Start your learning journeyAI Playbook
Log inStart learning
Role playbook

Marketing and content

Turn briefs into on-brand drafts, research SEO, analyse campaigns from your own exports, repurpose one asset into ten, watch competitors every week, set an image and video policy, and check every claim before it ships.

Checked 13 Oct 202621 min read16 prompts1 skill2 hooks2 routines

For Content marketers, campaign and growth managers, SEO specialists, brand and social teams

Ship on-brand content faster, with every claim checked before it goes out

  • Three channel drafts from one brief in about 20 minutes, in your brand voice
  • Every factual claim in a draft listed with its source, or cut before review
  • Campaign readouts computed from your exports, with totals that reconcile to each platform
  • A competitor digest waiting for you every Monday morning
  • A written policy for AI images and video that legal has signed
Works withClaudeChatGPTGeminiMicrosoft CopilotClaude CodeAhrefsGoogle Analytics
On this page
  1. Where AI helps in this job
  2. Start in 15 minutes
  3. The tools people in this role use
  4. Members only: Workflow: brief to on-brand drafts
  5. Members only: Workflow: SEO research to a content brief
  6. Members only: Workflow: campaign analysis from your exports
  7. Members only: Workflow: repurpose one asset into ten
  8. Members only: Fact-checking and claims compliance
  9. Members only: Image and video generation policy
  10. Members only: A skill for marketing
  11. Members only: Guardrails and hooks
  12. Members only: Routines that run without you
  13. Members only: Prompt ladder for marketing
  14. Members only: Measure it
  15. Members only: Risks and what not to automate

Most marketing weeks repeat the same work: turn briefs into drafts, find topics people search for, read campaign numbers, cut long assets into short ones, keep an eye on competitors and get everything past brand and legal review. AI is fast at the drafting and the reading. It is also fluent at writing claims you cannot prove, so the habit this page builds is simple: give the model your facts, compute numbers with code, and check every claim before it ships.

Where AI helps in this job#

TaskWhat AI does wellWhat you still own
Briefs to first draftsThree channel versions from one brief in your voice, in minutesThe angle, the final edit, the publish button
SEO researchClusters a keyword export by search intent, outlines pages, finds gapsChoosing topics you have real expertise in
Campaign analysisComputes CTR, CPA and ROAS from exports and lists the biggest changesWhat the numbers mean and where the budget goes
RepurposingCuts a webinar, report or podcast into posts, emails and clip scriptsChoosing what is worth saying again
Competitor monitoringReads pricing pages, launches and messaging each week and lists what changedDeciding whether to respond
Images and videoConcepts, backgrounds, storyboards and variations for testingRights, likeness, disclosure and brand fit
Fact-checking and claimsLists every claim in a draft and matches it to your evidenceHolding the evidence and signing off
Subject lines and ad variantsTen variants to test, labelled by angleThe test design and picking the winner

AI is weak at facts about your product, numbers it did not compute, anything newer than its sources, and brand judgment. It will write "the most trusted platform for finance teams" with nothing behind it, invent a statistic that sounds right, or quote a customer who never said that. It also drifts toward a generic, upbeat tone unless you give it your voice in writing. Each workflow below closes those gaps the same way: facts in, code for numbers, a claims table out, and a person who approves.

Start in 15 minutes#

These work in Claude, ChatGPT, Gemini or Microsoft Copilot (formerly Microsoft 365 Copilot). Paste your own text where you see brackets.

Prompt

Turn a paragraph into a LinkedIn post

Level 1Any assistant
Rewrite this paragraph as a LinkedIn post of at most 120 words for [AUDIENCE]. Keep every fact exactly as written. Do not add numbers, quotes, customer names or claims that are not in the paragraph. End with one question for the reader.

[PASTE PARAGRAPH]

Good output keeps every fact from your paragraph and adds none. Compare it line by line with the original and delete anything new.

Prompt

Ten subject lines in your voice

Level 2Any assistant
Write 10 subject lines for this email:

[PASTE THE EMAIL OR A SUMMARY]

These three subject lines performed well for us. Match their tone and length:
1. [PAST WINNER 1]
2. [PAST WINNER 2]
3. [PAST WINNER 3]

Rules: under 50 characters, no capitals for emphasis, no emoji, nothing the email does not deliver. Show the character count of each line and label the angle it uses: benefit, curiosity, urgency or social proof.

Good lines sound like your past winners and stay under 50 characters. Recount a few lengths yourself and cut any line that promises something the email does not contain.

Prompt

List every claim in a draft

Level 2Any assistant
List every factual claim in the draft below as a table with three columns: the claim quoted exactly, its type (statistic, comparison, superlative, product capability, customer result, third-party fact), and the evidence that would support it. Include claims hidden in adjectives such as "fastest" or "most trusted". Do not judge whether the claims are true and do not rewrite the draft.

