Using Prompt Reflection in ChatGPT: Booststraping Content


Reflection is a powerful technique in writing prompts. It allows you to create a step-by-step process to take an idea or analysis and expand on it. It creates a road map for ChatGPT to go from a poorly executed analysis to a highly efficient one. So what exactly is prompt reflection?

Prompt Reflection is a technique for writing prompts where a prompt directs ChatGPT to build content based on a step-by-step process. The initial content that ChatGPT creates is fed into a second step and that content is often sent to a third step or more.

So now you may be thinking that that sounds overly complicated. But it’s not. We often use this technique ourselves as human beings and in a number of areas.

For instance, if we’re building a customer Avatar we first have to decide who that person is going to be. Once we have an idea of who our ideal customer is in terms of their age and profession we can begin to understand them better. We can start understanding their needs and how we can serve them better.

(We’ll see how that works out below!)

So let’s dive in and see how to use prompt reflection with ChatGPT.

Table of Contents

Prompt Reflection Example #1: Writing an Article Title

Writing and optimized title that people want to click and only 60 characters is tough. The title has to have a little bit of a clickbait feel but it also has to be true.

Why 60 characters? Because that’s the amount of space that Google will give us when we show up in the SERPs for our title. Let’s take a look at an example.

Now you’ll remember that I said that the title has to be true. That means that it has to represent accurately the content of the article. If it doesn’t then you’ve been dishonest with your reader and they probably won’t come back.

One way to make sure that we are accurately creating a title is to use a persona that knows the material to create the title. What I mean by that is we’ll ask ChatGPT to be that expert persona and craft 10 titles for us.

Once we have those 10 titles will then ask a new persona that is an expert at optimizing titles for Google Search to refine the titles.

In terms of prompt reflection, we are having ChatGPT create some content and then modify it. This is the simplest type of prompt reflection.

(This prompt is written for ChatGPT 3.5 it does not work well on GPT4 without modification)

Acting as a ChatGPT and AI expert, write 10 sets of 60 character titles using a friendly and helpful tone, with the goal of creating titles that will be clicked on Google for the subject “How to Create a Neural Network”. After creating the titles, take on the persona of an expert in Google Organic Traffic Marketing and Title Optimization, and review the titles. Rewrite the titles to have a higher probability of being clicked if they appear for related search terms.

For the sake of space, I am only going to show 3 titles written by each persona

Sure! Here are 10 sets of 60 character titles for “How to Create a Neural Network”:

How to Create a Neural Network from Scratch
Building a Neural Network: Step-by-Step Guide
Simple Guide to Creating Your Own Neural Network
**********——— Removed For Space —————************

“Building a Neural Network: A Step-by-Step Guide for Beginners
“Create a Neural Network from Scratch: Easy Coding Tutorial”
“DIY Neural Network: How to Build Your Own AI Model”

***********——— Removed For Space —————************

The Many Personas of ChatGPT

So as you can see in this prompt, persona one is an expert on AI and created some titles and sent those titles to and expert persona on Google title optimization. The content of the titles was essentially the same. The difference of course is is that the second set of titles is much more likely to be clicked on.

It’s also important to note that the titles are true to the content that we said we had written.

Using multiple personas and having one Persona analyze the output of another can be used for many purposes. For instance, we may choose to have one persona write an article that’s an expert in a field and another persona that is an expert at communicating recycle it for use with young adults.

Another use of that same article might be to create a YouTube video. The article that we created initially would then be Rewritten by a persona that’s an expert on YouTube Scripts.

Yet another use of having different personas reflect on each other’s content is brainstorming. One Persona that might be an expert on ChatGPT might be paired with an persona expert on business within your prompts.

There are lots of possibilities for using different personas with prompt reflection I look forward to seeing some of yours.

Prompt Reflection Example #2: Creating A Customer / Reader Avatar

Sarah Our Avatar

For the next example of prompt reflection, we’re going to consider taking a small piece of data and expanding on it. In this case, we’re going to create a customer or reader Avatar. Let’s take a look at the prompt and the output.

Acting as an Expert Demographics Analyst, please take the [subject] I give you and use it to meet the goal.

Your goal is to create a comprehensive and highly detailed customer avatar, taking into account their demographics, pain points, personality characteristics, and preferences. Use this avatar to generate insights about their behaviors, needs, and interests related to the [subject]. Be sure to follow the instructions below step by step,

Identify the type of person most likely to be the [subject] and create a detailed Avatar, considering factors such as age, income, education, and geographical location.

Analyze your Avatar’s goals, values, challenges, and pain points in regard to the [subject].

Create a set of questions your Avatar would likely ask about the [subject], considering their challenges and pain points.

Identify your Avatar’s sources of information, including websites they would likely visit and influencers they might follow, to learn more about the [subject].

Suggest products or services that your Avatar would be likely to purchase to meet their wants and needs related to the [subject].

