The importance of a good AI prompt and how to write it 

In 2025, Artificial intelligence is becoming a default tool in every marketer’s toolkit. Tools like ChatGPT, Gemini, and Midjourney are now part of our everyday work life. However, it is still fairly easy to tell who is using AI correctly and who is not. The quality of AI-generated content depends entirely on the instructions it is given. A vague AI prompt will yield generic results, while a well-structured, detailed prompt can produce content that is precise, engaging, and useful. The process of refining these instructions is known as prompt engineering. 

By understanding the different prompting techniques and strategies, marketers, writers, and content creators can get the most out of their AI toolkit. In this article, we’re exploring best practices for prompt engineering, from basic principles to advanced techniques that improve both efficiency and output quality. Plus, we’re testing your AI-spotting eye.

How AI understands prompts 

AI does not possess human intuition. It follows patterns based on probability. When given a prompt, it predicts what text is most likely to follow based on the data it has been trained on. If a prompt lacks specificity, the AI will generate a response that is broadly relevant but may not align with the user’s expectations. However, when a prompt is structured clearly and provides sufficient context, the AI is more likely to produce a coherent and high-quality response. 

For example, a request like “write a product description” will result in a basic, generic response. However, adding specificity, “write a compelling product description for a high-end smartwatch designed for fitness enthusiasts, focusing on advanced health tracking, sleek design, and long battery life”, will guide the AI towards a more refined and useful output. 

Different prompting techniques 

Multiple approaches to writing prompts exist, each influencing the AI’s response in different ways. By understanding and experimenting with these techniques, users can refine their prompts to achieve better results. We’ve tested our own examples in two different AI text generators: ChatGPT (more creative) and Perplexity (more profound). 

Zero-shot prompting 

Zero-shot prompting involves providing AI with a request without any prior examples or context. This approach relies on the AI’s existing training to generate an answer. 

While this technique is useful for simple queries, it can result in generic answers that lack nuance or specificity.  

Few-shot prompting 

Few-shot prompting improves the AI’s performance by providing it with a few examples before making a request. This allows the AI to identify patterns and generate a response that is more aligned with the expected style or format. 

This approach is particularly useful when consistency in tone and structure is required, such as in branded content or social media posts. 

Chain-of-thought prompting 

For complex tasks that require reasoning or logical sequencing, chain-of-thought prompting encourages AI to break down the response step by step. Instead of providing a direct answer, the AI is guided through a structured thought process. 

This technique is effective for analytical writing, educational content, and problem-solving tasks. 

Meta prompting 

Meta prompting involves instructing the AI on how to think about the request before generating a response. By framing the task in a specific way, AI can be guided towards a more appropriate output. 

Role-based prompting 

By assigning AI a specific role, responses can be made more contextually aware and relevant to the task at hand. 

This approach is useful for ensuring that AI-generated content adheres to industry-specific standards and communication styles. 

Self-consistency prompting 

AI does not always produce the same answer for identical prompts. To ensure consistency, one technique involves prompting the AI multiple times with the same request and selecting the most accurate or useful response. This technique is particularly valuable for ensuring accuracy and reliability in AI-generated research or analytical writing. Prompt the AI tool with the same request and pick what suits your needs best.

Test yourself: Can you spot the AI? 

To demonstrate the impact of detailed prompting, we’ve put Google Imagen 3 to the test. Based on previous tests with different image generation tools, we’ve found Google Imagen 3 best suited for this job. So much so that we also struggled to find the AI, even though we prompted Imagen 3 ourselves. 

How good are you at spotting AI-generated images?

Test your AI knowledge. We’re showing you one real photo and one generated by AI based on an AI prompt from ChatGPT. Ready?

Test your skills

How it works

We're presenting five questions to test how sharp your AI-spotting eye is. Pick which picture you think is generated by AI

Question 1

Show prompt

"Man working on laptop outdoors with ocean view, modern patio."

Which picture is AI-generated?

Picture A

Picture B

Question 2

Show prompt

"Smiling child and mother in a vintage car, retro 70s style, warm sunlight, natural afros, joyful expressions, nostalgic road trip vibe."

Which one is AI-generated?

