- What Is Prompt Engineering?
- Why Is Prompt Engineering Important?
- The 5 Parts of a Good Prompt
- Prompt Engineering Techniques
- Examples of Prompt Engineering
- Prompt Engineering Best Practices
- What Does A Prompt Engineer Do?
- What Skills Are Needed For Prompt Engineering?
- How To Become a Prompt Engineer
- What Is The Average Salary For Prompt Engineers?
- Is Prompt Engineering Dead?
- Learn More About Prompt Engineering
What Is Prompt Engineering?
Prompt engineering is the practice of writing and refining the instructions you give a generative AI model, like ChatGPT, Claude, or Gemini, so that it produces accurate, relevant, and useful results. It involves choosing the right words, giving the right context, and structuring your request so the model understands exactly what you want.
A prompt is the natural-language text you give an AI model to tell it what to do. It can be a question, an instruction, or a longer brief with background information and examples. You can use prompts to generate text or images, analyze data, write code, summarize documents, and much more.
Here is an example of a prompt given to ChatGPT (for more ready-made prompts, visit our ChatGPT Prompt Library).
Why Is Prompt Engineering Important?
The quality of what you get from an AI model depends heavily on what you put in. A vague prompt gets a generic answer. A clear, well-structured prompt gets an answer you can actually use.
Good prompts give you:
More relevant results - When you explain the context and the goal, the model can tailor its answer to your situation instead of guessing.
Less back and forth - A complete prompt gets you close to the right answer on the first try, so you spend less time correcting and re-asking.
More consistent output - When you use the same structure every time (the same role, format, and rules), you get results in the same style and quality, which matters when you use AI for work.
Better use of each tool - Different AI tools have different strengths. Knowing how to prompt them helps you get the best out of each one.
The 5 Parts of a Good Prompt
Most effective prompts are built from the same five parts. You don’t need all five every time, but the more of them you include, the better the result usually is.
- Task: the action you want the AI to perform. Start with a clear verb, like write, summarize, compare, or explain. “Write a product description...”
- Context: the background the AI needs: who it is for, why you need it, and any facts it should use. “...for our new reusable water bottle, aimed at people who hike on weekends...”
- Persona: a role for the AI to take on, which sets the expertise and tone. “Act as an experienced ecommerce copywriter.”
- Format: the shape and limits of the output: length, structure, tone, and anything to avoid. “Use 3 short paragraphs, under 120 words, no jargon.”
- Examples: a sample of what a good answer looks like, so the AI can match it. “Here is a description we like: ...”
Put together, the prompt looks like this:
“Act as an experienced ecommerce copywriter. Write a product description for our new reusable water bottle, aimed at people who hike on weekends. It keeps drinks cold for 24 hours and weighs 300 grams. Use 3 short paragraphs, under 120 words, and no jargon. Match the tone of this description we like: ‘Light enough to forget, tough enough for any trail.’”
Prompt Engineering Techniques
Prompt engineering techniques are repeatable ways of structuring a prompt to get a better answer. These are the most common ones, with an example of each.
Zero-shot prompting - You ask the AI to do a task without giving it any examples. This works well for simple, common tasks.
Example: “Classify this customer review as positive, negative, or neutral: ‘Delivery was late, but the product is great.’”
Few-shot prompting - You give the AI a few examples of the input and the output you want, then ask it to do the same for a new input. This is the best way to get a specific style or format.
Example: “Here are two product titles in our style: ‘Trail Bottle 750ml, Keeps Cold 24h’ and ‘Camp Mug 350ml, Double-Walled Steel’. Write a title in the same style for our 1-liter insulated flask.”
Chain-of-thought prompting - You ask the AI to work through a problem step by step before giving its answer. This helps with math, logic, planning, and any task with several steps.
Example: “Our ad spend was $2,000 and it brought in 40 sales at $90 each. Work out the return on ad spend step by step, then tell me if the campaign was profitable.”
Role (persona) prompting - You tell the AI who to be. This changes the vocabulary, depth, and tone of the answer.
Example: “You are a financial advisor. Explain inflation to a high school student.”
Negative prompting - You tell the AI what to avoid. This is useful when the first answers keep including things you don’t want.
