Same AI. Same question. Completely different answer. Here's exactly why — and how to use it.
Two people open the same AI tool at the same time. They want the same thing — a compelling Instagram caption for a new product launch. The first person types their request and gets something generic, something they've essentially seen a hundred times before. The second person uses four lines of structured instruction and gets back a caption so on-brand and well-crafted that they paste it directly into the post without changing a word.
Same model. Same capability. One result that works, one that doesn't.
The difference is not luck. It's not a paid subscription. It's not some hidden setting buried in the menu. It's the structure of the prompt — specifically, whether it includes the four elements that tell an AI not just *what* to produce, but *who* to be, *why* this matters, *exactly* what's needed, and *what the limits are*.
That structure has a name: Role, Context, Task, Constraint. And this walkthrough is going to prove — with real text comparisons — why each element changes the output, what happens when you leave one out, and how to use the complete formula from the very next prompt you write.
The Problem With Open Questions
Before the formula, an honest look at what most beginners do.
An open question is a prompt with no structure beyond the basic ask:
> "Write a caption for my product launch."
Or:
> "Give me advice on starting a business."
Or:
> "Explain machine learning."
These aren't wrong. They're just incomplete. When an AI receives an open question, it has to make assumptions about everything the prompt didn't specify. Who is this for? What tone? What length? What does the person already know? What do they want to do with the answer? What should it avoid?
With no guidance, the AI chooses the most probable answer to all of those questions simultaneously — which means it defaults to the most average version of the response. The middle of the bell curve. The safest, most common interpretation.
And the most common interpretation is, by definition, the same response everyone else is getting.
The formula doesn't change what you're asking for. It changes how much the AI knows about what you actually need — and that changes everything.
Element One: Role
What it is: The identity, expertise, or professional perspective you assign to the AI before it starts.
What it does: Changes who is speaking. Shifts the vocabulary, depth, assumptions, and authority of the response.
The syntax: "Act as a [specific role]." or "You are a [specific role]."
The Text Block Proof
Look at these two prompts and their outputs side by side.
PROMPT A — No Role:
> Write me a tip for first-time job seekers.
OUTPUT A:
> When looking for your first job, make sure to update your resume with relevant skills and experience. Tailor your applications to each role and don't forget to follow up after interviews. Networking can also be a valuable tool for finding opportunities.
PROMPT B — With Role:
> Act as a career coach who has helped hundreds of recent graduates land their first professional roles. Write me one tip for first-time job seekers.
OUTPUT B:
> The biggest mistake first-time job seekers make is treating applications like a numbers game — sending the same generic resume to fifty companies and wondering why nobody calls back. Stop. Pick ten companies you'd genuinely be excited to work for. Research each one until you can speak about their challenges, their culture, and their recent wins. Then write an application that makes it obvious you did. One personalized application to a company that feels right will outperform fifty generic ones every single time.
Read those two outputs again. Both answered the question. But Output A sounds like a list of obvious things anyone might say. Output B sounds like someone who has watched this mistake happen dozens of times and developed a specific, hard-won opinion about it.
The role didn't change the topic. It changed the position from which the AI spoke — and that single change transformed a generic tip into a pointed, confident piece of advice with a real perspective behind it.
Why this happens: AI models are trained on vast quantities of human text, including content written by professionals in every field imaginable. When you assign a role, you're not creating a fictional character — you're activating a specific cluster of language patterns, reasoning styles, and knowledge domains associated with that expertise. The model draws on how career coaches actually write and speak and filters the response through that lens.
Role examples that reliably improve output:
| Your Goal | Strong Role Assignment |
|---|---|
| Marketing copy | "Act as a direct-response copywriter with 15 years of ecommerce experience" |
| Business advice | "Act as a startup founder who has raised venture funding and sold a company" |
| Simple explanation | "Act as a teacher explaining this to a curious 12-year-old" |
| Critical feedback | "Act as a senior editor who has no patience for weak writing" |
| Financial thinking | "Act as a CFO stress-testing this business decision" |
| Legal understanding | "Act as a lawyer explaining what this clause actually means in plain language" |
The role doesn't have to be a job title. It can be a perspective: *"Act as someone who tried this and failed"* or *"Act as the most skeptical person in the room."* Any position that creates a defined point of view improves on no position at all.
Element Two: Context
What it is: The background information that makes the response relevant to *your specific situation* rather than a general one.
What it does: Replaces the AI's assumptions with your actual facts. Narrows the response from the generic to the specific.
The syntax: One to three sentences describing who you are, who the audience is, and what the situation is.
The Text Block Proof
PROMPT A — No Context:
> Act as an experienced email copywriter. Write a follow-up email to a potential client.
