HOME | WRITING | NOW | SERVICES

My AI Copywriting Guide

As someone who has spent his entire career learning to write good copy and become a professional copywriter, AI has completely shaken the foundation of my world.

It’s both changed how I approach my day-to-day and my workflow. As well as how I view the future of my career – what will it look like to be a copywriter in 6 months? Or a few years? Or even beyond that?

In the past six months or so, the AI question has become completely inescapable. Both because it’s the LLMs have simply gotten better and because it’s entered into the public awareness.

There are a ton of professions being impacted by AI, but copywriters definitely are among those who feel it most pressingly.

I think this is partially because GPT was the first publicly-avaiable AI product that was actually good. Before that, the only popular AI use cases were images and videos – both of which had very weird, eerie and surreal results.

Even in the beginning days of GPT 3, you could generate some pretty great writing if you’re able to use it well.

Personally, I don’t think that AI will replace copywriters entirely. But I do think that writers who can utilize AI well will have a major advantage over writers who can’t.

And that’s the thing – right now, I’m not sure that most people think of using AI… or more specifically, I should say, prompting AI… as a “skill.”

But it is a skill. Because the way you prompt AI will significantly impact the quality of responses you get. In order to get better responses, you’ll need a better prompt.

But many people don’t view it that way. They prompt AI with something extremely vague and generic like “help me write a landing page for this supplement brand” and then when the response is boring, middle-of-the-road content, they say something like “AI just isn’t there yet.”

But it is. If you learn to use it correctly.

And this is a perspective – this thinking of prompting AI as a skill – that is helpful for copywriters.

It gives us agency in this world of AI. Instead of worrying about AI replacing us, we can get excited about new ways to practice using AI in our work.

It gives us a future worth working for. Copywriting has always made me decent money, but to be honest, I’ve always seen it as step 1 in a wider marketing career. I think that really getting to know AI well will be a nice launch pad into that wider marketing career.

And so in this guide, I’m basically going to give a run-down of the most beneficial ways I’ve found to use AI when writing copy, and how I’m approaching it moving forward. How I use it “well.”

The New Role of Copywriters

Copywriting is no longer just about stringing together catchy and creative words. Instead, it has evolved into a strategic role where copywriters act as the architects of content.

Think of yourself as the lead copywriter, and AI is your junior copywriter. You are the one giving the brief and making final decisions based on taste. AI can help you generate ideas for the actual words.

In this role, your focus should be on the purpose of the copy. Which may actually be a bit of a blessing in disguise. It helps you focus less on the pretty-ness of the words (I think all writers are guilty of overemphasizing this part) and more on their impact.

We’ve all been there. When you write something that feels really clever or witty or something like that. It made you feel like a good writer when you came up with it. And then when it comes to edit the piece, you realize it could be eliminated – but for some reason, it’s harder to remove than the other parts you edited out. So you justify keeping it – though it doesn’t truly have the impact you’re looking for.

I think that using AI to help write copy removes that piece a bit. I still feel proud of copy that I use AI to formulate, but I don’t think I have that same connection. Maybe we’re losing a piece of something there, but I think it does allow us to step back and really think about the purpose of the copy.

That’s why I’m personally approaching this next phase of my career by re-focusing on the basics… the timeless principles of persuasion and influence that make people buy.

Because now my value to companies isn’t in helping them generate the words… AI can generate plenty of nice words for them.

It’s in helping them use these words to move customers to action.

I think this has always sort of been our biggest value as copywriters, but it’s much more clear now. That’s for sure.

And every word, phrase, and sentence should have a clear purpose and move the reader toward action. When reviewing a piece of copy, ask yourself:

💭 What is the purpose of this?
💭 Does this move the reader toward action?
💭 Is this the most compelling way to say this?

And then from the reader’s perspective…

💭 Will this grab my attention?
💭 Will I give enough of a crap to keep reading?
💭 Is there a compelling enough reason to take further action?

Your role is to direct AI, refine the message, and ensure every sentence serves a purpose.

Rules For Writing With AI:

1️⃣ Use AI as a conversation partner, not a “one-and-done” writer.