[PASTE DRAFT]

Good output catches the adjectives as well as the numbers. Count the claims yourself first, then attach a source to each one or cut it.

Warning

Do not paste customer lists, unreleased results, embargoed launch details or anything under NDA into a personal chat account. Use your company's plan and check its data terms. Anthropic, for example, does not train on inputs or outputs from its commercial products (Claude Team, Enterprise and the API) by default. On a personal plan, check the training setting first.

The tools people in this role use#

  • A general assistant on your company plan. Claude, ChatGPT, Gemini or Microsoft Copilot handle drafting, research and analysis. For numbers, use one that runs code on your file instead of estimating.
  • Anthropic's marketing plugin. Free and open source, for Claude Cowork and Claude Code. Its skills cover draft-content, brand-review, campaign-plan, competitive-brief, performance-report, seo-audit, email-sequence and content-creation. It comes configured for the MCP servers of Ahrefs, Similarweb, HubSpot, Klaviyo, Canva, Figma, Notion, Supermetrics, Amplitude and Slack, so Claude can read your real data once you connect them. The plugins guide shows how to install and adapt one.
  • SEO and analytics data. Ahrefs runs an MCP server, and Google Analytics publishes an open-source MCP server on GitHub (Apache 2.0). Without either, a CSV export from your SEO tool or Search Console works for every workflow here.
  • Image and video tools. Use the ones your company approved, on a plan whose terms cover commercial use. The policy section below says what to check.

Workflow: brief to on-brand drafts

Run it for every new piece: a launch post, a newsletter section, a landing page. The inputs are the brief, your brand voice profile (it lives in the skill further down), and two or three pieces you already published and liked. 1. Fill in the ...

Members only below this line

Get the full Marketing and content playbook free

What you get on this page:

  • 4 step-by-step workflows
  • 13 prompts
  • 1 skill
  • 2 hooks
  • 2 routines

Free. You also get Stratiflux Weekly: one email a week, unsubscribe anytime.

Also in the members part

  1. Workflow: SEO research to a content brief
  2. Workflow: campaign analysis from your exports
  3. Workflow: repurpose one asset into ten
  4. Fact-checking and claims compliance
  5. Image and video generation policy
  6. A skill for marketing
  7. Guardrails and hooks
  8. Routines that run without you
  9. Prompt ladder for marketing
  10. Measure it
  11. Risks and what not to automate

Sources

Checked against these on 13 Oct 2026.

  1. Anthropic knowledge-work-plugins: marketing plugin (opens in a new tab)github.com
  2. Claude Help Center: Use skills in Claude (opens in a new tab)support.claude.com
  3. Claude Help Center: Use connectors (tool permissions) (opens in a new tab)support.claude.com
  4. Claude Help Center: Schedule recurring tasks in Claude Cowork (opens in a new tab)support.claude.com
  5. Claude Code docs: Skills (opens in a new tab)code.claude.com
  6. Claude Code docs: Hooks reference (opens in a new tab)code.claude.com
  7. Claude Code docs: Routines (opens in a new tab)code.claude.com
  8. Agent Skills: the open format and the products that read it (opens in a new tab)agentskills.io
  9. Anthropic Privacy Center: commercial products are not used for training by default (opens in a new tab)privacy.claude.com
  10. FTC: Final rule banning fake reviews and testimonials (August 2024) (opens in a new tab)ftc.gov
  11. FTC: Health Products Compliance Guidance (substantiation before an ad runs) (opens in a new tab)ftc.gov
  12. Google Search Central: Spam policies (scaled content abuse) (opens in a new tab)developers.google.com
  13. YouTube Help: Disclosing altered or synthetic content (opens in a new tab)support.google.com
  14. US Copyright Office: Copyright and Artificial Intelligence, Part 2 (January 2025) (opens in a new tab)copyright.gov
  15. C2PA: Content Credentials (opens in a new tab)c2pa.org
  16. European Commission: AI Act application timeline (opens in a new tab)digital-strategy.ec.europa.eu
  17. EUR-Lex: Regulation (EU) 2024/1689, the AI Act (Article 50) (opens in a new tab)eur-lex.europa.eu
  18. Google Analytics MCP server (GitHub) (opens in a new tab)github.com
  19. Gemini Apps Help: Schedule actions in Gemini Apps (opens in a new tab)support.google.com
  20. Microsoft Support: Schedule your most used Copilot prompts (opens in a new tab)support.microsoft.com

Related

GuidePrompting, from easy to hardHow the 16 prompts on this page workGuideSkills: teach an agent a procedure once, reuse it everywhereHow the skill on this page worksGuideHooks: guardrails that run every timeHow the 2 hooks on this page workGuideRoutines: scheduled agents that open pull requestsHow the 2 routines on this page work