Prompting Best Practices
Prompting Basics
Understand the Process
Understanding how Large Language Models (LLMs) work is crucial for crafting prompts that lead to the most usable and relevant outcomes. When you're unsure about what's happening under the hood with ChatGPT, Gemini, and Claude, using them can feel like navigating a complex system blindly, which often leads to frustration and subpar outputs. A quick review of our How Does an LLM Work page makes the general concept easy to grasp and will help you get more out of the information below.
Strive for Clarity Balanced with Details
Your mission is to find the sweet spot between details and clarity. When prompts are vague, ambiguous, or do not provide enough details, the AI will make assumptions. Long, complicated, and layered descriptions, requests, and examples in a prompt can confuse the LLM.
Keep it simple and have a conversation with the AI. It’d be counterintuitive to dominate a conversation with a human by overloading them with repetitive details and multiple explanations. Longer, detail-heavy prompts will rarely get you better results. Clarity beats complexity.
Likewise, even a slight lack of clarity could cause confusion. For example, when you want to know what day of the week someone was born on, and ask, “What day were you born?” The probability of them responding with a month, day, and year is nearly 100%. So, you clarify and ask again. Same thing here.
Remember, the LLM’s response to your prompt is determined by you. The often amazing, mind-blowing response you receive is a direct result of what you created. You utilized a creative tool to achieve a specific result.
Prompt Frameworks
You’ll find these frameworks to be helpful for specific outcomes, but keep in mind that they should be used when you have a clear objective, such as a script, a report, a list, or instructions delivered in a specific format, with multiple versions, etc.
The majority of your experiences with prompting any LLM will be conversational. When you're looking for information or want to generate ideas, explore new concepts, create lists, or similar tasks, a free-flowing conversation with the LLM is best.
The Best General Use Framework For ChatGPT, Claude, and Gemini
1. Role: Instruct the AI to"act as" or "take on the role of" a persona, or character, to influence the style, tone, and content of its response
2. Context: What’s the task? Who’s it for? What conditions should be adhered to?
3. Goal: This is the ‘why.’ A clear description of your goal deepens clarity, which in turn refines the response.
4. Output Rules: Specify tone, format, word count, or any boundaries.
Example:
Role: You are an expert social media strategist with a great sense of humor.
Context: Outline a four-week creative social content plan that increases tune-in for our station’s weekday afternoon“Drive at Five.”
Goal: Make our audience aware that we hand-pick the perfect songs for their drive home, as that may be the only thing that motivates them to go back to work tomorrow.
Output Rules: The outline should include ideas for posts with a message that evolves in a natural and entertaining way over a four-week. Include the best times to post on different social platforms.
The Mack Truck of Frameworks
Basic, reliable, easy to customize, good for generalized structured responses.
1. Context: What’s the task? Who’s it for? What should it achieve?
2. Guardrails: Specify tone, format, word count, or any boundaries.
3. Reference: Include an example or explain what good looks like.
Example:
Context: Make two lists of things parents think about; 1 - on their child’s first day of kindergarten, and 2 - on their child’s first day of their senior year in high school.
Objective: Make it relatable with sharp humor, do not use profanity.
Style/Tone: Format the lists side by side, like:
Kindergarten: Senior Year:
I hope they’ll be OK I hope they get a job
CO-STAR
(Context → Objective → Style → Tone → Audience → Response)
Best for general tasks that require clear, well-defined outputs.
Context: The background facts, inputs, and constraints the model must consider
Objective: The single outcome you want (what success looks like).
Style: Writing characteristics (e.g., concise, story-driven, technical).
Tone: Emotional vibe (e.g., friendly, authoritative, playful).
Audience: Who it’s for and their knowledge level.
Response format: The exact structure/length/filetype you want back.
Example:
Context: You’re writing a promo for a country radio station.
Objective: Write a 30-second script teasing a new contest announcement.
Style/Tone: Spark curiosity/high energy, fun.
Audience: Fans of our station and country music.
Response: 3 alt scripts, each with a one-line hook to open, and a close driving Friday 7 AM tune-in.
RTCCO
(Role → Task → Context → Constraints → Output)
Best for achieving precise and predictable deliverables.