This is the biggest mistake I see. Typically it’s from other marketers and just people who don’t really understand the role of AI and its power, but I have heard this from writers too.

It’s when they insert one prompt… typically a pretty vague one like “help me write this landing page for [X brand] selling [Y product].

And then they’ll say things like “AI isn’t quite there yet…”

But the truth is that prompting strategy is only going to get you vague, wishy-washy results whether you’re using ChatGPT 4.0 or 98.0 – it’s not that the AI is lacking, it’s that you’re not using it correctly.

Instead…

2️⃣ Use AI to expand your thinking.

As human writers, our measly little brains have limitations.

Our ideas are shaped by…

Personal experiences (with a bias toward recency).
Biased thinking (that we don’t know we have).
Moods (simply how we feel when we show up to the writing).

AI, on the other hand, can:

💡 Offer perspectives you wouldn’t have thought of.
💡 Surface counterintuitive angles.
💡 Change the direction of your train of thought.

3️⃣ Use AI as your assistant… not a boss.

It’s a thin line, but the difference is critical:

You must use AI as a tool for better thinking. NOT something you can delegate your thinking to.

When you outsource strategic decisions to AI, you’re choosing the average – the bland middle ground – because AI always defaults to the mean. It learns from massive amounts of generic content… most of which is bad.

So it needs to be you in the drivers seat, understanding the goal you’re trying to achieve, understanding the basic persuasion principles that’ll help you achieve that goal, and then ultimately deciding which will work best.

There’s also a selfish reason to not completely delegate your thinking to AI:

And just like muscles weaken when you stop using them, your writing and strategic skills will deteriorate if you rely entirely on AI.

At least that’s what I believe… and I’m too scared to experiment with it myself so I’ll assume that’s the truth.

4️⃣ Prompts are a tool to “think out” the problem.

This is a recurring theme about the way I’m thinking about AI these days:

The real value of AI doesn’t come from the content it generates. It comes from the clarity you gain by asking better questions:

What core problem am I really trying to solve with this copy?
Where is my reader emotionally and mentally when they see my message?

Good prompts help you take the objective from…

“A task I have to complete for X marketing initiative.”

“A marketing problem I’m trying to solve.”

And sometimes, that difference can be lost in the day-to-day monotony for copywriters, especially when you’re writing for a lot of different projects and a lot of different clients.

5️⃣ AI isn’t just a “work friend.”

A lot of my understanding about AI comes from the ways I’ve been able to use it outside of just copywriting or “work-related” things.

It definitely started there, don’t get me wrong. But taking this amazing tool to other aspects of my life has really helped me to understand its powers and be able to better utilize it in a work setting.

Some ways outside of work I’ve been able to use it:

1: Reflection journals → prompt like “give me 10 journal starters in the voice of David Whyte or Tim Ferriss or Liz Gilbert”

2: Recommendations → give it things you like, ask it to make suggestions based on that.

3: Themes → give it a lot of disparate information, and ask it to distill main themes.

4: Interactive learning → upload podcast transcripts / books / courses… have a discussion with AI based on that material – have it point back to the parts it references.

6️⃣ Always fact-check AI.

Hallucinations are simply part of the package – at least with today’s LLMs. Maybe it won’t always be the case, or the frequency will be less, but for now – there needs to be a human fact-checker for anything that is AI generated.

How It Works + Why It Matters

No need to get into the technical aspects of an LLM. But there is some very basic background knowledge about how this tool works that’ll help us better understand how to utilize it.

Here’s the simplest way to think about AI tools like ChatGPT:

AI is a massive aggregator that has read nearly every piece of publicly available content on the internet. That’s its superpower – the ability to comb through massive amounts of information. But it’s also something to be mindful of.

Because when it learns from all the content, it will also be learning from the bad, the bland, the mediocre.

So when you give it a vague prompt… the result is the very standard “bad” marketing. The cliche phrases… the tired claims that don’t mean much.

If your input is something basic like, “Write a marketing email for my product,” AI searches its database, averages out every mediocre marketing email it’s ever seen, and produces something bland, generic, and forgettable.