Picture A

Picture B

Question 3

Show prompt

"A bearded man with light skin, wearing a white hard hat and a navy blue padded vest over a long-sleeve dark blue sweater, is inspecting solar panels at a renewable energy farm. He holds a sleek, modern tablet in his left hand while his right hand gently touches the surface of a reflective solar panel. The sunlight casts natural highlights on his face and the metallic panels. The setting is an expansive solar farm with rows of photovoltaic panels stretching into the distance, surrounded by rolling hills under a clear blue sky. The composition follows the rule of thirds, with the man slightly off-centre. The camera angle is a medium close-up shot, slightly low, emphasising perspective and depth. The background is softly blurred, drawing attention to the worker and his task, with diagonal lines of the solar panels creating a strong leading line. The lighting is warm and natural, enhancing realism and texture in the scene."

Which picture is AI-generated?

Picture A

Picture B

Question 4

Show prompt

"A romantic outdoor wedding scene featuring two women dressed in elegant white dresses, exchanging vows in a natural setting. The woman in the foreground has voluminous curly hair adorned with white and purple flowers and is slightly out of focus, turned towards her partner. The main subject, a woman with long dark hair styled with a floral hairpiece, smiles radiantly while holding a piece of paper, possibly reciting vows. The setting includes a rustic floral wedding arch draped in white and purple fabric, surrounded by trees and earthy tones. The sunlight filters through the trees, casting a warm and soft natural glow on the couple. The composition follows the rule of thirds, with a shallow depth of field blurring the background and foreground to emphasise the emotion and intimacy of the moment. The medium close-up shot captures the joy and love between the couple, framed by a dreamy, nature-inspired backdrop."

Which one is AI-generated?

Picture A

Picture B

Question 5

Show prompt

"A diverse group of five young adults, mid-jump, celebrating joyfully in an urban university campus setting. Each person is holding a coloured notebook or folder (red, blue, yellow, purple), with expressive faces radiating excitement. They wear trendy casual outfits, including striped tops, denim jeans, plaid jackets, oversized sweaters, and headphones. The scene takes place in a narrow modern alleyway, framed by sleek glass buildings and classic brick structures. The shot is taken from a low-angle perspective, enhancing the sense of movement and energy. The natural daylight casts soft, even lighting with well-balanced shadows. The background is slightly blurred, keeping the focus on the subjects while maintaining architectural depth. The image is vibrant, dynamic, and full of youthful enthusiasm."

Which one is AI-generated?

Picture A

Picture B

How many did you get right?

These images are AI-generated:

Comments

This field is for validation purposes and should be left unchanged.

Versturen

How we created this test 

To be fully transparent on how we created this test on AI, we’ve decided to add our step-by-step action plan to this article. Use it for context, for inspiration, or just as an FYI.  

• Step 1: We crawled Freepik for images that varied in complexity (composition, lighting, depth, subjects, etc.)  
• Step 2: We fed the images to ChatGPT one by one and asked it to extrapolate a prompt from each image. 
• Step 3: In Google Imagen 3, we used this prompt to generate images. 
• Step 4: Repeat, but request increased complexity of the AI prompt by ChatGPT.

The principle of prompt engineering applies to text generation as well. The more precise the prompt, the more accurate and useful the AI’s response. 

Best practices for writing effective prompts 

Based on (our) experimentation, a structured approach to crafting an AI prompt has proven most effective. The following framework provides clarity and relevance in AI-generated content:

• Define the content type: Specify whether it is a blog, ad, image, email, or another format. 
• Provide context: Explain the purpose and audience for the content. 
• Set the tone and style: Indicate whether it should be professional, casual, technical, or persuasive. 
• Include key details: Highlight important features, benefits, or themes. 
• Establish constraints: Specify word limits, exclusions, or formatting requirements. 

Conclusion  

Writing an effective AI prompt is both a science and an art. A well-structured prompt ensures that AI-generated content is accurate, engaging, and fit for purpose. By experimenting with different prompting techniques, such as few-shot prompting, chain-of-thought prompting, and role-based prompting, users can achieve more precise and meaningful outputs. 

Artificial Intelligence is a helpful marketing tool, but it requires clear instructions to function effectively. By refining & reiterating your prompts, you can use AI as a valuable asset in your content generation process.