Skip to main content

๐Ÿ“ข Marketing Copy Prompt

A marketing copy prompt directs an LLM to generate persuasive, brand-consistent content โ€” ad copy, landing pages, email campaigns, product descriptions, and A/B test variants โ€” all aligned to a specific audience, voice, and conversion goal.

Why This Mattersโ€‹

The average person encounters 6,000โ€“10,000 ads per day. To cut through the noise, marketing copy must be specific, emotionally compelling, and laser-focused on the audience's pain points. A well-engineered prompt consistently produces copy that converts โ€” because it encodes the brand voice, audience psychology, and conversion principles that make great marketing work.

The Production Promptโ€‹

Marketing Copy โ€” Full System Prompt
You are an expert direct-response copywriter with 15 years of experience writing high-converting copy for B2B SaaS, DTC ecommerce, and digital products.

**Core Frameworks You Apply:**
- **AIDA:** Attention โ†’ Interest โ†’ Desire โ†’ Action
- **PAS:** Problem โ†’ Agitation โ†’ Solution
- **BAB:** Before โ†’ After โ†’ Bridge
- Choose the framework that best fits the content type and goal.

**Brand Context:**
- Brand name: {{brand_name}}
- Brand voice: {{voice}} (e.g., "confident and witty, like a smart friend who works in tech")
- Target audience: {{audience}}
- Key value proposition: {{value_prop}}
- Competitors to differentiate from: {{competitors}}
- Tone words: {{tone_words}} (e.g., bold, approachable, no-nonsense, empathetic)
- Words to avoid: {{avoid_words}} (e.g., "synergy", "revolutionary", "game-changer")

**Task:** {{task_type}}

**Instructions by Content Type:**

**For Ad Copy (Google/Meta/LinkedIn):**
- Write 5 headline variants (under 30 characters each)
- Write 3 description variants (under 90 characters each)
- Include the primary CTA keyword in at least 3 headlines
- Use power words: free, proven, instant, exclusive, limited
- Each variant should test a different angle: pain point, benefit, social proof, urgency

**For Landing Page Copy:**
- Hero section: headline (under 10 words), subheadline (under 25 words), CTA button text
- Problem section: 3 pain points the audience faces
- Solution section: how the product solves each pain point
- Social proof section: suggest testimonial themes and trust signals
- CTA section: urgency-driven closing with clear next step

**For Email Campaigns:**
- Subject line: 5 variants (under 50 characters, optimize for open rate)
- Preview text: matching preview for each subject line
- Email body: hook โ†’ value โ†’ CTA structure, under 200 words
- P.S. line with an additional incentive or urgency element

**Quality Standards:**
- Every sentence must serve a purpose โ€” no filler
- Lead with benefits, not features
- Use specific numbers over vague claims ("saves 4 hours/week" not "saves time")
- Write at a 6th-grade reading level (Flesch-Kincaid)
- Include at least one element of social proof or specificity per piece

Bad vs. Improved Promptsโ€‹

โŒ Bad Promptโ€‹

Write marketing copy for my project management app.

Why it fails: No audience, no brand voice, no content type, no value proposition, no conversion goal. The output will be generic "streamline your workflow" copy that every competitor also uses.

โœ… Improved Promptโ€‹

You are an expert copywriter specializing in B2B SaaS.

Write Google Ads copy for "TaskForge" โ€” a project management tool built specifically for remote engineering teams of 10-50 people.

Target audience: Engineering managers frustrated with Jira's complexity.
Value prop: Set up in 5 minutes, not 5 days. Built for how engineers actually work.
Brand voice: Confident, slightly irreverent, developer-friendly. Think Stripe's marketing tone.
Words to avoid: "revolutionary", "game-changer", "synergy"

Deliverables:
1. 5 headline variants (under 30 chars each) โ€” test these angles: pain point, speed, simplicity, social proof, contrast with Jira
2. 3 description variants (under 90 chars each) โ€” lead with a benefit + specific number
3. For each variant, note which psychological trigger it uses (urgency, social proof, curiosity, fear of missing out, pain avoidance)

Output as a numbered list grouped by headlines and descriptions.

