AI Prompting for Writing: How to Get High-Quality Output

Want to improve the quality and consistency of AI’s written output? Wonder how prompts, context, and templates can help?  

In this article, we explore AI prompting techniques for getting quality written results.

Why It’s Important to Balance AI Assistance With Human Creativity and Expertise

There are currently two extreme views on artificial intelligence (AI) in content creation: those who argue against using AI at all–claiming it produces low-quality content and lacks the human touch and those who are using tools that automate the entire content creation process, which could potentially flood the internet with AI-generated content.

Looking ahead, humans will remain crucial to the content creation process, at least in the near future. The approach is to use AI as an editor or assistant rather than relying on it to create content from scratch.

This balanced approach allows content creators to leverage AI's strengths (such as rapid information processing and language generation) while maintaining the unique insights, creativity, and emotional intelligence humans bring to the process.

As AI technology improves, people may care less about whether content was created by AI or humans as long as it provides value. 

The following tips will help you deliver truly valuable content that stands out because of its strong human influence.

#1: Start With Clear, Specific AI Prompts That Include Detailed Instructions

When writing effective AI prompts, it's crucial to be very clear and specific with your instructions. If you think of AI as a super knowledgeable, almost genius intern, but one that lacks judgment and tends to be “lazy,” doing exactly what you say without questioning it, you'll be on the right track for communicating with it. You need to spell out exactly what you want.

For example, instead of just saying, “Write a blog post about my morning routine,” provide your AI writing assistant with these instructional guides:

• The goal of the content (e.g., to inform, persuade, or entertain)
• The target audience (e.g., busy entrepreneurs, college students, new parents)
• Writing guidelines (including tone, style, and any specific requirements)
• A detailed template or structure for the content
• Examples of what to avoid (e.g., clichés, overly promotional language)
• References to writing styles or publications that exemplify the desired output

For example:

Write a 1,000-word blog post about a productive morning routine for busy entrepreneurs. The tone should be motivational yet practical. Include 5 key steps, each with a brief explanation and a real-life example. Conclude with a call to action, encouraging readers to try this routine for a week. Avoid using cliché phrases or overly promotional language. The writing style should be conversational but professional, similar to articles found in Harvard Business Review.

#2: Give Your AI Writing Assistant Context

Another key tip is to provide relevant background information to the AI. This might include details about your business, target audience, or personal preferences. 

Keep these context cues relatively short—just a few lines or a paragraph—to avoid overwhelming the AI tool with too much information. 

In ChatGPT, you can save this information in the “Custom Instructions” feature, which automatically applies to every new chat. Alternatively, you can save these contextual cues in a notepad to easily add them to your AI writing prompts when needed. However, don't rely too heavily on this feature, as it can sometimes lead to repetitive or overly constrained outputs.

For example:

I'm a solopreneur and content creator on Twitter. My content focuses on productivity hacks and time management techniques for freelancers and small business owners. My audience consists primarily of millennials and Gen Z professionals who are looking to optimize their work processes and achieve better work-life balance. My writing style is typically informal and engaging, with a focus on actionable advice and real-world examples. I often use humor and pop culture references to make complex topics more accessible.

#3: Use Templates for Consistent Structure in Your Content

A template gives your AI assistant the structure you want it to follow, almost line by line. This helps ensure you get consistent, high-quality results each time.

To create a template, you can ask the AI to analyze a piece of content you like and turn it into a general format. Here's an example:

Analyze the structure of this blog post and create a general template that can be used for similar posts on different topics. Identify key elements like the opening hook, main points, transitions, and conclusion. For each element, provide a brief description of its purpose and how it contributes to the overall effectiveness of the post.