Role: The hat the model wears (DJ, Social Media writer, Producer)
Task: The concrete action to perform (write, compare, outline).
Context: Relevant details, inputs, and situational info.
Constraints: Non-negotiable rules (length, do/don’t, citations, sources).
Output: Required shape of the result (bullets, table, script, article, etc.)
Example:
Role: Marketing copywriter.
Task: Draft a 30 video script.
Context: Find out how to win $500 five times a day from (station), listen on Friday at 7 AM.
Constraints: Make script VO fill 20 seconds total, include two 5-second listener drops, upbeat tone in
plain English, no buzzwords, no cheesy radio-sounding language, suggest visuals.
Output: 3 script versions, attention-grabbing opener, 3 CTA options throughout all scripts.
AIDA
(Attention → Interest → Desire → Action)
Best for promos, spots, social posts, landing page, app notifications.
Attention: Hook
Interest: Why it matters, The emotion it sends
Desire: Benefits/proof
Action: CTA
Example:
Attention: Write a Facebook post for a $500 contest
Interest: Free money for the taking, splurge!
Desire: $500 goes a long way if you know how to have fun.
Action: Learn how to win, Friday 7AM
PAS
(Problem → Agitate → Solve)
Best for short, persuasive copy and intros.
Problem: The core pain or friction.
Agitate: Make the impact feel real (stakes, consequences).
Solve: Present the solution and how to start.
Example:
Problem: You missed (morning show bit) today
Agitate: Very sad for you, it was the best one we’ve ever done.
Solve: Lucky you! We posted the entire bit on our website, mobile app, and Facebook.
RATT
(Role → Audience → Topic → Task)
Best for quick, lightweight prompts when you’re in a hurry.
Role: The perspective/expertise to adopt.
Audience: Who will consume it and what they know.
Topic: The subject or theme to cover.
Task: The action/output required on that topic.
Example:
Role: PD/Ops Mgr.
Audience: New on-air staff.
Topic: Localization tips.
Task: Write a 200-word cheat sheet with 5 do’s and 3 don’ts.
Prompt Templates
Copy, Customize, Deploy
-
Using the playbooks of top business influencers with huge email lists, build a 180 day plan I can use to add 10k email subscribers to my list. Whenever the influencers refer to their niche or industry in their strategy, please use the broadcast radio industry.
-
We're stuck. Our design thinking group has been brainstorming for 20 minutes and we keep circling back to the same predictable solutions.
Here's our challenge:
[Describe the specific problem you’re working on, include:
-what needs to be created or fixed?
-why?
-who benefits?
-what roadblocks are you facing?
-what happens if you fail?
-what happens if you succeed?
-(others) ]Here are the ideas they've generated so far: [List current ideas]
I need you to be the devil's advocate and breakthrough catalyst. Please:
1. Challenge one key assumption underlying their current ideas
2. Suggest 3 completely different angles they haven't considered
3. Pose one provocative question that could shift their entire approach
4. Recommend a specific exercise or constraint that might unlock fresh thinking
Help me get this group past surface-level solutions to something that could actually make a difference.
-
I create content for [very detailed description of your target listener*]. Provide me with 30 problems that these people are likely experiencing right now. Include a mix of practical problems and emotional ones. Complete the entire list and review it to ensure there are no duplicates. When you have completed the first list, then make a second list of ideas about how to solve or lessen the problem. Provide at least three ideas for each problem.
*The more detailed you are with the description of the listener, the better the list and ideas will be. Consider the season, your market, life stages, etc. -
You are a customer research expert, now working for me at my radio station, as my assistant. I want you to do audience research for me. Use my target listener persona and business profile for reference, provided below.
I want to know 10 frustrations, 10 deepest desires, 10 hidden thoughts, 10 past failures, 10 burning questions, 10 common enemies and 10 fears that my target audience experiences that relates to [What you want to talk about, provide as many details as possible].
Do the above while mentioning the awareness level of that group of audience in detail.
[Copy + paste your target listener persona and business profile here.]
