Prompt writing: how to write better AI prompts (with examples)

The difference between “AI does not work for us” and “AI saves us hours every week” rarely sits in the tool and almost always in the prompt. Ask a vague question and you get the generic average of the internet back; write a good prompt and you get usable work in your own voice. The good news: prompt writing is not a technical craft but a habit you can learn, built on one simple formula. In this guide you learn that formula in four steps, get copy-paste example prompts for the most common marketing tasks of an SME, and read when your own language or English pays off. At ClickForest we use and teach this approach daily; everything below comes from practice, tested on real tasks of SMEs in Flanders.

What is a prompt?

A prompt is the instruction you give to an AI model: the question, the instructions and all the context you send along. ChatGPT, Claude and Gemini generate their answer entirely from what that prompt contains (plus any files or settings), so the prompt is literally your steering wheel. Vague input produces vague output; sharp input produces sharp work.

“The hottest new programming language is English”

— Andrej Karpathy, AI researcher (former Tesla AI director, later OpenAI), on X in January 2023

It helps to see an AI assistant as a lightning-fast, broadly trained freelancer who does not know your business and cannot read minds. You would never hand such a freelancer a five-word briefing. Everything you would normally put in a good briefing (goal, audience, style, examples, format) therefore belongs in your prompt. Which assistant you choose matters little for this skill; we discuss the differences between the big models in ChatGPT vs Claude vs Perplexity, but a good prompt works everywhere. If you are unsure which assistant fits which task, our practical guide to choosing one helps.

How do you write a good prompt? The formula in 4 steps

Use the formula Role + Goal + Context + Output. Tell the AI who it should be (role), what exactly you want to achieve (goal), what it needs to know about your situation (context) and in which form the answer should come (output). Together those four elements catch nine out of ten prompt problems.

This is how each step works:

1. Role: “You are an experienced copywriter for Flemish webshops.” A role steers the tone, level and perspective of the answer.
2. Goal: “Write a product description that convinces doubters.” One concrete goal per prompt; ask for three things at once and you get three half answers.
3. Context: “The product is X, the audience is Y, our tone of voice is Z, this is an example we like.” This is the step almost everyone skips and the one that makes the biggest difference.
4. Output: “Give me 150 words, in Dutch, with a heading and three bullets.” Making format, length, language and structure explicit saves endless rework.

That context step in particular deserves repeating, every single time. As Wharton professor Ethan Mollick puts it:

“treat AI just like an infinitely patient new coworker who forgets everything you tell them each new conversation.”

— Ethan Mollick, professor at the Wharton School, in “Getting started with AI: Good enough prompting”

The formula is deliberately simple; more elaborate frameworks exist, and if you like working with a fixed template, our article on the RACE model for ChatGPT prompts offers a variant in the same spirit. Pick one and use it consistently: consistency beats perfection.

Which example prompts can you use right away?

Below are five copy-paste example prompts for the most common marketing tasks of an SME, each built along the formula Role + Goal + Context + Output. Replace the brackets with your own details and add an example of your own wherever possible; in our experience that last step doubles the usability of the output.

Task	Example prompt (shortening is fine, cutting context is not)
Product copy	”You are a copywriter for our webshop in [sector]. Write a product description of 120 words for [product], aimed at [audience]. Tone: [description or example]. Structure: catchy opening line, 3 concrete benefits, closing line with a call to action.”
Social post	”You are the social media manager of [business]. Write 3 variants of a LinkedIn post about [topic], each max 100 words, aimed at [audience]. Start each variant from a different angle: question, number, mini story.”
Email	”You are [role] at [business]. Write a follow-up email to a customer who [situation]. Goal: [desired action]. Tone: warm and professional, no sales talk. Max 120 words, in the first person.”
Translation with tone	”Translate the text below into [language] for [audience]. Keep the meaning but rewrite idioms naturally; no literal translation. Our tone of voice: [description]. Text: [text]“
Analysis	”You are a marketing analyst. Below are [data/reviews/questions]. Summarise the 5 most important patterns, each with a concrete example from the data and one action we as [type of business] can take.”

