12 prompts
AI prompts for marketing agencies
Updated
These AI prompts for marketing agencies cover the client lifecycle: discovery, proposals, planning, production, reporting and QA. They work best with a per-client brand sheet pasted into the [BRACKETS]. Always remove personal data before pasting, and have a strategist review outputs before anything reaches the client.
Winning and onboarding clients
Discovery call summary
Use it right after a sales or kickoff call.
You are an agency strategist. Here are notes or a transcript from a discovery call with [CLIENT_NAME]: [PASTE_NOTES]. Summarize: business goals, the metric they care about most, budget signals, past agency experience, internal decision makers, risks, and open questions. Then draft a 5-line follow-up email confirming what we heard and the next step: [NEXT_STEP].
Proposal scope section
Use it when turning a discovery call into a proposal.
Act as an agency account director. Client goals: [GOALS]. Budget range: [BUDGET]. Timeline: [TIMELINE]. Our services: [SERVICE_LIST_WITH_RATES]. Write the scope section: 3 phases, deliverables per phase, what is out of scope, assumptions, and what we need from the client. Write it so a non-marketer can understand it. Flag any goal that the budget cannot realistically support.
Client onboarding questionnaire
Use it to collect what you need in one pass.
You are an agency onboarding lead for a [SERVICE] engagement. Write a questionnaire of no more than 20 questions covering: business and offers, audience, competitors, brand voice, access and tools, approvals, and success metrics. Group the questions, mark required ones, and add an example answer under the 5 hardest questions so clients know what level of detail we need.
Planning and production
Quarterly content calendar
Use it when a client signs a retainer and needs a plan.
Act as a content strategist for [CLIENT], who sells [OFFER] to [AUDIENCE]. Their content pillars: [PILLARS]. Channels: [CHANNELS]. Monthly volume: [VOLUME_PER_CHANNEL]. Build a 12-week calendar as a table: week, channel, pillar, working title, format, goal (traffic, leads, trust), and owner. Spread pillars evenly and tie at least 3 items to these dates: [KEY_DATES].
Brief a freelancer
Use it when handing work to a contractor so it comes back right the first time.
You are a project manager at an agency. Write a brief for a freelance [ROLE] working on [CLIENT]. Task: [TASK]. Audience: [AUDIENCE]. Voice rules: [VOICE_RULES]. Examples of good work: [LINKS_OR_DESCRIPTIONS]. Deadline: [DATE]. Budget: [RATE]. Include a definition of done and a 6-point checklist they should run before submitting.
Repurpose one asset for a client
Use it to stretch a single piece of content across channels.
Act as a content producer for [CLIENT]. Source asset: [PASTE_ARTICLE_OR_TRANSCRIPT]. Create: 3 LinkedIn posts, 1 X thread of 5 to 7 posts, 1 newsletter section of 150 words and 3 short video hooks. Each piece should stand alone and use a different point from the source. Follow these voice rules: [VOICE_RULES].
Ad variation matrix
Use it when a client wants to test several angles quickly.
You are a paid social copywriter. Product: [PRODUCT]. Audience: [AUDIENCE]. Offer: [OFFER]. Platform: [PLATFORM_AND_CHARACTER_LIMITS]. Write 4 angles (pain, outcome, proof, objection). For each angle write 3 primary texts and 2 headlines within the platform limits. Return a table with angle, primary text, headline and character counts.
Reporting and retention
Monthly client report narrative
Use it so reports explain results instead of listing charts.
Act as an account manager writing to [CLIENT_CONTACT_ROLE]. Data this month vs last: [PASTE_METRICS]. Work delivered: [DELIVERABLES]. Write: a 3-sentence executive summary, what worked and why, what did not and what we are changing, and next month's plan. Tie each result to the client's stated goal: [GOAL]. Do not hide weak results.
Churn risk check
Use it when a client relationship feels shaky.
You are an agency director. Here are recent signals from [CLIENT]: [EMAILS, MEETING NOTES, MISSED APPROVALS, RESULTS]. Rate churn risk as low, medium or high with reasons. Then suggest 3 actions for the next 2 weeks, and a short, honest email opening a conversation about how things are going.
Case study from a finished engagement
Use it when a client has agreed to be featured.
Act as an agency marketer. Engagement details: client [CLIENT], starting situation [BEFORE], what we did [WORK], results [VERIFIED_RESULTS_AND_TIMEFRAME], client quote [APPROVED_QUOTE]. Write a case study with: headline stating the result, challenge, approach in 3 steps, results with numbers, and quote. Use only the results provided. Under 600 words.
Quality control
Pre-delivery QA review
Use it on any deliverable before it goes to the client.
You are a QA editor at an agency. Client brand sheet: [BRAND_SHEET]. Deliverable: [PASTE_DELIVERABLE]. Check for: off-voice wording, banned words, claims without sources, wrong product names or prices, broken formatting, missing CTA, and spelling. Return a list of issues with location and fix. End with pass or fail.
Competitor watch digest
Use it to give clients a short monthly competitive view.
Act as a market analyst for [CLIENT]. Here are notes on what competitors published, launched or advertised this month: [PASTE_NOTES]. Summarize into: notable moves, messaging changes, offers, and 2 opportunities for our client to respond. Keep it under 250 words.
Run these automatically
Thig lets agencies set up one workspace per client, each with its own strategy docs, brand voice and agent feed. See how it fits digital agencies, or start with the SEO agent and Writer agent.
Frequently asked questions
How do marketing agencies use AI prompts?
Agencies use prompts to speed up repeatable work: discovery summaries, proposals, monthly reports, content calendars and quality checks, with a strategist reviewing every output.
Is it safe to paste client data into AI tools?
Check your client contracts and the tool's data policy first. Remove personal data and confidential numbers where you can, and use business plans that do not train on your inputs.
How do I keep AI output consistent across many clients?
Keep a short brand sheet per client (voice rules, audience, offers, banned words) and paste it into every prompt for that client.