If you’ve ever typed “write me a marketing email” into ChatGPT and gotten back something flat, generic, and unusable, the problem isn’t the AI. It’s the prompt.
Most marketers treat ChatGPT like a search bar. They type a vague request and hope for a great result. But large language models don’t “guess” what you mean; they predict the most statistically likely response to the words you give them. Vague input equals generic output, every time.
In this guide, and in the video below, I break down a complete system for writing prompts that work: a 5-part prompt formula, the RTF and CARE frameworks, and two advanced techniques called role-based prompting and chain-based prompting. Every section links to the exact timestamp in the video so you can watch the live demo alongside the write-up.
Video timeline:
- 00:00 Introduction: Why your ChatGPT results are average
- 01:10 What is a Prompt?
- 01:45 Bad Prompt Example (Live Demo)
- 03:00 Good Prompt Example (Live Demo) and the Golden Rule
- 05:30 What is Prompt Engineering?
- 06:00 The 5-Part Formula: Role, Task, Context, Format, Goal
- 08:00 Instagram Content Calendar Prompt (Live Demo)
- 09:30 Best Prompt Frameworks Explained
- 09:45 RTF Framework (Role, Task, Format)
- 10:15 CARE Framework (Context, Action, Result, Example)
- 11:00 RTF Live Demo: SEO Strategy Prompt
- 12:30 Role-Based Prompting Explained
- 13:30 Chain-Based Prompting Explained (5-Step Blog Process)
What Is a Prompt?
(Watch this section: 01:10)
A prompt is simply the instruction you give an AI tool like ChatGPT. But most people write prompts the way they’d type a search on Google, a few keywords and a hope for the best. ChatGPT isn’t a search engine. It’s a prediction engine that responds to the quality and clarity of what you feed it. A better prompt produces a better output, every single time.
Bad Prompt vs. Good Prompt (Live Demo)
(Watch this section: 01:45 and 03:00)
In the video, I demonstrate this live with a side-by-side comparison.
Bad prompt example: “Write a social media post about my product.”
This gives ChatGPT almost nothing to work with: no audience, no tone, no goal, no format. The output is generic and could apply to any product, from anyone, for anyone.
Good prompt example: A rewritten version of the same request that adds a role, an audience, a tone, a clear goal, and a format. The difference in output quality is immediate and obvious.
The Golden Rule of Prompting
The golden rule I share in the video is this: the more specific your input, the more useful your output. ChatGPT cannot read your mind or your brand guidelines. Every detail you leave out is a detail it has to guess, and guesses produce generic content.
What Is Prompt Engineering?
(Watch this section: 05:30)
Prompt engineering is the practice of structuring your instructions to an AI model so it produces the output you actually want, on the first try, instead of the fifth. For marketers, this isn’t a technical or coding skill. It’s closer to writing a strong creative brief for a freelancer or agency: the clearer the brief, the better the work that comes back.
The 5-Part Prompt Formula: Role, Task, Context, Format, Goal
(Watch this section: 06:00)
This is the core formula behind every strong prompt, and it expands on the golden rule above. Each part answers a question ChatGPT would otherwise have to guess.
| Component | Question It Answers | Example |
|---|---|---|
| Role | Who should the AI act as? | “Act as a senior social media strategist for D2C beauty brands.” |
| Task | What exactly do you want done? | “Create a 7-day Instagram content calendar.” |
| Context | What background does it need? | “The brand sells vegan skincare to women aged 22 to 35 in India.” |
| Format | How should the output be structured? | “Present as a table with Day, Post Type, Caption Idea, and Hashtags.” |
| Goal | What outcome are you optimizing for? | “The goal is to drive profile visits and saves, not just likes.” |
Put together as one prompt, these five elements remove almost all the ambiguity that leads to generic ChatGPT output.
Instagram Content Calendar Prompt (Live Demo)
(Watch this section: 08:00)
In the video, I use this exact 5-part formula to generate a full Instagram content calendar live, showing how each of the five elements shapes the final output. A template version:
Role: Act as a senior social media strategist for D2C beauty brands. Task: Create a 7-day Instagram content calendar for a new product launch. Context: The brand sells vegan skincare to women aged 22 to 35 in India. The launch product is a niacinamide serum. Format: Present as a table with columns for Day, Post Type, Caption Idea, and Hashtags. Goal: Optimize for profile visits and saves, not just likes.