Example: “Write a LinkedIn post about our webinar. Don’t use hashtags, emojis, or the words ‘excited’ and ‘thrilled’.”
Prompt chaining - You break a big task into smaller prompts and use the output of one as the input for the next. For example, first ask for an outline, then ask for each section, then ask for an edit of the full draft.
Examples of Prompt Engineering
The easiest way to understand prompt engineering is to compare a basic prompt with an engineered one.
Prompts for AI text models like ChatGPT, Claude, and Gemini
When you give an AI text model this prompt: “Write a short paragraph on energy-saving tips.”, you get this result:
When you engineer the prompt by adding an audience, a length, and a format, like this: “Write a 200-word paragraph for beginners about the basics of energy-saving. Include 4 tips in a list format on reducing waste and conserving energy.”
You get a result that is more useful and closer to what you asked for:
We used our AI paragraph generator for these examples, but you can test the same prompts in ChatGPT, Claude, or Gemini and compare the answers.
Here is another pair you can try:
- Basic prompt: “Explain photosynthesis.”
- Engineered prompt: “Explain the process of photosynthesis to a 10-year-old, using simple language and examples, such as how a plant uses sunlight to make food.”
Prompts for image models
Prompts matter just as much for images. Image generators are built into ChatGPT and Gemini, and there are dedicated tools like Midjourney. We created the example below with OpenAI’s DALL-E.
A generic prompt like “Create a digital artwork” gives you a nice but generic image.
An engineered prompt that describes the subject, style, and setting gives you something much closer to what you had in mind: “Create a digital artwork inspired by the concept of ‘time travel,’ blending elements of futuristic technology with historical landmarks in a surreal landscape.”
Types of prompts, with examples
Most everyday prompts fall into one of these types:
| Type | What it does | Example |
|---|---|---|
| Instructional | Gives a direct command | “Summarize this article in three bullet points.” |
| Question | Asks for information or an explanation | “What is the difference between SEO and PPC?” |
| Role-based | Asks the AI to answer as a specific expert | “You are a career coach. Review my CV summary and suggest improvements.” |
| Contextual | Adds background that frames the answer | “This text is for a beginner cooking class. Rewrite it using simple steps.” |
| Few-shot | Shows examples for the AI to follow | “Here are two email subject lines we like. Write five more in the same style.” |
For ready-to-use marketing examples, see our ChatGPT prompt library, which includes marketing prompts.
Prompt Engineering Best Practices
Follow these tips when writing your AI prompts:
1. Understand What The Specific AI Model Can Do
Each AI tool has different strengths and limits. Knowing them helps you pick the right tool and write prompts that play to its strengths.
Here is a summary of the most popular AI tools and what they are good at:
- ChatGPT, Claude, Gemini, and Microsoft Copilot: general-purpose AI assistants. They can write and edit text, answer questions, search the web, analyze files and data, and write code.
- Image generators: ChatGPT and Gemini can create images from text prompts, and dedicated tools like Midjourney specialize in them.
- Coding assistants: tools like GitHub Copilot and Claude Code suggest, write, and fix code based on what you are working on.
Each provider also publishes its own prompting guide, which is worth reading for the tool you use most. If you work with Claude, our list of free Claude courses is a good place to start.
2. Provide All Required Details
Make your prompts specific and give the AI all the details it needs. A good prompt has:
A clear objective - State exactly what you want the AI to do. Ambiguity leads to broad or irrelevant answers.
Natural language - Write the way you would brief a colleague. Plain, direct sentences work best.
Specific details - Include the audience, the purpose, the length, and any facts the AI should use.
Clear limits - Say what to avoid and how long the answer should be.
Here is an example of a detailed prompt: “Create an email targeted to seniors, inviting them to join a webinar on the use of AI. Give 3 examples of how AI can be used to make their lives easier. Don’t use more than 200 words.”
3. Provide Feedback and Follow Up Instructions
Your first prompt rarely gets a perfect result, and that’s fine. Treat it as a conversation: tell the AI what it got right, what it got wrong, and what to change.
Be specific with your corrections. “Make it shorter, drop the second paragraph, and use a friendlier tone” works much better than “Try again.”