Recommend a specific product or service a blogger could offer to cater to your Avatar’s needs and preferences related to the [subject].

If the Avatar purchased our product what would the physical effect be if any to their body? Describe a before and after state.

If the Avatar purchased our product how would they feel? Describe a before and after state.

If the Avatar purchased our product how would it change their day? Describe a before and after state.

If the Avatar purchased our product how would it change their status in society? Describe a before and after state.

What are the most likely USPs for this Avatar for our product?

Are there any particularly good hooks that we could use to advertise?

Propose a lead magnet, such as a free resource or downloadable, that a blogger could provide to your Avatar in exchange for their email address and subscription to an email list.

Please create an image description of the Avatar in great detail for a person that is legally blind to understand them better.

[subject] Person Reading an Article on Prompt Engineering

Avatar:

Name: Sarah
Age: 32
Gender: Female
Occupation: Software Engineer
Income: $80,000 per year
Education: Bachelor’s degree in Computer Science
Geographical Location: San Francisco, California

Goals:

Stay up-to-date with the latest trends and best practices in software engineering
Improve her coding skills and productivity
Get promoted to a senior software engineer role
Find a better work-life balance
Values:

Knowledge and skill development
Work-life balance
Professional growth and career advancement


Challenges and Pain Points:

Feeling overwhelmed with work demands and a busy schedule
Struggling to keep up with new technologies and trends
Difficulty in balancing work and personal life


Questions:

What are the latest best practices in software engineering?
How can I improve my coding skills and productivity?
What tools or resources can I use to stay up-to-date with the latest trends in software engineering?
How can I balance my work and personal life more effectively?


Sources of Information:

Medium, Hacker Noon, and other tech-related blogs and websites
LinkedIn and GitHub profiles of industry experts and influencers
Online forums and communities for software engineers

Products and Services:

Online coding courses and tutorials
Coding bootcamps
Productivity and time management tools
Personal development books and resources


Recommended Product or Service:


An online coding bootcamp that offers personalized coaching and support to help Sarah improve her coding skills and achieve her career goals.

Physical Effects:

There would be no significant physical effect on Sarah’s body.

Emotional Effects

Sarah would feel more confident and motivated in her work, as she would have gained new skills and knowledge that would make her a better software engineer.

Daily Effects:

Sarah’s work would become more efficient and effective, allowing her to better balance her work and personal life.

Status Effects:

Sarah’s status as a software engineer would improve, making her more attractive to potential employers and clients.

USPs:

Personalized coaching and support
Access to industry experts and influencers
Convenient online learning platform
Hooks for Advertisement:

“Get Ahead in Software Engineering with Personalized Coaching and Support”
“Master the Latest Technologies and Best Practices in Software Engineering”
“Achieve Your Career Goals with an Online Coding Bootcamp”

Lead Magnet:
A free ebook on “10 Strategies for Balancing Work and Personal Life as a Software Engineer.”

Image Description:

Sarah is a 32-year-old woman with shoulder-length brown hair and green eyes. She wears glasses and a black turtleneck sweater. She sits at a desk with a computer and a notebook, surrounded by books on software engineering and productivity. She looks focused and determined, with a hint of stress on her face.

The description sounds pretty accurate. The only thing that’s a little out is the fact that prompt engineering was not a well-known term in 2021 when gpt’s data cut off occurred. I would expect for a better known subject that slightly better suggestions would be made.

What’s important about the example is this, we have a single persona a demographics expert analyzing the most probable person to be reading a article. That person has certain proclivities and pain points that are important for our purposes.

Based only on the information of what our Avatar was reading ChatGPT was able to create a plan on how to reach out to that person.

That’s incredible! One small bit of data was used to create more data which was used to create even more data until we get a full picture of our ideal reader.

Prompt Reflection Example #3: Creating a Summary

Now it’s possible to give some content to chat GPT and simply ask it to create a summary. The problem is the summary that it creates is poor. To make the summary better we’re going to use prompt reflection.

Take a minute to think for yourself how you would use prompt reflection to improve a summary.

Okay here we go then. We’re going to first have chat GPT figure out the main topics covered in the content. Then using those topics we’re going to have ChatGPT summarize the information about them.

So what I’m going to do is give you an example of a reflective summary prompt output versus and non-reflective prompt output.

Let’s take a look first at a non-reflective summary of this youtube video.

Get SEO Backlinks for FREE in 2024 NEW!
3 AUTHENTIC methods: 100+ DoFollow Backlinks for SEO [Niche Relevant Free Backlink Building]

ChatGPT please give me a summary of the content below.

“Transcript of the Youtube Video” (you can get a transcript at at https://youtubetranscript.com/)

The content is about methods to get more organic traffic to a website or blog by obtaining high-quality, niche-relevant backlinks. The author shares three methods to achieve this goal, including finding broken pages on competitor websites, stealing competitors’ backlinks, and reaching out to website owners with resources related to a broad keyword search. The author recommends trying all three methods and then finding the best topics to write about and writing them at scale. Two videos on the author’s channel are suggested as additional resources for finding quick-rank keywords and writing articles with AI optimization.

