why ordinary prompting fails

“Create a campaign brief from this research” leaves too many decisions unstated. The model may combine unlike customer segments, treat one memorable comment as a pattern, invent supporting facts, or jump directly to slogans. The result can sound polished while making it difficult to see why a message was chosen.

A better prompt defines the audience decision, separates evidence from interpretation, and requires citations back to the supplied material. It also asks for unresolved questions. This makes the brief easier to challenge before creative work begins.

Reusable prompt

Copy
ROLE
You are a senior marketing strategist. Turn supplied customer research into a focused campaign brief. Be specific, conservative with claims, and faithful to the source material.

REQUIRED INPUTS
1. Campaign objective and desired customer action
2. Product or offer, including relevant limitations
3. Target segment or buying situation
4. Customer research with source labels, such as INT-01, SURVEY-12, or CALL-07
5. Brand and legal constraints
6. Channels under consideration
7. Any existing positioning or message that must be retained

STEPS
1. Check whether the inputs are sufficient. List missing information that could materially change the brief.
2. Extract customer problems, desired outcomes, objections, triggers, and language. Cite each item with its source label.
3. Group repeated observations, but do not call something a pattern unless it appears in at least two independent sources. State the source count.
4. Separate direct evidence from your interpretation.
5. Recommend one primary audience, one campaign problem, and one desired action. Explain why these are the best-supported choices.
6. Develop a message hierarchy: core proposition, three supporting points, proof available, objections to address, and words or claims to avoid.
7. Suggest three campaign concepts. For each, explain the evidence behind it and the risk of misinterpretation.
8. Recommend the strongest concept and outline channel adaptations without writing final creative.

OUTPUT FORMAT
A. Input gaps
B. Research synthesis table: finding | type | source labels | frequency | confidence
C. Campaign brief: objective | audience | situation | problem | desired action
D. Message hierarchy
E. Three concept directions
F. Recommended direction and rationale
G. Evidence still needed
H. Approval questions

EVIDENCE AND UNCERTAINTY RULES
- Use only the supplied inputs.
- Never invent quotes, customer facts, product capabilities, or performance claims.
- Put direct customer wording in quotation marks and cite its source label.
- Label interpretations as “Interpretation.”
- Use High, Medium, or Low confidence and explain each rating.
- If evidence conflicts, show the conflict rather than resolving it silently.
- If the target segment is too broad, say so and propose a narrower choice.

What to provide

Provide the rawest useful research you can, with labels that survive copying into the prompt. Include who each participant is in business terms, but remove personal information that is not needed. Add the campaign objective, the action you want the audience to take, product boundaries, channel constraints, and claims that require legal review.

Do not provide only a previous summary if the underlying notes are available. Summaries often erase frequency, disagreement, and exact customer language. If the source set is large, include an index explaining each source and split the work into batches before asking for a combined brief.

How to review the output

Start with the synthesis table, not the campaign concepts. Check every important finding against its cited source. Confirm that repeated observations really come from independent sources and that the selected audience matches the objective. Then review the boundary between evidence and interpretation.

Ask the people closest to the research to challenge the core proposition. Could they point to the inputs that support it? Does it promise more than the product can do? Would the audience recognize the problem in its own language? Finally, treat concept directions as options for discussion, not approved copy.

Where it fails

This prompt cannot repair weak or biased research. If all interviews came from enthusiastic customers, the brief may understate objections. If source labels are missing, traceability will be poor. It also cannot decide strategic tradeoffs that depend on budget, sales capacity, legal advice, or company priorities that were never provided.

The prompt is less useful when several distinct segments are mixed together. In that case, run it once per segment and compare the resulting briefs. Human review remains essential for claims, brand judgment, and final channel choices.

Practical takeaway

The best campaign-brief prompt does not ask AI to be imaginative first. It asks AI to show its evidence, narrow the decision, and expose uncertainty. That gives marketing a brief it can inspect rather than a polished story it has to reverse-engineer.

Try the prompt in Agentic Workers with a small, clearly labeled research set, then review every recommendation beside its source before expanding the campaign.

<!-- X derivative: The best prompt for a campaign brief turns labeled customer research into a traceable recommendation, with evidence, confidence, and gaps shown before creative work starts. -->