Most marketers are using 10% of what AI prompting can do.
Most people type one instruction and hope for the best.
No role.
No structure.
No constraints.
Then they wonder why the output feels generic and flat.
They rewrite the same prompt five times and get five average answers.
The gap usually isn't the AI.
It's the prompt.
I spent years writing briefs for creative and media teams.
Long documents.
Endless rounds of back and forth.
Still missing the mark half the time, even with a full page of notes.
AI prompting works exactly the same way.
Vague brief in, vague output out.
Once I started treating prompts like proper briefs, the output changed overnight.
Less editing.
Fewer rewrites.
Answers I could actually use first time.
That's the shift most marketers still haven't made.
Sixteen years in marketing taught me one thing about instructions.
The clearer the brief, the better the work, every single time.
AI runs on the exact same rule.
Give it role, context, constraints, and format, and the output jumps a level.
This isn't a hack.
It's just proper briefing, aimed at a machine instead of a person.
Most people skip straight to the ask and wonder why the answer feels thin.
The people getting real value from AI aren't smarter prompters.
They've just stopped winging it.
Twelve of my favourites from the thirty in this infographic:
→ Role Prompting
Give it an identity before the task.
→ Goal First Prompting
State the outcome before the instructions.
→ Context Injection
Add background before you ask the question.
→ Constraint Prompting
Set limits on tone, length, and style.
→ Audience Prompting
Tell it exactly who the output is for.
→ Step by Step Prompting
Break the task into a clear sequence.
→ Self Critique Prompting
Ask it to find its own weaknesses.
→ Chain of Verification
Make it check the facts before finalising.
→ Comparative Prompting
Get it to weigh options before recommending one.
→ Progressive Disclosure
Outline first, then expand section by section.
→ Ask Before Answer
Get it to ask you what's missing first.
→ Reflection Prompting
ask for one improvement before you accept the answer.
The other eighteen are in the infographic above, saved and ready to use.
Save yourself the trial and error I went through learning these the slow way.
Print it, screenshot it, whatever gets it in front of you next time you open a chat window.
More like this every week, if it's useful, stick around.
Send this to a teammate who's still typing one line prompts and hoping.
Which of these are you already using? Genuinely curious.
30 AI PROMPTING HACKS
You Can Use On ChatGPT, Claude, Grok, Gemini, Perplexity, Mistral, and other AI models, LLMs, and AI tools
1. Role Prompting
Explain: Assign the AI a specific expert identity before giving the task.
Example: "Act as a senior cybersecurity consultant and audit this system."
2. Goal First Prompting
Explain: State the exact outcome before providing instructions.
Example: "Your goal is to increase this email's reply rate to over 20%."
3. Context Injection
Explain: Give relevant background information before asking the question.
Example: "This article is for beginner entrepreneurs with no AI experience."
4. Few Shot Prompting
Explain: Show a few examples so the AI follows the same pattern.
Example: "Example 1... Example 2... Now generate Example 3."
5. Zero Shot Prompting
Explain: Ask the AI to perform a task without providing examples.
Example: "Summarize this report in five bullet points."
6. Delimiter Prompting
Explain: Separate instructions and data using clear markers.
Example: "Summarize only the text inside ```"
7. Output Format Prompting
Explain: Specify exactly how the response should be structured.
Example: "Return the answer as JSON with title, summary, and keywords."
8. Constraint Prompting
Explain: Specify exactly how the response should be structured.
Example: "Return the answer as JSON with title, summary, and keywords."
9. Audience Prompting
Explain: Tell the AI who the content is for.
Example: "Explain quantum computing to a 12 year old."
10. Step by Step Prompting
Explain: Ask the AI to solve the task one step at a time.
Example: "Break the business plan into clear sequential steps."
11. Self Critique Prompting
Explain: Ask the AI to review and improve its own output.
Example: "Now identify three weaknesses and improve the answer."
12. Multi Persona Prompting
Explain: Make multiple experts collaborate on one problem.
Example: "Answer as a marketer, lawyer, and product designer."
13. Reverse Prompting
Explain: Start with the desired output and work backwards.
Example: "What prompt would produce this exact result?"
14. Socratic Prompting
Explain: Make the AI ask questions before answering.
Example: "Ask five questions before creating the strategy."