-
Headshot Prompts for ChatGPT
There are four prompts here, customized for Facebook, X, TikTok, and LinkedIn. You’ll see some Power Ups below, too.Prompt for Facebook
Edit the attached image into a professional-but-casual Facebook profile headshot. Preserve 100% of the person’s facial features and expression (face shape, hairline, skin tone/texture, eyes, nose, mouth/teeth, facial hair, ears).
• Lighting/scene: Natural outdoor lighting (soft, open shade or golden hour). Shallow depth-of-field bokeh with a blurred everyday background (park/sidewalk/café—no landmarks).
• Wardrobe: Switch from suit to casual wear—well-fitted crew-neck tee or casual button-down/henley (solid medium blue/olive/charcoal), no logos.
• Framing: Chest-up, camera at eye level, eyes on the upper third, subtle headroom.
• Color/retouch: Neutral white balance, gentle contrast, preserve skin texture, minimal cleanup only (no face reshaping or age changes).
• Output: 4:5 portrait crop, sRGB JPG, high quality; also provide one alternate with a different casual top/background variation.
Prompt for X (dynamic + dramatic)
“Edit the attached image into a high-resolution dynamic headshot. Preserve 100% of the person’s facial features and expression (face shape, hairline, skin tone/texture, eyes, nose, mouth/teeth, facial hair, ears).
• Lighting/scene: Dramatic side lighting (soft “short” or Rembrandt-style; defined shadow falloff), subtle vignette. Minimalist urban background (out-of-focus concrete, glass, or steel; clean lines; no logos/landmarks).
• Wardrobe: Modern business casual—well-fitted knit polo or open-collar button-down with an optional unstructured blazer (solid charcoal/navy/olive); no tie, no loud patterns.
• Framing/pose: Chest-up, camera at eye level, shoulders at slight angle, confident posture, eyes sharp; hint of a smile to convey sharp intellect with subtle humor.
• Color/retouch: Neutral/contrast-forward grade; keep skin texture natural. Minimal cleanup only (stray hairs, lint). No face reshaping or age changes.
• Output: Square 1:1, sRGB JPG, high quality (≥1500×1500). Provide one alternate with a slightly different urban background and blazer on/off.”Prompt for TikTok (vibrant + playful)
“Edit the attached image into a vibrant, playful headshot. Preserve the person’s exact facial features and expression (face shape, hairline, skin tone/texture, eyes, nose, mouth/teeth, facial hair, ears).
• Lighting/background: Colorful, energetic lighting with soft rim or gel accents; fun gradient backdrop (e.g., teal→purple or coral→gold). Subtle bokeh/confetti-style light blobs acceptable, but keep it clean.
• Wardrobe: Trendy casual—well-fitted crew or henley, or lightweight bomber over a tee (solid mid-tones; avoid logos; tasteful texture OK).
• Framing/pose: Chest-up, eye-level camera, lively yet natural smile; micro-tilt of shoulders for energy.
• Color/retouch: Punchy saturation and crisp contrast while keeping skin tones realistic. Minimal cleanup only; no alteration of facial structure or age.
• Output: Square 1:1, sRGB JPG, high quality (≥1080×1080). Also provide a second version with a different gradient palette.”Prompt for LinkedIn (studio + formal)
Edit the attached image into a professional LinkedIn headshot. Preserve 100% of the person’s facial features and expression (face shape, hairline, skin tone/texture, eyes, nose, mouth/teeth, facial hair, ears).
• Lighting/background: Studio-quality soft key with gentle fill and catchlights; soft neutral background (light gray or warm gray, clean and even).
• Wardrobe: Formal professional—tailored suit jacket (navy/charcoal), crisp light shirt, optional understated tie; no lapel pins or loud patterns.
• Framing/pose: Chest-up, eye-level, relaxed shoulders, confident and approachable smile; eyes on the upper third with subtle headroom.
• Color/retouch: True-to-life white balance; refined but natural finish. Minor blemish/lint cleanup only; no facial reshaping or age reduction.
• Output: Square 1:1, sRGB JPG, high quality (≥2000×2000). Provide a second crop at 4:5 for versatility.Power ups:
Ask ChatGPT to give you color palette suggestions to match your skin tone.
Change or add image dimensions to get perfectly sized images for any need.
Customize any part of the prompt to make it more ‘you.’
