Generate a landing page for a newsletter in 17 minutes using ChatGPT or GPT-3

Imagine we need to generate a landing page for a newsletter for software engineers with five years of experience who want to switch to data engineering. We’ll use AI - OpenAI’s ChatGPT and GPT-3 to generate the text.

The AIDA marketing model

The copywriting world is full of writing formulas, for example, AIDA. AIDA stands for Awareness, Interest, Desire, and Action. It is a hierarchical marketing model consisting of stages to convince a consumer to purchase. Of course, the action we want to “sell” doesn’t require selling anything. We can use the same model to convince people to subscribe to a free newsletter, donate to a charity, or watch a movie.

• First, we make the consumer aware of a problem they have.
• Second, we get them interested in solving the problem. We tell them the solution exists and how it can improve their situation.
• In the Desire stage, we describe the solution by promoting the benefits of solving the problem with the product we offer.
• In the end, we write a Call-To-Action. In the case of a newsletter, we want them to give us their email address. If we had an expansive product, we could offer a free webinar or a free call with a consultant to give them more information and deepen the desire to get the product.

Naturally, the model forms a funnel, and some consumers will drop out at every stage of the funnel. First, they may overlook the title or the first sentence. We fail at the awareness stage if the header doesn’t catch their attention. That’s why our GPT model must generate a catchy first paragraph or, even better, a catchy first line. Next, the consumers decide whether they are interested. Most people will drop out at this stage. It’s ok. You want to reach the right people, not all people. To create a Desire for your product, you must show your Unique Selling Preposition. Why would they buy from you instead of the hundred people who seemingly offer the same thing? In the end, you need a compelling Call-To-Action. The consumer may not be ready to buy after reading 1000 words long landing page, but the text may be convincing enough to make them subscribe to a newsletter and read more about your offer.

How does GPT work, and why does it matter for Prompt Engineering?

Before we start, we need to know how GPT works. Not because it’s important when we generate the text, but to understand what we can expect from the AI model and what expectations are unrealistic.

GPT is a text completion model. Given any input, GPT will produce the most likely next word in the sentence. Only the next word? Yes. Tools like ChatGPT send your input with the previously generated word to the model again to get entire paragraphs of text, but the model generates words one by one. If you use the Chain of Thought prompting technique, GPT can even do some logical reasoning, but the model won’t handle such a difficult task without your help.

We can lead GPT towards generating what we want by “role-playing.” In short, we provide the context describing a situation and ask GPT to fill out blanks in the script of the conversation. We will use the technique described by Swaroop Mishra from Arizona State University and Swaroop Mishra from Microsoft Research in their research paper “HELP ME THINK: A Simple Prompting Strategy for Non-experts to Create Customized Content with Models” arXiv:2208.08232.

That’s only one of many Prompt Engineering techniques. If you want to learn more about them, subscribe to my newsletter and get a notification when I publish the next article about Prompt Engineering and other Machine Learning-related topics.

To start, we describe the role played by AI in first-person singular and ask GPT to generate questions. Later, we answer those questions until we gather enough data to generate the output we want from our responses. In the text below, the regular font is the prompt written by me (context and answers). The bold text was generated by AI (questions):

I’m an expert copywriter. I will ask you several questions to gather information required to write a landing page for your newsletter for programmers with 5 years of experience who want to switch to data engineering:

Q: What is the goal of the landing page?
A: I want them to subscribe to my newsletter.
Q: What topics will the newsletter cover?
A: I will write about data engineering tools, software architecture relevant in data engineering, data lakes, data warehouses, batch data processing, and machine learning.
Q: What benefits will the subscribers get from the newsletter?
A: They will get all of the information they need in one place in a form of step-by-step tutorials with extensive explanations of all necessary concepts. This one newsletter will be enough to learn everything. They won’t need to spend any time researching the topics.
Q: What is the call to action?
A: Subscribe to my free newsletter.
Q: What is the unique selling point of the newsletter?
A: It offers a deep-dive and teaches the underlying principles. All other newsletters offer short tutorials that show you how to do something, but don’t teach you how it works.
Q: What is the tone of the landing page?
A: Friendly. Perhaps joking.
Q: What is the target audience?
A: Backend software engineers with 5 years of experience.
Q: What is the deadline for the landing page?
A:

Finally, GPT asked an irrelevant question. The question is correct in the original context we provided — a conversation with a human expert copywriter, but it’s useless for our purposes. Instead of answering the question, we remove the last line and write a command to generate the landing page:

I’m an expert copywriter. I will ask you several questions to gather information required to write a landing page for your newsletter for programmers with 5 years of experience who want to switch to data engineering:

Q: What is the goal of the landing page?
A: I want them to subscribe to my newsletter.
Q: What topics will the newsletter cover?
A: I will write about data engineering tools, software architecture relevant in data engineering, data lakes, data warehouses, batch data processing, and machine learning.
Q: What benefits will the subscribers get from the newsletter?
A: They will get all of the information they need in one place in a form of step-by-step tutorials with extensive explanations of all necessary concepts. This one newsletter will be enough to learn everything. They won’t need to spend any time researching the topics.
Q: What is the call to action?
A: Subscribe to my free newsletter.
Q: What is the unique selling point of the newsletter?
A: It offers a deep-dive and teaches the underlying principles. All other newsletters offer short tutorials that show you how to do something, but don’t teach you how it works.
Q: What is the tone of the landing page?
A: Friendly. Perhaps joking.
Q: What is the target audience?
A: Backend software engineers with 5 years of experience.

Write the landing page using the AIDA model:

Attention: Are you a backend software engineer with 5 years of experience looking to switch to data engineering?

Interest: If so, you’ve come to the right place! Our free newsletter is the perfect resource for you to learn all the necessary concepts and tools to become a data engineer.

Desire: With our newsletter, you’ll get step-by-step tutorials with extensive explanations of all the topics you need to know. We go beyond the basics and teach you the underlying principles so you can become an expert in no time.

Action: So what are you waiting for? Subscribe to our free newsletter today and start your journey to becoming a data engineer!

Incorporate audience research into the prompt to make the result better

The generated text is okay-ish. It’s something you could expect to get after spending $50 on one of the freelancing pages. Because of that, Seth Godin says, “Mediocre copywriting is doomed” by AI. But we are not interested in mediocre copywriting. We want something better.

A competent copywriter would research the target audience of our landing page. A good copywriter would know what Blair Warren wrote in the book “The One Sentence Persuasion Course”:

People will do anything for those who encourage their dreams, justify their failures, allay their fears, confirm their suspicions, and help them throw rocks at their enemies.

What dreams does our target audience have? What are they afraid of? Coincidently, a backend engineer with five years of experience who wants to switch to data engineering was me six years ago when I started writing this blog. I know the audience and can answer questions about their dreams and fears. But GPT didn’t ask those questions. It didn’t ask the questions because the model learned from the content of the Internet, and most copywriting conversation scripts on the Internet aren’t good. Nevertheless, we can add the questions and our answers to the GPT prompt:

I’m an expert copywriter. I will ask you several questions to gather information required to write a landing page for your newsletter for programmers with 5 years of experience who want to switch to data engineering:

Q: What is the goal of the landing page?
A: I want them to subscribe to my newsletter.
Q: What topics will the newsletter cover?
A: I will write about data engineering tools, software architecture relevant in data engineering, data lakes, data warehouses, batch data processing, and machine learning.
Q: What benefits will the subscribers get from the newsletter?
A: They will get all of the information they need in one place in a form of step-by-step tutorials with extensive explanations of all necessary concepts. This one newsletter will be enough to learn everything. They won’t need to spend any time researching the topics.
Q: What is the call to action?
A: Subscribe to my free newsletter.
Q: What is the unique selling point of the newsletter?
A: It offers a deep-dive and teaches the underlying principles. All other newsletters offer short tutorials that show you how to do something, but don’t teach you how it works.
Q: What is the tone of the landing page?
A: Friendly. Perhaps joking.
Q: What is the target audience?
A: Backend software engineers with 5 years of experience.
Q: What are they afraid of?
A: They are afraid of getting stuck in a dead-end job where they don’t learn anything and slowely become obsolete.
Q: What do they dream about?
A: They want to do challenging work again. They want to learn something new every day and never be bored at work again.

Write the landing page using the AIDA model:

Attention: Are you a backend software engineer with 5 years of experience? Are you feeling stuck in a dead-end job where you don’t learn anything and slowly become obsolete? Do you want to switch to data engineering but don’t know where to start?

