Artificial intelligence adoption is at an all-time high. Generative AI, in particular, surged from a 33% adoption rate in 2023 to 65% in 2024 and continues to rise.
There are myriad ways to use Gen AI, from content creation to data processing. However, the most commonly reported functions of the technology by businesses that have weaved it into their workflows are content support and strategy (16%).
While using Gen AI means doing more with fewer resources, the quality you get out of it is directly correlated to how effective your prompts are. With so many people prompting Gen AI every day for all kinds of different things, there are bound to be a few feeble attempts. However, those shouldn’t be a reflection of Gen AI as a whole.
The other side of that proverbial coin is that the better you make your prompts, the better your results will be. Let’s learn how to create compelling copy to feed the machine and what makes a good prompt so you start seeing better, more relevant outcomes.
5 Universal Prompt Writing Tips and Best Practices
Before we dive into a selection of different types of prompts and when they’re the most useful, there are universal truths and best practices for writing an effective prompt that apply to nearly every kind.
1. Be Clear and Specific
Your prompts should clearly define the task you wish to accomplish or the question you want answered. Use plain language and don’t be afraid to add supporting context if necessary. For example, instead of: “Tell me about AI.” Try, “Explain the basic concepts of AI and its applications in health care.” More specificity is always better unless you’re looking for general information about a broad topic.
Subscribe to
The Content Marketer
Get weekly insights, advice and opinions about all things digital marketing.
Thanks for subscribing! Keep an eye out for a Welcome email from us shortly. If you don’t see it come through, check your spam folder and mark the email as “not spam.”
2. Define the Role or Perspective
Giving the AI a persona or role can help make its responses more relevant. If you’re asking it about high school education, tell it to pretend it’s a teacher before asking any questions. Here’s an example: “You’re a high school teacher. What are some of the common challenges today’s educators face in the classroom?”
3. Set the Desired Output Format
If you don’t tell the AI to produce its answer in a specific format, it’ll do whatever it thinks is best. So, before you begin, specify the structure, such as bullet points, tables or a narrative format to get what you want right from the start.
4. Set Boundaries and Constraints
While it doesn’t always listen (at least not perfectly), setting guidelines can help keep the AI on track. If you want it to avoid a specific topic in its response, say so. If you have a word limit in mind, set that boundary. It may not always get it spot on, but it will try — and it will generally be closer to what you want if you didn’t provide that guidance at all. It’s less back and forth!
5. Iterate, Don’t Overload
Generative AI is pretty powerful, but it’s not immune to confusion. Overloading your AI writing tool of choice with too much information in one prompt will ultimately work against you. Instead, break longer or more complex tasks into smaller, manageable prompts and refine as you go. Tools like ChaptGPT remember what you’ve told it as long as you remain in the same chat window.
8 Types of Generative AI Prompts and When To Use Them
There are tons of different kinds of prompts — and I mean tons. You can ask generative AI tools like ChatGPT and Google Gemini for just about anything, so you have the freedom and flexibility inherent in language itself to craft your inputs.
With that in mind, generative AI is very much a ‘you get out what you put in’ type of deal. If you hit your enter key on a lackluster, incomplete or vague prompt, that’s likely what the machine will return to you in its response.
Enter prompt engineering, a term that describes the practice of creating, testing and refining prompts from generative AI to produce the best results possible. The cool part? Anyone can be a prompt engineer.
These common prompt types are great starting points on your way to becoming an effective prompt engineer who can bend Gen AI to your will.
1. Text Completion Prompts
A text completion prompt sees you using a partial or incomplete bit of text and asking your AI tool to complete it. This type of text prompt is great for brainstorming, storytelling or bringing yourself up to speed on a topic, like the benefits of exercise. Here are some examples:
- “The benefits of exercise include increased energy, improved mood, and…”
- “Once upon a time, in a distant kingdom…”
2. Instructional Prompts
If you want your AI generator to perform a specific task, like summarizing a paragraph or writing a specific piece of content, you’ll want to craft some instructional prompts. This type of prompt is excellent for personal education, content creation and task automation. Examples include:
- “Summarize the following article in one paragraph.”
- “Write a poem about a rainy day.”
3. Creative Writing Prompts
While I’m hesitant, nay, certain that generative AI cannot be truly creative, creative prompts can help you beat writer’s block by inspiring or unlocking ideas that you just can’t coax out. And since they’re meant to be creative, you can have a little fun and experiment however you like:
- “Write a sci-fi story about an uncontacted lifeform living in the depths of Earth’s oceans.”
- “Create a poem in the style of Shakespeare about love and loss.”
4. Informational Prompts
Informational prompts are designed to elicit detailed, factual or explanatory responses. Back when AI tools like ChatGPT didn’t connect to the internet, it was a bit more difficult to take what it said as fact. To be clear, you still shouldn’t take what any Gen AI tool says as fact even if they can search the web. Always cross-reference your results. That said, informational prompts are amazing research tools that can help expedite your traditional processes. Here are a couple of examples:
- “Describe the cultural significance of the Renaissance.”
- “What are the main causes of climate change?”
- “What are the differences between renewable and non-renewable energy sources?”
5. Image Generation Prompts
If you’re using generative AI to create images for or to inspire graphic design, for example, prompting looks a little different than it would if asking for text-based content. Describing visual elements, composition, colors, style and sometimes emotions isn’t the same as asking ChatGPT to finish a sentence; it’s much more involved.
Since visual interpretation is highly literal, image prompts require explicit detail to achieve precise results. You have to be more of everything — more descriptive, more detailed, more clear.
Instead of typing “Generate a serene mountain scene.” Image prompts should look more like the following example:
- “Generate an image of a snowy mountain peak at sunrise, with climbers reaching the summit, under a vivid orange sky.”
From there, you can iterate:
- “Make the sky more vibrant, add more details to the mountains and include a flowing river in the foreground.”
Generating images isn’t a replacement for human talent, but it can be great for unlocking ideas for graphic design projects and even aid storytelling by providing visual inspiration.
6. Analytic Prompts
We’ve used traditional AI for data-related tasks long before the present, but the proliferation of generative AI unlocks a new way to use the technology for analysis. If you need help with critical thinking, data analysis, logical reasoning or problem-solving, analytic prompts are there to help extract trends, forecast or perform cause-and-effect analysis:
- “Analyze the following sales data to identify trends: [data].”
- “What are the likely economic impacts of a prolonged supply chain disruption?”
- “How might implementing a hybrid work model impact employee productivity?”
7. Conversational Prompts
Using Gen AI bots as a conversation partner has lots of great applications, from interactive learning experiences to offering virtual customer assistance and coaching. Conversational prompts should simulate or guide natural back-and-forth dialogue and encourage the AI to respond in a conversational tone or assume a specific role within a discussion:
- “Take on the role of a personal trainer. Let’s have a conversation about creating a tailored workout routine for me.”
8. Programming Prompts
Yes, generative AI can help you write code. It can even explain how a particular piece of code works and provide debugging support.
- “Create a JavaScript snippet that validates email input using a regular expression.”
- “Explain how this Python generator function works.”
- “Here’s my Python code for a binary search function, but it’s not working correctly. Can you debug it?”
Go Out and Get Prompting
As I mentioned earlier, there are many, many different kinds of prompts. You have the freedom to try and tweak all kinds of inputs to see what works best for you. Even if you think generative AI couldn’t possibly help you with a particular task or problem, give it a try anyway; see what happens! You might be surprised.