OUTPUT A:
> Subject: Following Up on Our Previous Conversation
>
> Hi [Name],
>
> I hope this message finds you well. I wanted to follow up on our recent conversation and see if you had any questions about our services. We'd love the opportunity to work with you and help you achieve your goals.
>
> Please don't hesitate to reach out if you need any additional information. Looking forward to hearing from you!
>
> Best regards,
> [Your Name]
---
PROMPT B — With Context:
> Act as an experienced email copywriter. Context: I'm a freelance video editor. Two weeks ago I sent a proposal to a boutique fitness brand for a series of six social media videos. The proposal was $2,400 total. They seemed enthusiastic in our call but haven't responded to the proposal email. Write a follow-up email.
OUTPUT B:
> Subject: Quick check-in on the social video series
>
> Hi [Name],
>
> Wanted to check in on the video proposal — it's been a couple of weeks and I know things get busy.
>
> I'm still excited about this project. The six-video format we discussed would give you a strong foundation for the next quarter's social content, and I've got availability in my schedule that would work well for your timeline.
>
> If you have any questions about the proposal or want to adjust the scope, I'm happy to talk through it. Just let me know.
>
> [Name]
---
The difference is stark. Output A is a template — something that could be sent by anyone about anything. It's so generic it actually signals to the reader that minimal effort was made. Output B is a real email: specific about the project, warm without being desperate, confident in the work, and moves toward a specific outcome.
That specificity came entirely from context. The AI didn't invent any of those details — it used the ones you provided.
What good context always includes:
Who you are: Your relevant role or relationship to this task. First-time founder? Experienced teacher? Freelancer? Employee?
Who the audience is: Who will read, hear, or use what the AI produces? Age, expertise, relationship to you, what they already know.
The specific situation: What happened, what you're trying to achieve, where this will go, what constraints the real world is placing on this.
Facts the AI can't know: Numbers, names, history, specifics that only you have access to.
Context is the difference between the AI guessing about your situation and actually knowing it. Three sentences of specific, honest context will improve output quality more than almost any other single change.
Element Three: Task
What it is: The precise description of what you want produced — the format, structure, length, and specific deliverable.
What it does: Tells the AI exactly what object to make, not just what topic to address.
The syntax: Specific nouns and verbs. Format. Length. Structure. What it should accomplish.
The Text Block Proof
PROMPT A — Vague Task:
> Act as a business strategist. Context: I run a small handmade soap business selling at local markets and online. Write something to help me grow.
OUTPUT A:
> Growing a small business requires a multi-pronged approach. Consider expanding your online presence through social media platforms like Instagram and TikTok. You might also want to look into wholesale opportunities with local boutiques. Building an email list of loyal customers can provide a direct line of communication and drive repeat purchases. Don't overlook the power of collaborations with complementary brands...
(continues for several more paragraphs covering every possible growth strategy at surface level)
PROMPT B — Specific Task:
> Act as a business strategist. Context: I run a small handmade soap business selling at local markets and online. My revenue is around $1,500/month. I have 8 hours a week available for business development. Write a numbered list of exactly three specific actions I should take in the next 30 days to increase monthly revenue — each action explained in two sentences maximum.
OUTPUT B:
> 1. Launch a "Market Regular" loyalty card. Create a simple paper stamp card offering a free bar after ten purchases — regulars become repeat buyers and tell friends, costing you almost nothing to implement this weekend.
>
> 2. Email your past online customers with a "back in stock" message. Write a 100-word email this week announcing your two most popular scents — past buyers who already trust your product are your lowest-effort path to an immediate revenue bump.
>
> 3. Approach two local boutiques with a wholesale minimum order. Identify two shops in your area whose customers overlap with yours, email a one-page sell sheet, and offer a minimum opening order of $150 — one wholesale account can add $300–$600/month with no additional market time.
Output A tried to be helpful by covering everything. Output B is useful because it covers exactly what was asked — three specific actions, explained concisely, with enough detail to actually take action on Monday morning.
The task didn't change the topic. It changed the *shape* of the answer — and the shape of the answer is what determines whether you can actually use it.
Task specification checklist:
- Format: What form should this take? Email, numbered list, table, paragraph, script, outline, caption, bullet points?
- Length: How many words, sentences, items, or paragraphs?
- Structure: "Start with X, then Y, end with Z"
- Purpose: What should this accomplish for the reader or for you?
- Specificity level: Detailed and actionable, or high-level overview?
When you're specific about the task, you're essentially designing the deliverable before the AI produces it. The output slots into what you actually needed.
Element Four: Constraint
What it is: The rules, restrictions, and limits that keep the output inside the boundaries that matter to you.
What it does: Prevents the most common frustrations before they happen. Focuses the AI on what matters by telling it what to exclude.