Ok…. As a brief overview that’s not bad. But it’s not really what you want as a summary. Let’s try a reflective prompt.

Acting as an expert at creating concise but detailed summaries please use the content below and the content I asked you to you store in memory.

I would like you to follow these steps.

1) Create a list of the most important topics found in the content. Send this to me as a response.

2) Using your generated topic list, focus on each topic and summarize all the data in the content about each of those topics separately. State each topic and create a bulleted list. Send me this list as a response.

Topic List:

Importance of high-quality backlinks for website traffic and revenue
Three methods to obtain free, high-quality backlinks at scale
Method 1: Finding broken pages on competitors’ websites and offering a replacement link
Method 2: Stealing competitors’ backlinks by analyzing their backlink profiles
Method 3: Finding resource pages related to your niche and requesting to be added as a link
Importance of writing SEO-optimized articles at scale
Using AI to write SEO-optimized articles
Summary of Methods for Obtaining Backlinks:

Method 1: Finding Broken Pages on Competitors’ Websites

Find competitors’ broken pages using SEMRush’s backlink analytics tool
Identify websites that are linking to the broken page
Reach out to those websites and offer your website as a replacement link

Method 2: Stealing Competitors’ Backlinks

Identify competitors in your niche and analyze their backlink profiles using SEMRush
Reach out to websites that are linking to your competitors and offer your website as a replacement link

Method 3: Finding Resource Pages Related to Your Niche

Use Google to find resource pages related to your niche
Reach out to the owners of those pages and request to be added as a link

Other Tips:

Use SEO-optimized articles to attract more organic traffic to your website
Use AI to write articles at scale and optimize for SEO.

Part of the difference that you are seeing between the two summaries is formatting the prompt reflection summary actually has a lot more information in it. It actually suggests how what’s being spoken in the video can be done. To me that makes it much more useful. Overall it’s very clear that the second summary is better than the first.

So that’s the difference. Using prompt reflection allows you to have ChatGPT analyze data and a step-by-step way to create better content.

So How Can You Build Prompt Reflection Prompt?

Pulling Yourself Up by Your Bootstraps

So let’s take a look first of all how the process works.

  • We ask ChatGPT to analyze either some data that we give it or some data that it has internally.
  • We ask ChatGPT to take that data and use it to create a refined analysis.
  • We may then take that content from the refined analysis to go to another step or we may do another analysis using the content we’ve created.

The process of taking a small amount of data and then using it to grow and create more and more data is called bootstrapping. That’s the term that came into wide use in the beginning of the Computer Revolution a long time ago.

Of course from the image it looks a bit ridiculous but computers as they boot up read in a small bit of code which then reads in more and more code until the computer is booted up. Similar processes occurring with prompt reflection.

To apply this to your own prompts what you need to do is figure out what the most important data is first.

In the case of our summary prompt we needed to know the topics covered first, in the case of our Avatar prompt we need to know who the person was, and in the case of the titles we needed to have titles that were about the correct information first.

  1. So the first step in using a prompt reflection is to determine what is the most important data to accomplish your task.
  2. Once you have that data you’re going to use it by asking chat GPT to analyze content. That content may be titles that you’re giving it or it might be the behaviors and thoughts of a person that you asked ChatGPT to analyze.

In the end it’s that simple and that complex. Knowing exactly when to use prompt reflection is more of an art at this point than a science, my basis for determining whether to use it or not it is the quality of the content if I don’t use prompt reflection.

So if you’re unhappy with the quality of the content that ChatGTP is producing you should consider using prompt reflection.

So What Are the Benefits of Using Prompt Reflection?

I think it’s pretty clear from what we’ve already said that there are numerous benefits to using prompt reflection. Here just a few.

  • Improved content creation by using multiple personas
  • The ability to take a small amount of data and expand on it using ChatGPT’s existing knowledge base
  • The ability to direct ChatGPT’s analysis to make it more detailed and more useful

For these reasons alone I think that prompt reflection should be in your toolbox.

Conclusions

So in this article we defined what prompt reflection is. How we could use it and practical prompts. And of course how you can build your own prompts using prompt reflection.

I hope that you found this article informative and that you’ll be able to use prompt reflection the next time you’re unhappy with the quality of the content that ChatGTP produces. Good luck and I look forward to seeing some of the prompts that you create.

Chris

Chris Chenault trained as a physicist at NMSU and did his doctoral work in biophysics at Emory. After studying medicine but deciding not to pursue an MD at Emory medical school Chris started a successful online business. In the past 10 years Chris's interests and studies have been focused on AI as applied to search engines, and LLM models. He has spent more than a thousand hours studying ChatGPT, GPT 3.5, and GPT4. He is currently working on a research paper on AI hallucinations and reducing their effects in large language models.

Recent Posts