You can then use this template repeatedly, plugging in different topics as needed. A detailed template might look like this:

1. Opening Hook: Start with a surprising statistic, provocative question, or brief anecdote related to the topic (2–3 sentences)

2. Introduction: Briefly explain the importance of the topic and what readers will gain from the post (1 paragraph)

3. Main Point 1: Present the first key idea, including:

• Clear explanation of the concept (2–3 sentences)
• Real-world example or case study (1–2 sentences)
• Actionable tip for implementing this idea (1–2 sentences)

4. Main Point 2: Follow the same structure as Main Point 1

5. Main Point 3: Follow the same structure as Main Point 1

6. Potential Challenges: Address 1–2 common obstacles readers might face when implementing these ideas (1 paragraph)

7. Solutions: Provide practical solutions or workarounds for the challenges mentioned (1 paragraph)

8. Conclusion: Summarize the main points and reinforce the benefits of implementing these ideas (1 paragraph)

9. Call to Action: Encourage readers to take a specific next step, such as trying one of the techniques or sharing their own experiences (1–2 sentences)

#4: Provide Multiple Examples of Good and Bad Content to Guide your AI Assistant

Show the AI good and bad examples of what you're looking for. This helps the AI understand your expectations more clearly and learn what you're looking for and what to avoid.

To enter examples, you can copy and paste or upload PDFs and other documents to your AI assistant.

The more relevant examples you provide, the better the AI can tailor its output to your needs. For instance, you might provide five examples of good transformations (where content has been successfully turned into a template) and one example of a bad transformation.

Here's an example of how you might include examples in an AI prompt:

Here are two examples of effective email newsletters:

Example 1: [Insert or upload full text of a well-crafted newsletter]

Example 2: [Insert or upload full text of another well-crafted newsletter]

Notice the following elements in these examples:

• Conversational, friendly tone
• Short paragraphs (2–3 sentences each)
• Use of subheadings to break up content
• Inclusion of personal anecdotes
• Clear call to action at the end

Now, here's an example of a poor newsletter:

[Insert or upload full text of a poorly crafted newsletter]

Notice these issues with the poor example:

• Overly formal tone
• Long, dense paragraphs
• Lack of personal connection
• No clear structure or subheadings
• Vague or missing call to action

When you write my newsletter, please emulate the style and structure of the good examples and avoid the mistakes in the poor example. Additionally, incorporate my brand voice as described in the context I provided earlier.

#5: Iterate on and Test Your Prompt Writing and Refine It Based on the Results

Don't expect perfect results on your first try. If you're unhappy with what you get, analyze what's wrong and adjust your instructions accordingly. Initially, expect to test and adjust a prompt at least 10 times before finalizing it. 

Follow this sample framework to get started:

• Start with a basic version of the prompt
• Analyze the output for strengths and weaknesses
• Refine the prompt based on this analysis
• Test the refined prompt multiple times
• Continue this cycle until you consistently get satisfactory results

Word Choice Matters

You might need to experiment with different words or phrases in your detailed prompt. For instance, replacing “creative writer” with “professional writer” could significantly change the content style you receive. 

Similarly, specifying “write in the style of Malcolm Gladwell” or “use language appropriate for a technical white paper” can dramatically alter the output.

Use the Lever Technique to Fine-Tune Specific Aspects of AI-Generated Content

For more precise control over the output, apply the Lever Technique:

• Get an initial output from the AI

• Ask the AI to rate a specific aspect of the output on a scale (e.g., formality from 1–10)
• Request the AI to adjust that aspect to a different point on the scale

For example, if you get an email that's too formal, you might say:

On a scale of 1 to 10, with 1 being very casual and 10 being very formal, how would you rate the tone of this email?

If the AI responds with “7,” you might then say:

Please rewrite the email, adjusting the tone to a 3 on the same scale. Maintain the same information and overall structure, but make the language more conversational and friendly.

Other Notes From This Episode

• Connect with Michael Stelzner @Stelzner on Facebook and @Mike_Stelzner on X.
• Watch this interview and other exclusive content from Social Media Examiner on YouTube.

Where to subscribe: Apple Podcasts | Spotify | YouTube Music | YouTube | Amazon Music | RSS

✋🏽 If you enjoyed this episode of the AI Explored podcast, please head over to Apple Podcasts, leave a rating, write a review, and subscribe.