Try It Yourselfโ€‹

๐Ÿงช Try It Yourself

Edit the prompt and click Run to see the AI response.

Tips for Customizationโ€‹

CustomizationHow to Modify the Prompt
A/B testingAdd: "For each piece of copy, create Variant A (benefit-focused) and Variant B (pain-focused) for split testing"
Seasonal campaignAdd: "This is for a Black Friday campaign running Nov 25โ€“28. Incorporate urgency and limited-time messaging"
Product launchSwitch task: "Write a product launch email sequence: 3 emails over 5 days โ€” teaser, launch day, last chance"
Brand voice sampleAdd: "Here is a sample of our brand voice: [paste 2-3 sentences]. Match this tone exactly."
LocalizationAdd: "Write copy for the German market. Use formal 'Sie' address. Adapt cultural references โ€” don't just translate."
ComplianceAdd: "This is for financial services โ€” avoid guaranteed returns language. Include required disclaimer: {{disclaimer}}"

Practice Challengeโ€‹

Challenge

Pick a product you use daily. Write a marketing copy prompt that:

  1. Defines the target audience with specifics (demographics, psychographics, pain points)
  2. Specifies the brand voice with a comparison ("like X but more Y")
  3. Requests 5 headline variants testing different psychological triggers
  4. Sets character limits for each piece

Run it and evaluate: Which headline would you actually click? Which one sounds most like the real brand? Refine and re-run.

Real-World Scenarioโ€‹

Scenario: An ecommerce brand needs to generate product descriptions for 500 SKUs across 4 languages for their holiday catalog.

Implementation approach:

  1. Product data feed: export SKU data โ€” product name, category, features, price, materials, dimensions
  2. Template prompt: one master copy prompt per category (electronics, apparel, home goods) with brand voice locked in
  3. Per-SKU generation: inject each product's data into the template and generate: 50-word description, 3 bullet points, meta description
  4. A/B variants: generate 2 variants per SKU โ€” one benefit-led, one feature-led โ€” for multivariate testing
  5. Multi-language: run a second pass with localization prompts (not translation โ€” cultural adaptation)
  6. Quality gates: automated checks for character limits, banned words, reading level score
  7. Human review: copywriters review 10% random sample to calibrate quality

Temperature: 0.8 for creative variety across variants, 0.4 for brand-consistent descriptions.

This produces 500 product descriptions in under 2 hours vs. 2 weeks of manual copywriting.

Interview Questionโ€‹

Interview Question

Q: How would you maintain brand voice consistency across hundreds of AI-generated marketing assets?

A: Brand voice consistency requires constraints at multiple levels:

  1. Voice definition โ€” define the brand voice with specific, testable attributes: "We are witty but never sarcastic. We use contractions. We never use exclamation marks. We reference technology analogies." Include 3โ€“5 examples of on-brand and off-brand sentences
  2. Tone words as guardrails โ€” provide explicit lists: "Always: confident, clear, specific. Never: hyperbolic, vague, corporate."
  3. Reference samples โ€” include 2โ€“3 paragraphs of approved copy in the system prompt and instruct: "Match this voice exactly"
  4. Banned word list โ€” explicitly list words and phrases the brand never uses
  5. Automated scoring โ€” build a post-generation classifier that scores new copy against the brand voice rubric, flagging anything below threshold for human review
  6. Few-shot anchoring โ€” include examples of the exact output format with the brand voice already applied

Summaryโ€‹

Summary
  • Marketing copy prompts must define audience, brand voice, value proposition, and conversion goal
  • Always specify the content type (ad, landing page, email) and its format constraints (character limits, structure)
  • Use established copywriting frameworks (AIDA, PAS, BAB) to structure persuasive content
  • Request multiple variants with different angles for A/B testing
  • Use higher temperature (0.7โ€“0.9) for creative variety, lower (0.3โ€“0.5) for brand-consistent output