Key AI Prompting Techniques

Most people think getting good results from AI is all about choosing the right model. In reality, the biggest breakthroughs often come from how you ask, not what you ask.
To really unlock an AI's potential, it helps to master the craft of clear, structured prompting. Whether it's formatting your questions with bullet points, using examples, or refining outputs in stages, the way you shape the conversation determines the quality of the outcome. Here are a few methods that consistently elevate results when working with AI:

1. Chain-of-Thought Prompting
Get the AI to reason step by step before giving an answer. This leads to more thoughtful recommendations and helps ensure decisions are aligned with your priorities.

Prompt:
“We’re launching a new product. Walk through the factors we should consider before recommending a marketing channel.”

Example:
“You mentioned a limited budget, a younger target audience, and the need for measurable results. Paid search might be too expensive upfront, and traditional media doesn’t offer detailed tracking.
Social platforms like Instagram or TikTok offer strong reach among younger demographics and allow for performance-based targeting.
Recommendation: Focus initial efforts on TikTok ads.”


2. Audience Persona Pattern
Tailors responses based on the user’s background, attention span, or expertise.

Prompts:
“Explain LLMs to me as if I have no background in computer science.”

“Explain LLMs to me, I get distracted easily so I need engaging content to stay focused (e.g images/animations/clearly formatted text).”


3. Better Question Suggestion Pattern
Encourages the AI to refine or improve user questions for clarity or intent.

Prompt:
“Whenever I ask a question, suggest a better version.”
(Optional) “Ask if I want to use the better version instead.”

Tailored Examples:
“If I ask who the greatest footballer is, reframe to compare specific accomplishments.”
“When I ask about dieting, suggest questions focusing on healthy eating habits.”


4. Scoring Pattern / Self-Evaluation Prompting
It gets the AI to:
- Use a rubric or scoring criteria.
- Evaluate (and optionally explain) how well something meets those criteria.
- Output a total or weighted score and ranking.

Basic Version:
“Evaluate the following [text/ideas/options] using this scoring rubric:
- Criterion A (out of X points),
- Criterion B (out of Y points),
- Criterion C (out of Z points).
Total the score out of [X+Y+Z] and explain the score for each criterion.”

Advanced Version:
“Score each option below using the following criteria. Assign a score 1–10 for each and give reasoning:
- Relevance
- Originality
- Clarity
- Overall Usefulness
Rank the options based on their total score out of 40.”