So the solution is to be precise with your prompts, and give the AI the information needed to succeed.

If you want it to help write “good” copy…

#1: Tell the AI what good copy is. Provide AI examples of your top-performing copy.

#2: Tell the AI who you’re talking to. Detailed descriptions of your ideal target audience, and your brand’s relationship to them.

#3: Tell the AI what makes your brand unique. If you don’t focus on factors that actually make your brand unique and different from competitors, it’s inevitable that the AI will provide super generic and bland responses.

#4: Show the AI your brand’s unique voice. This is the anecdote to AI’s tendency to write very vague, generic, fluffy responses. Give examples of how you write… it will follow suit pretty nicely.

–

All of this to say, you need to remember what AI actually is and how it functions:

These tools don’t “think” in the way you and I do. They merely reflect and recombine what they’ve learned.

Your job, as the strategic leader, is to clearly define your standards and expectations upfront. The more carefully and thoughtfully you brief AI, the better the results.

3 Ways to Use AI for Copy:

1️⃣: Ideation: AI is a great way to spark ideas & expand your thinking. It can ideate a ton of options for whatever you need (angles. hooks, headlines, CTAs), essentially speeding up the creative process.

On some level, I think that creativity is throwing a bunch of shit at the wall and then having good enough taste to preserve what works.

In that case, AI can help tremendously with the “throwing a bunch of shit” portion of that.

A lot of the time I’ll start off with my prompt → if I’m writing an email, I’ll give all the context of that email (the list I’m writing to, who the sender is, what I’m trying to accomplish) and then give a prompt like

“give me 10 angles/ideas to approach this email with that would achieve my goals for this email”

It’ll give me some good ideas. Some bad ones. But if I can get my brain stretched in some different directions that it wouldn’t otherwise have, than it’s worth it to me.

Sometimes I’ll like one of the angles it gives me. Many times it sparks an idea that I have, not quite what the AI generated but adjacent to it.

So in that sense, it’s back to that recurring theme:

AI isn’t just a great tool for writers in the stuff it generates. It’s the way it makes you think.

2️⃣: Research: Using AI as a super-quick & fairly reliable research assistant.

To clarify, I do NOT mean using AI to gather research. As I mentioned before, AI will hallucinate facts and it will always need to be fact-checked → so not a great research assistant in that sense.

But when you aggregate a bunch of data for AI… that’s when it can be gold.

As a copywriter, one of the best use-cases I’ve found is for customer research. Basically, I’ll:

1: Aggregate a shit ton of audience reviews about the brand I’m writing for

2: Paste them into GPT and ask questions like…

“What are the top 3 positive themes that loyal customers praise about this brand?”
“What are the most frequent words or phrases that loyal customers use to describe this brand?”
“What are the strongest emotional drivers behind the positive reviews”

This approach can be used for basically any sort of research. When writing for health brands, I’ve given my handy AI research assistant the task of combing through lengthy research articles and podcast transcripts to…

1: Summarize for me → “explain it to me like I have a very basic understanding of science”

2: Extract relevant information → “paste all the excerpts from this piece that are relevant to [whatever I’m writing about]”

When it comes to using AI for research, always ask it to source its answers. Agan, AI is pretty unreliable when it comes to facts… which is a pretty big flaw for a research assistant.

But it can comb through thousands of pages of data in seconds… so it’s a tradeoff.

3️⃣ Refinement: Using AI as a way to strengthen existing copy.

I love using AI in the editing phase. Since it’s such a great tool for creating variations (which we explored in the first use case, ideation), it is really good for refining what you’re trying to say.

So let’s start with the basics.

In the editing phase, when I see a sentence I want to change…

Maybe I like the gist of the idea I’m trying to express, but it’s not expressed clearly enough. In that case, I’ll enter the sentence into GPT with a prompt like “Give me 5 alternative rewrite suggestions that make this more clear and straightforward for this audience.”

Maybe I like the copy, but I think it’s a bit too long and may lose some of the readers. “Give me 5 alternative rewrite suggestions that make this more concise and compelling to read for this target audience. Rewrite it so that each sentence of copy has a goal of getting the reader to NEED to read the next sentence in order to satisfy their curiosity.”