The syntax: A short list of what to avoid, what not to include, what limits apply.
The Text Block Proof
PROMPT A — No Constraints:
> Act as an experienced copywriter. Context: I'm launching a new line of sustainable water bottles. My customers are environmentally-conscious millennials. Write a product description for a 32oz insulated bottle made from recycled stainless steel.
OUTPUT A:
> Introducing our revolutionary new 32oz insulated water bottle — the perfect sustainable choice for the eco-conscious individual! Crafted from premium recycled stainless steel, this luxurious bottle keeps your beverages cold for 24 hours and hot for 12. Whether you're hitting the gym, heading to the office, or embarking on an outdoor adventure, this game-changing bottle is your perfect companion. It's not just a water bottle — it's a statement. A statement that you care about the planet and refuse to compromise on quality. Join the movement. Hydrate responsibly. Live sustainably.
PROMPT B — With Constraints:
> Act as an experienced copywriter. Context: I'm launching a new line of sustainable water bottles. My customers are environmentally-conscious millennials. Write a product description for a 32oz insulated bottle made from recycled stainless steel. Constraint: Maximum 60 words. No exclamation marks. Avoid the words "revolutionary," "game-changing," "perfect," and "luxurious." Don't open with "Introducing." The tone should be confident and direct, not enthusiastic.
OUTPUT B:
> Made from 100% recycled stainless steel, this 32oz bottle keeps cold drinks cold for 24 hours — no ice required. Double-wall insulation, a leak-proof lid, and a design built to last ten years, not one season. Because the most sustainable product is the one you never have to replace.
Output A hits every single cliché in sustainable product marketing. "Revolutionary." "Game-changing." "Perfect companion." "Join the movement." It's the language of a hundred other eco products — which means it disappears into the noise.
Output B is confident, clean, and ends with a line that's actually memorable. The constraints didn't restrict the creativity — they focused it by removing the lazy defaults.
Most useful constraint types:
Length limits: "Under 100 words." "Exactly three paragraphs." "No more than five bullet points."
Tone restrictions: "No corporate jargon." "Conversational, not formal." "Serious and direct."
Banned words/phrases: List the specific words or phrases you never want to see — especially common AI output defaults like "delve," "crucial," "seamlessly," "it's worth noting," "in today's world."
Format rules: "No bullet points — prose only." "No headers." "No bold text."
Audience rules: "Write as if the reader has never heard of this topic." "Assume the reader is already convinced — skip the persuasion."
Content restrictions: "Don't mention competitors." "Don't make any specific income claims." "Avoid medical advice language."
The Complete Formula in Action
Here's the RCTC formula applied in full — all four elements, combined, in one prompt:
The scenario: You run a small online bakery and want an Instagram caption for a new seasonal product.
The complete RCTC prompt:
> [Role] Act as a social media copywriter who specializes in food brands and understands how to write captions that drive saves and shares without sounding like an ad.
>
> [Context] I run a small online bakery with 4,200 followers. My audience is mostly women aged 25–40 who love home baking and supporting small businesses. My tone is warm, real, and slightly playful — not polished or corporate.
>
> [Task] Write an Instagram caption for a limited-edition brown butter apple crumble that I'm launching for autumn. The caption should feel like it's coming from a real person who made something they're excited about, not a marketing department.
>
> [Constraint] Maximum 80 words. No emojis. No hashtags (I'll add my own). Don't open with "Introducing" or "We're excited to announce." Avoid the word "delicious" and any sentence that starts with "This."
Output:
> Brown butter apple crumble is back, and honestly — it's the smell that gets me every time. That toasty, nutty warmth from the browned butter combined with the tartness of the apples hitting a hot oven. We only make these in small batches while good apples are still around. Link in bio to grab yours before they're gone.
That's a caption that sounds like a real person, feels warm without trying too hard, creates urgency without being pushy, and fits the voice of a small bakery talking to people who care about food.
It took one structured prompt.
Your Copy-Paste Template
Save this. Use it for any prompt where the output actually matters.
[Role] Act as a [specific expertise, role, or perspective].
[Context] [Who you are. Who the audience is. The specific situation. Relevant facts the AI can't know.]
[Task] [Exactly what you want produced. The format. The length. The structure. What it needs to accomplish.]
[Constraint] [What to avoid. Word limits. Tone rules. Banned phrases. Format restrictions.]
Fill in each section before you type any serious prompt. Not because the formula is magic — but because the formula forces you to think clearly about what you actually need before you ask for it. And when you know what you actually need, the AI can give it to you.
That's the whole secret.
Build a prompt using the RCTC formula and share your before-and-after result in the comments. We'd love to see what changes for you.
0 Comments