AI text generator tools sometimes use words a human wouldn’t use. The best way to make the writing sound more like you is to give feedback and examples of your own writing.
4. Experiment With Different Prompts
One of the best ways to improve your prompts is to experiment. Try different wording, add or remove context, change the format, and compare the results. Over time you’ll learn what works for your tasks and your tools, and you can save your best prompts to reuse.
5. Ask The AI For Help!
The AI can help you write prompts, too. Explain what you want to achieve and ask it to write the prompt for you, or ask it which information it needs from you to do the task well.
Use our free AI prompt generator to get examples and prompt ideas.
What Does A Prompt Engineer Do?
A prompt engineer designs, tests, and improves the prompts that AI systems use, so that they give reliable, high-quality results. In many companies, this means writing the instructions (often called system prompts) behind chatbots, AI features, and automated workflows.
Typical responsibilities include:
- Writing clear and effective prompts
- Testing and refining prompts against real examples
- Understanding what different AI models can and can’t do
- Evaluating AI responses for accuracy and quality
- Training colleagues to use AI tools well
- Working with developers who build AI products
- Making sure AI is used safely and ethically
Is using ChatGPT prompt engineering? ChatGPT is a tool, not prompt engineering itself. But every time you write a prompt for ChatGPT, Claude, or Gemini and improve it to get a better answer, you are doing prompt engineering.
What Skills Are Needed For Prompt Engineering?
Anyone can learn to write better prompts. For a job with “prompt engineer” in the title, employers usually look for:
- Excellent writing and communication skills
- A solid understanding of how large language models (LLMs) and generative AI work
- Hands-on experience with AI assistants and chatbots
- Analytical skills to test prompts and measure the quality of the results
- For technical roles, programming skills (usually Python) and experience with AI APIs
Some technical roles also ask for a degree in computer science or a related field. In many other jobs, especially in marketing and content, prompt engineering is expected as part of the role rather than a separate job title.
How To Become a Prompt Engineer
To get started, follow these steps:
1. Understand the Basics of AI and Machine Learning
Start with the fundamentals: what AI models are, how they are trained, and what they are good and bad at. Our guide on how to learn AI gives you a step-by-step roadmap.
The AI providers’ own guides are the best free resources on prompting. Read the prompt engineering guide from OpenAI and Anthropic’s prompt engineering overview for Claude.
2. Get Hands-On with AI Models
Use ChatGPT, Claude, and Gemini for real tasks. Give them the same prompt, compare the results, and notice how small changes in your prompt change the output.
3. Learn Programming and Scripting
For technical prompt engineering roles, learn a programming language such as Python, which is widely used in AI and data science. It lets you call AI models through their APIs, automate tasks, and test many prompts at once.
4. Practice, Practice, Practice
The key to becoming a good prompt engineer is practice. Use AI to write content, analyze data, automate tasks, or build a custom assistant, and keep a library of the prompts that work best.
5. Prompt Engineering Courses And Certifications
If you prefer a structured way to learn, a course gives you a clear path and a certificate for your CV. These are good options:
- Generative AI Fundamentals Specialization (IBM)
- Prompt Engineering Specialization (Vanderbilt University)
- AI for Everyone (DeepLearning.AI)
For more options, including free ones, see our list of free AI courses.
What Is The Average Salary For Prompt Engineers?
According to Indeed, the average base salary for a prompt engineer in the United States is $114,368 per year, with a typical range from about $74,000 to $177,000. The figure is based on salaries in job postings on Indeed over the past 36 months (updated September 2026).
Pay depends heavily on the role. Technical prompt engineers who build AI products earn more than people who use prompting as one skill in a marketing or content role.
Is Prompt Engineering Dead?
No, but it is changing. AI models have become much better at understanding plain requests, so you no longer need special tricks or exact phrases to get a good answer.
What matters more now is giving the model the right information. Anthropic describes context engineering as the natural progression of prompt engineering: instead of focusing only on the wording of a prompt, you decide which instructions, documents, examples, and tools the AI has access to while it works. This matters most for AI agents, which run through many steps on their own.
The basics in this guide still apply. A clear task, the right context, a defined format, and good examples are the foundation of both.