This refinement phase is where your taste really comes into play as a copywriter. The editing phase can simply be a period of putting yourself in the shoes of your audience (as much as possible), reading through what you’ve written, and simply seeing what comes up…

What do you like? Maybe you like a sentence that describes a certain benefit of the product you’re selling. Maybe you could ask AI to give you 5 variations of sentences that highlight that benefit.

What don’t you like? Maybe it’s a sentence that feels important, but like it’s a bit out of place in the current context of your copy. Maybe the prompt can be “give me 5 alternative rewrite suggestions for this copy that make it flow better with the rest of the piece.”

This is where it becomes super important to think of yourself as the architect of your copy. Because that’s truly what we are and where our value lies.

It’s not in the words. It’s in making the words work.

Some of the things I’m typically refining copy for:

– More persuasive
– More compelling
– More emotionally driven
– More authoritative
– More concise
– More readable
– More relatable to [specific target audience]

Conversations

Now let’s get into the details of your different conversations with GPT – essentially, every time you click ‘New Chat’ is a new conversation.

Every time you start a new conversation, you need to give the AI context all over again.

Think of the context at the front-of-mind information that the AI is using to generate responses.

The AI has this endless information bank, and because of that, it can only use so much of that information at one time to generate responses.

To make that a bit more concrete, let’s think of ourselves. You know how you know a whole bunch of stuff… but most of it isn’t top-of-mind. Meaning, you know a piece of information is true, but haven’t thought about it recently enough where you would use that piece of information to answer a question.

So every time you start a new conversation, you need to give AI the context to help you succeed. Refresh its memory with the most critical information needed to write for the brand you’re writing for – bring this information top-of-mind.

So then comes the question – when do we want to start a new conversation?

I find that the more an AI is prompted different things – for example, if you’re writing for the same brand but about different topics… the more and more topics you prompt it to help you write for, the more confused the responses will get.

Details from one of the prompts may seep into another.

When that starts to happen, I start a new conversation.

That’s why I think it’s super helpful to create brand guides for each of the brands you’re writing for. And like with many parts of copywriting with AI, the practice of creating these brand guides is just as helpful as being able to actually use them with the AI.

But I think that these brand guides should be a comprehensive background of all the context the AI could possibly use to best write copy for this brand. Incluidng things like:

– Who is your target customer? What are their pains? What are their aspirations? What do they care most about your brand?

– What is your product? What makes it concretely different from competitors? Why do loyal customers say that they prefer your product?

– What is the tone of voice? What are some great examples of that tone of voice? What do you NOT want the voice to be?

And so on and so forth.

The things you include in each guide will be different for every brand, depending on what you’re selling, who you’re selling to, and the quirks involved with each.

The Future of Copywriting

As AI makes generic content easy to produce, authentic human experiences and personal stories will stand out even more.

As writers, we have to think about what AI has access to and what we have unique access to. AI has access to all the information… so it wins there. And that’s no small win.

But we have access to our unique experiences and our unique ways of processing those experiences. That’s our competitive advantage, and I think we must use it.

I think this will lead to the accelerated importance of personal brands… something that was already becoming more important, but I believe will be even more important.

Some other predictions about AI…

I think that understanding how to prompt well will be a valuable skill.

Right now, I don’t think most people value it as a skill. They treat it as binary… something that is “good” or “bad.” They don’t understand it as a tool that will deliver different results depending on the person wielding it.

This is a massive distinction with a ton of implications, especially for copywriters.

Understanding AI as a tool and learning to best use it gives you your agency back as a writer. Because if we’re not “practicing” using AI in the best way possible… then what are we doing?

Just waiting until it gets good enough to replace us?

I don’t believe that’s the case. I don’t think we’re just waiting for AI to replace writers. But I DO think that writers who can best utilize AI will replace the writers who don’t.

Luckily, I know the majority perspective on AI will change with time, and being able to articulate a prompt that delivers the proper results will not only be valued as a skill… it’ll be a highly compensated one.

We’ll see, of course. But that’s sort of what I’m hoping.