Save your best-performing prompts in a shared document with your team: that way you build a prompt library and nobody has to start from scratch every time. It is also the bridge to structural AI use, covered in our AI marketing guide. Which tools are strongest per marketing task is compared in the best AI tools for marketers; and if you would rather outsource content production structurally, that is possible through AI content production.

Should you prompt in your own language or in English?

For Dutch or French output: simply prompt in that language. The current generation of models understands both excellently, and prompting in English for a text in another language builds in an unnecessary translation step in which nuance and tone of voice get lost. English does pay off for technical tasks, code and content that has to be in English anyway.

Two practical refinements. For Dutch output, also state the region explicitly (“Belgian Dutch, no Netherlands-Dutch terms”) if that matters for your brand; models often write northern Dutch by default. And when in doubt: ask the model itself for two versions, one based on an instruction in your own language and one based on an English instruction, and compare. Testing takes thirty seconds and settles the debate for your use case better than any rule of thumb.

Want your whole team to learn to prompt better? ClickForest gives hands-on AI training to SMEs in Flanders, from prompt basics to working routines per department. Take a look at the AI training or book a no-obligation video call.

What are the most common prompt mistakes?

Five classics: too little context (the number one), several tasks in one prompt, no format specified, treating the first answer as the final product, and sharing sensitive data without thinking. All of them avoidable with the formula plus two habits: iterating and checking.

“Working with AI is a dialogue, not an order.”

— Ethan Mollick, professor at the Wharton School

Iterating means: treat the first answer as a rough draft and steer (“make it shorter”, “less formal”, “give three alternatives for the opening line”). A three-turn conversation almost always delivers better work than one perfect mega-prompt. Checking means: verify facts, figures and claims in AI output before publishing, because a fluently worded answer can still be wrong. And about that sensitive data: do not share customer data or confidential figures in tools whose settings you do not know; make team agreements about this before someone does it by accident.

How do you use prompts for recurring tasks?

Turn your best prompts into templates: a fixed prompt structure per recurring task, with fill-in fields for whatever changes each time. That way your whole team gets the same quality, and per task you build on what has proven to work instead of improvising every time. OpenAI CEO Sam Altman named the value of that investment early on:

“writing a really great prompt for a chatbot persona is an amazingly high-leverage skill and an early example of programming in a little bit of natural language”

— Sam Altman, CEO of OpenAI, on X in February 2023

In practice it works like this: pick your three most recurring writing tasks, work out one template for each along the formula, test it for a week, and sharpen it based on what the output misses. Store the templates centrally (a shared document is enough) and agree that improvements are kept there. From a handful of templates onwards, this becomes the foundation of your AI way of working; how you scale that way of working towards automation and agents is covered in our guide on AI agents in marketing.

When are you ready for prompt engineering?

As soon as you notice the basic formula falling short: with complex reasoning tasks, output that is never quite good enough, or workflows you want to make so reliable that you can automate them. Then it is time for techniques such as giving examples (few-shot), step-by-step reasoning and chaining prompts together.

The distinction is simple: prompt writing is a habit for everyone who works with AI; prompt engineering is the deep dive for those who build systems with it. The full explanation of those techniques, including when to use which one, is in our follow-up article what is prompt engineering?; a short definition is also in our glossary. Only start there once the basics are in place; most teams get the bulk of the gains from the formula and the templates above.

What should you remember about writing prompts?

Writing better prompts is not a talent but a habit: Role + Goal + Context + Output, with context as the step that makes the biggest difference. Prompt in your own language, save what works in templates, iterate instead of starting over, and check facts before publishing. Whoever gets that rhythm going gets multiples out of the same AI tools compared to a colleague who “just asks something”.

ClickForest helps teams of SMEs in Flanders prompt better and use AI structurally, through hands-on training and guidance on the broader AI way of working. Curious what is in it for your team? Book a no-obligation video call and we will look at your most time-consuming tasks together.