id: marketing-email-campaign
framework_id: 4d-framework
version: "1.0.0"
contributor: "Rick Dakan & Joseph Feller / Anthropic"
title: "AI-Assisted Marketing Email Campaigns"
domain: writing
dimensions:
  delegation:
    description: The human marketing strategist designs the campaign's core message, audience segmentation logic, and brand voice. AI then executes variation drafting — subject line alternatives, body copy variants for A/B testing, personalization token suggestions, and CTA phrasing options. The human selects, refines, and approves; AI accelerates the variation space.
    example: A marketing manager defines the campaign goal (re-engage lapsed subscribers with a 20% discount), the audience segment (users inactive for 90+ days), and the brand voice (friendly, direct, never pushy). She then asks AI to draft five subject line options and three body copy variants, which she narrows to one of each after testing against campaign criteria.
    antipattern: Having AI independently design the campaign messaging strategy, choose the audience, set the offer, and draft the emails — then simply forwarding the AI output to the email platform without human strategic review or brand alignment check.
  description:
    description: "Provide AI with a complete campaign brief: target audience demographics and psychographics, campaign objective and success metrics, desired brand voice with examples, the specific offer or message, any compliance constraints (CAN-SPAM, GDPR), and the desired call to action. The more precisely you define success, the more useful the AI drafts."
    example: "Prompt: 'Draft three subject line options and two body copy variants for a re-engagement email. Audience: B2B SaaS users, mid-level product managers, inactive 90+ days. Offer: 20% discount on annual plan upgrade. Voice: confident, peer-to-peer, not salesy — like a trusted colleague nudging, not a pushy vendor. CTA: Schedule a 15-min call. Compliance: include unsubscribe link placeholder. Max 150 words body copy.'"
    antipattern: "Prompting: 'Write a marketing email for our product.' With no audience definition, no campaign goal, no voice guidance, and no offer details, AI produces a generic promotional template that fits no specific audience and advances no particular business goal."
  discernment:
    description: Evaluate each AI-drafted variant against your campaign goals, brand standards, audience fit, and legal requirements before sending. Check that claims are factually accurate, that the tone matches your brand, that the offer details are correct, and that no language could mislead recipients or create compliance issues.
    example: "A campaign manager reviews an AI-drafted email and flags two issues: the subject line 'You''re missing out!' violates the company's brand guideline against FOMO language, and a product claim ('the #1 tool for project managers') is unsubstantiated. She revises both before the email enters the approval workflow."
    antipattern: Sending AI-drafted campaign emails after only a quick skim for spelling errors — missing off-brand phrasing, unverified product claims, incorrect offer details, or compliance issues that could damage brand trust or trigger regulatory scrutiny.
  diligence:
    description: Validate all factual claims, offer details, pricing, and legal disclosures before the email goes out. Consider the ethics of AI use in persuasive marketing communications and whether your audience has any expectation of human-crafted outreach. Maintain a human approval step in the campaign workflow regardless of AI involvement.
    example: Before deploying a 50,000-recipient campaign, the marketing team runs the AI-drafted email through their compliance checklist (unsubscribe link present, sender identification correct, no deceptive subject lines), has the legal team verify the offer terms, and gets final sign-off from the brand director.
    antipattern: Auto-deploying AI-drafted emails through an automated campaign platform with no human review step, sending incorrect pricing or expired offers to thousands of recipients, or using AI-generated urgency language ('Act now before midnight!') for a deadline that doesn't actually exist.
score_hints:
  delegation: 0.25
  description: 0.35
  discernment: 0.20
  diligence: 0.20
tags:
  - marketing
  - email-campaign
  - copywriting
  - brand-voice
reference: "https://aifluencyframework.org"

collaboration:
  pattern: linear_with_loops
  description: "Marketer delegates copy generation, provides audience and goal context, evaluates AI copy for brand fit, then approves send — looping back to description if messaging misses the mark."
  sequence:
    - step: 1
      d: delegation
      label: "Define AI copy role"
      example_prompts:
        - speaker: human
          text: "Draft 3 subject line variants and the body for a re-engagement campaign. You choose the CTA."
        - speaker: ai
          text: "Should I optimize the CTA for clicks or for replies?"
        - speaker: human
          text: "Clicks — we want traffic to the new pricing page."
      triggers_next: "AI's copy role and optimization goal agreed."
    - step: 2
      d: description
      label: "Provide audience segment and brand context"
      example_prompts:
        - speaker: human
          text: "Audience: lapsed users, last active 90+ days ago. Brand voice: friendly, direct, no urgency language. Goal: bring them back to see the new plan."
        - speaker: ai
          text: "Is there a specific pain point from exit surveys I should address?"
        - speaker: human
          text: "Yes — 60% cited pricing confusion. Lead with clarity on the new plan."
      triggers_next: "AI has audience, brand voice, and specific messaging anchor."
    - step: 3
      d: discernment
      label: "Evaluate copy for brand fit and accuracy"
      example_prompts:
        - speaker: human
          text: "Subject line 2 has urgency language — 'Don't miss out' — remove it."
        - speaker: ai
          text: "Replaced with 'See what's new.' Does the body CTA feel too salesy?"
        - speaker: human
          text: "Slightly — soften to 'Take a look' from 'Upgrade now'."
      triggers_next: "Copy aligns with brand voice and factual accuracy confirmed."
      loop_back:
        to: description
        condition: "Copy consistently misses tone or misrepresents product features."
        reason: "Messaging anchor or brand constraints need to be re-provided with examples."
    - step: 4
      d: diligence
      label: "Approve and send"
      example_prompts:
        - speaker: human
          text: "Legal sign-off on claims — approved. Scheduling send for Tuesday 10am."
        - speaker: human
          text: "Logging AI-generated copy version and override history for compliance."
      triggers_next: "Campaign sent. Cycle restarts for next campaign."
      can_restart: true
  transitions:
    - from: delegation
      to: description
      trigger: "Copy role and optimization goal agreed."
    - from: description
      to: discernment
      trigger: "AI delivers copy variants."
    - from: discernment
      to: diligence
      trigger: "Copy validated for brand and accuracy."
    - from: discernment
      to: description
      trigger: "Messaging consistently off — re-provide anchor."
      is_loop_back: true
    - from: diligence
      to: delegation
      trigger: "Next campaign — restart."
      is_cycle_restart: true