Best Prompt Frameworks Explained
(Watch this section: 09:30)
Once you understand the 5-part formula, the RTF and CARE frameworks are simply structured, memorable shortcuts for applying it. RTF is the fast, everyday version. CARE is the deeper version for tasks that need more nuance.
RTF Framework: Role, Task, Format
(Watch this section: 09:45)
RTF strips the 5-part formula down to its three fastest-moving pieces: Role, Task, and Format. It is the quickest framework to learn and covers most day-to-day marketing prompts, such as captions, subject lines, and short-form copy.
| Component | What It Means |
|---|---|
| Role | Tell the AI who to “be” |
| Task | State exactly what you want done |
| Format | Specify how the output should look |
CARE Framework: Context, Action, Result, Example
(Watch this section: 10:15)
CARE is built for tasks where tone, background, and outcome matter more, like campaign plans, nurture sequences, or positioning documents.
| Component | What It Means |
|---|---|
| Context | The background ChatGPT needs: your company, audience, goals, or constraints |
| Action | The specific, practical instruction |
| Result | The outcome you want, so the model knows what success looks like |
| Example | A sample of the tone, format, or quality you expect |
RTF Live Demo: SEO Strategy Prompt
(Watch this section: 11:00)
In the video, I apply RTF live to build an SEO strategy prompt. A template version:
Role: Act as an experienced SEO strategist for local service businesses. Task: Build a 90-day SEO content plan to improve rankings for a plumbing business in Mumbai. Format: Present as a table with columns for Month, Focus Keyword, Content Type, and Target Search Intent.
RTF vs. CARE: Which One Should You Use?
| RTF | CARE | |
|---|---|---|
| Best for | Quick, single-output tasks | Multi-step or strategic tasks |
| Components | 3 (Role, Task, Format) | 4 (Context, Action, Result, Example) |
| Example use case | SEO strategy outline, captions, subject lines | Campaign plans, nurture sequences, positioning docs |
A simple rule of thumb: start with RTF. If the output feels close but not quite right, missing your brand voice, missing background, missing nuance, add Context and an Example and you’ve upgraded to CARE.
Role-Based Prompting Explained
(Watch this section: 12:30)
Role-based prompting is the practice of assigning ChatGPT a specific persona before giving it a task, such as “Act as a 15-year email marketing veteran” or “Act as a skeptical customer reading this ad for the first time.” Assigning a role narrows the model’s tone, vocabulary, and frame of reference, which is why it consistently produces sharper, more specific output than an unassigned prompt. This is the same “Role” component from the 5-part formula and RTF, used on its own as a standalone technique for quick tone or perspective shifts.
Chain-Based Prompting Explained (5-Step Blog Process)
(Watch this section: 13:30)
Chain-based prompting means breaking a large task into a sequence of smaller prompts, where the output of one step becomes the input for the next, instead of trying to get a perfect result from a single prompt. In the video, I demonstrate this with a 5-step process for writing a blog post from scratch: outlining, drafting section by section, tightening the language, adding examples, and finishing with a title and meta description pass. Chaining prevents the shallow, rushed output you tend to get from asking ChatGPT to “write a full blog post” in one shot.
Distinction between RTF and CARE extremely practical.
For example:
RTF = Role + Task + Format
Best for straightforward, structured marketing tasks.
CARE = Context + Action + Result + Example
Better when you need the AI to understand the situation, desired outcome and style/example.
Frequently Asked Questions
What is the 5-part prompt formula?
What is the RTF framework in prompt engineering?
What is the CARE framework for ChatGPT prompts?
What is role-based prompting?
What is chain-based prompting?
Key Takeaways
- The golden rule of prompting: the more specific your input, the more useful your output.
- The 5-part formula (Role, Task, Context, Format, Goal) is the foundation for every strong prompt.
- RTF is the fast, everyday framework for quick marketing tasks.
- CARE adds depth for strategic, multi-step marketing work.
- Role-based prompting sharpens tone and perspective by assigning ChatGPT a persona.
- Chain-based prompting breaks large tasks, like a full blog post, into a connected sequence of smaller prompts.
Previous Topic: How to Audit a Website for SEO in 2026