-Make me hold a microphone-Add recording studio headphones
-Make me wear mirrored aviator sunglasses.
-Make the background a long shot of an outdoor amphitheater concert at dusk.
-Add these quick fit & styling tips where needed:
Choose mid-tone solids; they separate nicely from your skin and hair.
If you wear glasses, anti-glare helps with dramatic or colorful lighting.
Keep contrast near the face (darker jacket, lighter shirt) for LinkedIn/X; go single-layer mid-tone for Facebook/TikTok.
Avoid bright yellow/green near the face; they can cast color on skin/teeth.
-
I want to improve how I delegate tasks effectively within my team. Please provide me with a detailed step-by-step guide on identifying which tasks to delegate and selecting the right team members for those tasks. Also please include tips on communicating the delegation in a clear, motivating style and clearly and following up to ensure tasks are progressing smoothly.
-
I would like a comprehensive breakdown of the marketing strategies utilized by my key broadcast radio competitors in [your market]. Please analyze their online presence, including social media engagement, content marketing efforts, and any notable promotional campaigns. Additionally, evaluate their use of SEO, AEO, and paid advertising. Highlight what appears to be working well for them and areas where they may be lacking. Based on your analysis, suggest innovative marketing tactics that I could implement to stand out and attract more customers.
[Your brand name]
[Competitor brand name #1]
[Competitor brand name #2]
[Competitor brand name #3] -
This social media post did well: [the post]. Identify the factors that contributed to its success, and then develop 15 post ideas that incorporate the same factors. Make them distinctly different from the original.
-
You are an expert marketing copywriter skilled at creating engaging, clever and persuasive promo scripts for radio station featured content promos. Your task is to craft three versions of a compelling audio promo script (30 seconds, 15 seconds and 10 seconds) that quickly captures attention, highlights key features, and ends with a clear call-to-action.
Context:
You are writing for an audience that is unfamiliar with the product and may have short attention spans. The product could be specialized programming content (shows, contests), digital platforms (mobile app, social media, website) or on location station events (live broadcasts, concerts) Assume the script will be used live on the air so it must be fast-paced, upbeat, fun and highly engaging.Structure:
Start with a hook (a question, bold statement, or relatable problem).Introduce the product/content and its main benefit.
Focus on 2-3 key features or advantages.
End with a strong call-to-action (CTA), urging listeners to tune in at a specific day and time.
Tone:
Use an energetic, confident, and relatable tone to keep listeners excited and engaged.
Format:
Write in short, punchy sentences suitable for radio, written to cut through the noise. Aim for around 75–90 words maximum.
Content:
[add yours] -
Role: You are a seasoned, talented content innovation consultant and creative strategist with experience helping top radio brands solve complex creative challenges.
Context: You are facilitating a brainstorming session to generate innovative ideas and solutions for content idea and new angle challenges.
Instructions: Generate diverse, actionable ideas for the given challenge or opportunity, thinking from multiple angles and considering various constraints.
Constraints:
Provide at least 8-10 distinct ideas
Mix practical and creative approaches
Consider budget, time, and resource limitations
Include both short-term and long-term solutions
Offer creative suggestions for funding ideas beyond the scope of the budget
Output Format:
## Quick Wins (Low effort, low or no cost, immediate impact):
- [Idea 1]
- [Idea 2]
- [Idea 3]
## Big Ideas and Solutions (Moderate investment):
- [Idea 4]
- [Idea 5]
- [Idea 6]
## Big Swings (Higher impact, long lasting effects):
- [Idea 7]
- [Idea 8]
## Wild Cards (Unconventional, left of center approaches):
- [Idea 9]
- [Idea 10]
Reasoning: Use divergent thinking combined with constraint-based reasoning - first generate without limitations, then apply practical filters. Employ analogical reasoning to draw inspiration from different domains.
User Input: [Describe your challenge, goal, or brainstorming topic]
How to Help the LLM Help You
Prompting is far from a one-size-fits-all experience. You still need to stay in a ‘learn and adapt’ mindset, but generally, prompting continues to get easier thanks to the improved reasoning capabilities of the newest large language models (LLMs) used by ChatGPT, Claude, Gemini, and others.
