How I Used ChatGPT to Rewrite OpenAI’s Website Copy (Experienced Copywriter Tutorial)

Here’s an “over the shoulder” tutorial on how an experienced copywriter applies ChatGPT to the entire copywriting process – research, data organization, idea generation and writing.

Want to know how to ACTUALLY use ChatGPT for copywriting?

Most of what you read out there isn't written by real practitioners who use AI in their daily grind. I’m here to change that.

This guide shows you how I (an experienced copywriter) integrate ChatGPT into my workflow for research, data organization, idea generation and (of course) writing. 

This is more or less how I use ChatGPT in my client work. Sometimes I use the outputs, sometimes I don’t. But every step adds value to the process.

To give you a concrete example, I’ll walk you through a hypothetical scenario: rewriting OpenAI’s website copy using ChatGPT.

Full disclosure: I didn’t work with OpenAI on this, but this example will show you how a pro would leverage ChatGPT in a real-world context (OpenAI, if you’re reading this, let’s talk!).

Ready to see how ChatGPT is an asset to the entire copywriting process from research to final draft?

Let’s get going.

Let’s say OpenAI was running some A/B tests (they definitely have the traffic for it) and wanted to experiment with different positioning for their ChatGPT for Enterprise page.

What strategic hunches might they want to test? 

What type of messaging might they try out? 

Maybe they’re building an enterprise sales/marketing team and want to see how different messaging resonates with each segment.

They want to reposition their story to make it clear that companies are using ChatGPT to solve some of their hardest problems. 

They need to overcome the objection that ChatGPT is just a cute toy and show the world the concrete use cases and bigger possibilities.

Step 1: Gather unique, original data:

I don’t have access to any internal data, but where are some obvious places to look? 

Their case study library is full of quotes and data I can feed the model. Let’s use that.

I go into the case studies, one by one and pluck out the sticky quotes with usable language.

Key: start with very human data.

Now where are more places online where business people talk about AI solving their problems?

Amazon book reviews about enterprise AI.

Tooooooooooons of results:

Lots more datapoints to dig into.

Data = leverage in AI.

The more, better, unique, proprietary data you have the better you’re going to do with AI.

Ok so either keep your data in GPT thread or paste into a spreadsheet. Use that repository of quotes to write your copy variations. 

And this is the output:

Time saved: ~3 hours

Step 2: Write copy variations.

The goal is to answer people’s pains, questions and objections (which we just figured out what they are through VOC data mining).

This is how I generate a big list of questions to work from:

ChatGPT’s output:

Pro tip: Never give a model a prompt without context or starting points to build on. If I only asked it for objections without a starting list it's not going to give me the best result.

Time saved: ~1 hour

Your page would be a mile long if you tried to get to all of these, but I’m going to choose the ones that make the most sense for this “project”.

Each section rewrite will answer the following objections:

• “How will this actually help us with our work?”

• “How do we approach this tool?”

• “How does this fit into our big picture?”

• “How do I know this won’t replace our people over time?”

VOC used:

Objection addressed: “How will this actually help us with our work?”

---

Let’s work our way down the list:

VOC used:

Objection addressed: “How do we approach this tool?”

---

And the main headline:

Objection addressed: “How does this fit into our big picture?”

---

With their traffic, this is a shot to really mic drop and reinforce their positioning. The privacy and security stuff can go below the fold.

VOC used:

Objection addressed: “How do I know this won’t replace our people over time?”

Final thoughts

No one has really laid out solid frameworks and principles for thinking about AI—stuff that doesn’t change. People get wrapped up in the news, tech updates, and new functionalities, which are out of your control and constantly changing. 

I'm focusing on the fundamental skills and thought process.

This is the type of thing that makes a copywriting skill set defensible. You have experience across a lot of clients. You’ve seen businesses. You understand buyer psychology. Use that. 

If you have strategic ideas, use that to edit the copy via natural language and simply explain the changes you’d like to make, why, and how, and have ChatGPT do it. 

It’s an immediate iteration and application of your ideas applied to something concrete like website copy. 

ChatGPT is this weird, scalable machine that can help you scale your creative process if you know how to use it right.

Use your expertise to guide these experiments. Think about the fundamentals of buyer behavior and apply that to your copy.

That’s where the magic happens. 

And there’s plenty of weirdness left to explore.

Like this guide? Subscribe and get more of them over time.