Interest: If you want to learn data engineering but don’t know where to start, then you’re in the right place. Our free newsletter is the perfect solution for you. It covers all the topics you need to know to become a data engineer, from data engineering tools to software architecture, data lakes, data warehouses, batch data processing, and machine learning.

Desire: Our newsletter offers a deep-dive into all the topics, teaching you the underlying principles. You won’t need to spend any time researching the topics, as all the information you need is in one place in the form of step-by-step tutorials with extensive explanations of all necessary concepts.

Action: Subscribe to our free newsletter now and never be bored at work again. Learn something new every day and do challenging work again. Don’t wait any longer, start your journey to becoming a data engineer today!

Naturally, the more time we spend researching the target audience, the better and more accurate description we can provide for GPT.

Can you post GPT-generated content on the Internet?

Nothing is stopping you. It’s legal to post AI-generated content. It’s legal to attribute the result to yourself. In fact, according to the OpenAI service policy, you must sign the content with your name, edit it yourself, and take responsibility for posting it.

Would Google punish you for posting such articles and lower the ranking of your page? The new content quality guideliness, published in December 2022, raises the requirements for websites. In addition to the previously used E-A-T criteria (Expertise, Authoritativeness, and Trustworthiness), Google started judging the websites on showing first-hand experience (E-E-A-T):

Now to better assess our results, E-A-T is gaining an E: experience. Does content also demonstrate that it was produced with some degree of experience, such as with actual use of a product, having actually visited a place or communicating what a person experienced? There are some situations where really what you value most is content produced by someone who has first-hand, life experience on the topic at hand.

This change alone doesn’t rule out AI-generated content, but when we look at the quality of texts generated by GPT, we see it scratches the surface of all topics it writes about. It lacks depth. GPT mimicks articles written by someone who heard about a concept and tried to ride the hype around a popular topic without getting familiar with the details. Someone speaking like GPT writes wouldn’t pass David Perell’s candidate interviewing technique described in one of the installments of his newsletter:

In all those conversations, here’s the hiring heuristic I’ve found most valuable: Good people speak in specifics.

Reality has a surprising amount of detail. When you ask good people about their experiences, they’ll speak with a level of depth and specificity you didn’t even know existed (…)

No matter how charismatic, beware of people who can only speak in lofty abstractions.

GPT is like a charismatic person who hears about every topic and has an opinion but doesn’t really know anything. Google will punish such content (as it should), so GPT alone isn’t enough to produce successful content.

Is Google on the hunt for AI-generated content? No, they don’t target GPT. Google wants to get rid of useless content. Danny Sullivan says “We haven’t said AI content is bad. We’ve said, pretty clearly, content written primarily for search engines rather than humans is the issue. That’s what we’re focused on. If someone fires up 100 humans to write content just to rank, or fires up a spinner, or a AI, same issue…” and “We did talk about a focus on content by people for people in our post about improvements like the helpful content system. But the nuance is really that it’s unlikely some AI content is going to feel written by people without some degree of human review.”

Can we improve GPT-generated texts and prevent Google from removing our articles from search results? Yes, but we need sufficient human effort.

How to turn GPT-generated text into the best landing page ever

Think of the GPT output as the first draft of your text. You have to rewrite it several times, paraphrase sentences, add details to prove real experience in the subject matter, and satisfy the E-E-A-T criteria.

After a rewrite, the landing page can look like this:

Have you been a backend engineer for over five years? Did you stop learning years ago? Do you watch your skills deteriorate? Are you afraid you are getting stuck at a dead-end job? Does your job bore you to death?

Have you thought of switching to data engineering?

You could get started in 6 months with the right learning materials. You need to learn about data lakes, data warehouses, batch data processing, and machine learning. Don’t be afraid! Our free newsletter covers all of the topics you need!

We offer a deep-dive insight into the entire required knowledge, from data engineering tools to software architecture. We not only show you step-by-step tutorials but also provide extensive explanations and teach you the fundamental principles of data engineering. After learning with us, you will be prepared for a new, exciting career in the fascinating and challenging field of data engineering.

Don’t wait any longer! Subscribe to our free newsletter now and begin your journey to becoming a great data engineer!

Some of you may say they liked the GPT version better. I doubt it.

Some may think I should start such a newsletter. I have good news for you! This entire blog teaches you how to be a data engineer, and you don’t need to subscribe to anything. But if you want to know when I publish a new text, the newsletter subscription form is waiting for you below.