Contrary to what you may have heard, there is no single “best” way to write a prompt. We’ve included prompt frameworks above because they can be a significant help and have been proven to work. We’ve learned some AI users may find them somewhat cumbersome. Trying to accurately craft prompt structure “ingredients” can slow the process down for some people, which is the exact opposite of what you should expect when using AI.
To that end …
The more you understand how the AI platform, tool, or LLM you're using was taught to absorb and interpret what you’re asking it to do, the better you become at prompting. Try these techniques a few times before relying solely on a prompt structure:
Prompt AI to improve your prompt
This is the single most effective way to learn how to get more useful responses and actions. Enter a prompt like this: “I’m trying to accomplish (X). This is the prompt I’ve created.” (insert the prompt you intend to use) “How would you improve this prompt to get a better answer from you?”
This technique makes the LLM reveal what it knows to you, so you can apply that understanding in the future. You’re basically asking it to show you what it needs in order to provide responses you can use. You’re using a prompt in your prompt to get a prompting lesson.
Ask for a clarity check
“This is the prompt I’ve created.” (insert the prompt you intend to use). “What parts of this request are unclear, vague, or ambiguous? What assumptions are you making? Is there additional information that would help you improve the accuracy of your response?”
By asking these questions, you’re making the LLM reveal its uncertainties that it would otherwise infer or assume, so it can proceed. But now you’ve got it talking about the info, data, examples, context, and other things you may not have realized it needed. Asking for a clarity check allows you to fill in the blanks you didn’t know existed, rather than turning those decisions over to the LLM (which can be the source of AI hallucinations).
Capability discovery
This is a great technique to use when you've prompted a few times and feel like you're banging your head on the wall because the LLM’s responses aren’t what you're looking for. The prompt: “How would you approach this if you had no constraints? What would be your ideal process? What tools or information would help?”
This technique can sometimes reveal prompting approaches you hadn’t thought of and information you can provide to help improve its output. Asking more questions about the tools and information the LLM mentions, like, “Tell me how you’d use that information,” or, “Why is that tool helpful?” can help the LLM draw a clearer picture of how you can help it help you.
Request a step-by-step plan before asking for execution
This allows the LLM to lay out its thinking for you to critique and adjust before it begins working on your request. This can make a big difference in the outcome of a layered or detail-heavy prompt. This concept is similar to the technique known as Chain Prompting: asking the LLM to ‘think step by step’ in a single prompt that includes many planning and execution requests. Chain prompting is quite effective, but we also recommend breaking chain prompt requests into individual prompts, allowing you to review the responses one by one, and, if needed, modify them or ask questions before moving to the next step of your request.
Chain Prompt Example:
"Write a research paper on the impact of social media on mental health. First, outline the key sections of the research paper (introduction, literature review, methodology, results, discussion, conclusion). Second, suggest a logical flow for each section. Third, suggest potential research questions and hypotheses. Finally, create an outline with key points for each section."
Plan-Solve Example:
Prompt #1 - Help me create a plan for a research paper on the impact of social media on mental health. First, outline the key sections of the research paper (introduction, literature review, methodology, results, discussion, conclusion).
Review the response, ask for changes or explanations. Move to Prompt #2 when approved.
Prompt #2 - Suggest a logical flow for each section.
Review
Prompt #3 - Suggest potential research questions and hypotheses.
Review
Prompt #4 - Create an outline with key points for each section.
Considering the nature of these four prompts, one can see how the outline being created in Prompt #4 will be clearly defined and on point with the user’s expectations, as they were able to guide the LLM’s thinking and decisions on the key sections, logical flow, and research.
Pro Tips
Stop typing. Start talking.
Our brains move faster than we can type. Because ChatGPT, Claude, Gemini, and others are built to work with plain language prompts, simply clicking the microphone and talking to them is the fastest and easiest way to get what you're looking for and pick up new skills
If you’re uncomfortable recording your voice directly into any AI platform, use the speech-to-text function on your computer or phone. Five years ago, it seemed odd to see someone looking at their phone while they talked to it to write a text message. It's become the norm today and should be the norm when you're prompting AI.
Choose Your AI Wisely