AI Prompt Library for Marketers

What this is

A library of 42 prompts, organised by marketing function, that I use on live retainers. Each one is a structured system prompt with variables, constraints and a set output format, so you get something more usable than a “write me a blog post” request.

I’ve tested every prompt on both Claude (Sonnet/Opus) and GPT-4o, and each has notes on which model handles it better and why.

What is inside

Content strategy (10 prompts)

• Content brief generator: takes a target keyword and produces a full brief with angle, outline, sources to cite, internal links to include and a word count target
• Content refresh auditor: reads an existing post and tells you what to update, add or cut to recover lost rankings
• Topic cluster mapper: turns a pillar topic into a hub-and-spoke content plan with keyword targets for each spoke
• Editorial calendar builder: produces a 90-day content calendar from seasonal trends, competitor gaps and business goals
• Headline A/B generator: writes 10 headline variants scored on clarity, curiosity and keyword inclusion

Ad copy & conversion (8 prompts)

• Google Ads RSA generator: 15 headlines and 4 descriptions per ad group, within character limits
• Landing page copy framework: hero, social proof, features, objection handling and CTA, structured for above-the-fold impact
• Email subject line tester: generates 20 subject lines with its reasoning on likely open rates
• Meta description writer: batch-writes SEO meta descriptions aimed at CTR, within 155 characters
• Social proof synthesiser: reworks raw testimonials for landing pages, ads and email

Technical SEO (8 prompts)

• Schema.org generator: produces JSON-LD for any page type from a URL and content description
• Redirect map builder: takes old and new URL lists and writes redirect rules with regex patterns
• Robots.txt auditor: reviews robots.txt and flags crawl budget waste or accidental blocks
• Log file analyst: summarises crawl patterns from server logs, including bot frequency, crawl budget allocation and error rates
• Core Web Vitals diagnostician: takes PageSpeed Insights JSON and returns a prioritised fix list

Competitor analysis (8 prompts)

• Competitor content gap finder: compares your content inventory with a competitor’s and lists the topics you’re missing
• SERP feature analyst: for a given keyword, shows which SERP features appear and how to target each one
• Competitor positioning decoder: reads a competitor’s homepage or about page and pulls out their positioning, messaging hierarchy and weak points
• Backlink opportunity finder: given competitor backlink data, picks out the link sources you’re most likely to get too

Reporting & strategy (8 prompts)

• Monthly report narrator: turns raw GA4/GSC data into an executive summary with insights and recommendations
• QBR slide writer: writes presenter notes and talking points for quarterly business reviews
• KPI anomaly explainer: given a spike or drop in a metric, offers three hypotheses ranked by likelihood
• Budget reallocation advisor: takes channel performance data and recommends budget shifts with expected impact

How to use it

Every prompt follows the same structure:

[SYSTEM] Role and constraints
[CONTEXT] Background information you paste in
[VARIABLES] {{keyword}}, {{url}}, {{competitor}} — replace with your data
[OUTPUT FORMAT] Exactly how the response should be structured
[QUALITY GATES] Self-check criteria the AI applies before responding

Best practices

• Always fill in the context section. Generic context gets you generic output.
• Keep the quality gates in. They catch the most common AI failures: made-up facts, generic advice and missing specifics.
• Edit instead of regenerating. If the first output is 70% right, fix the other 30% rather than starting over.
• Each prompt has its own Claude vs GPT note. In general, Claude is better with nuance and longer outputs, and GPT-4o is faster for batch work.

Version history

Version	Date	Notes
3.0	June 2026	Added quality gates to all prompts, updated for Claude Opus 4
2.5	April 2026	Added competitor analysis section
2.0	January 2026	Restructured around marketing functions