15. Checklist Prompting
Explain: Give the AI a checklist to verify before responding.
Example: "Ensure accuracy, clarity, and actionable advice before replying."
16. Decision Matrix Prompting
Explain: Ask the AI to compare options using predefined criteria.
Example: "Compare these laptops based on price, battery, and performance."
17. Chain of Verification Prompting
Explain: Instruct the AI to verify important claims before finalizing.
Example: "Check for factual errors before giving the final answer."
18. Iterative Refinement Prompting
Explain: Improve the output through multiple revisions.
Example: "Create Version 1, then improve it twice."
19. Persona Simulation Prompting
Explain: Simulate a specific customer or user type.
Example: "Respond as a skeptical enterprise buyer."
20. Tree of Ideas Prompting
Explain: Generate multiple solution branches before selecting one.
Example: "Suggest five different marketing strategies, then recommend the best."
21. Negative Prompting
Explain: Clearly state what should not appear in the response.
Example: "Do not use jargon, emojis, or marketing buzzwords."
22. Evaluation Prompting
Explain: Ask the AI to score or rate its output using criteria.
Example: "Rate this landing page from 1 to 10 for conversions."
23. Comparative Prompting
Explain: Compare multiple alternatives before making a recommendation.
Example: "Compare Claude, ChatGPT, and Gemini for coding."
24. Progressive Disclosure Prompting
Explain: Reveal information in stages instead of all at once.
Example: "Give only the outline first, then expand each section."
25. Error Finding Prompting
Explain: Ask the AI to identify mistakes in text or code.
Example: "Find every logical error in this business proposal."
26. Hypothetical Scenario Prompting
Explain: Use imagined situations to test reasoning.
Example: "How would this strategy change if the budget were zero?"
27. Template Prompting
Explain: Force the AI to follow a reusable template.
Example: "Use Problem, Solution, Benefits, and CTA for every response."
28. Ask Before Answer Prompting
Explain: Require the AI to clarify missing information first.
Example: "If anything is unclear, ask questions before answering."
29. Rubric Prompting
Explain: Give detailed evaluation criteria before generating content.
Example: "Optimize this article for clarity, SEO, originality, and readability."
30. Reflection Prompting
Explain: Ask the AI to reflect on whether the response fully solves the task.
Example: "Review your answer and make one final improvement before submitting."
Most are using it as a ChatGPT replacement.
Claude’s real power is that it can replace your entire marketing system.
I spent 30 days going deep. Testing every feature. Breaking things. Rebuilding them.
What I found changed how I run marketing permanently.
Here are 9 Claude features most marketers don't know exist:
→ Projects — Your marketing command center. Upload brand voice, ICPs, positioning docs. Every chat remembers them. Every output sounds like your brand from prompt one.
→ Custom Instructions — Tell Claude exactly who your buyer is. "VP of Marketing, mid-market SaaS, 50-200 employees." Now every response is written for that person. Not some generic audience.
→ Extended Thinking — Turn this on for strategy work. I ran a competitive analysis using Playing To Win. The output read like a $20K consulting brief. Not a blog post.
→ Search — Claude pulls live data mid-conversation. I asked for our competitor's latest product launches. Had a competitive brief built without opening a browser tab.
→ Artifacts — I asked Claude to build a campaign ROI calculator. It built one I could edit live inside the chat. Input ad spend and ACV. Get projected pipeline instantly.
→ File Uploads — I dropped in 3 months of email data. Claude found the subject line patterns driving our highest open rates. Took 45 seconds.
→ Connectors — I linked Google Drive. Asked Claude to find our Q3 performance deck and summarize wins and misses. No uploading. No screenshots. Just answers.
→ Cowork — Pointed Claude at a folder of campaign assets. It read our positioning deck and built a full campaign brief as a Word doc. Right into the folder.
→ Styles — Created a custom style from my top 3 LinkedIn posts. Now every draft sounds like me without touching a word.
Last Tuesday I built a full campaign plan in one afternoon.
Messaging hierarchy. Channel strategy. 30 assets. Landing page wireframe.
One brief. One tool.
The marketers figuring this out right now will look like geniuses to their boards by year end.
If you want a hi res version of this, follow me and DM me “How to Claude” and I’ll send it to you.