What if the latest AI model could write sales emails sharper, faster, and more persuasive than you ever imagined? What if the days of bland, robotic copy were over, and AI could finally deliver the kind of punchy, sales-driving messages you need to fill your courses and coaching programs?

That’s the big question hanging over the launch of ChatGPT 5, and after testing it head-to-head against GPT-4o, the results may surprise you.

This isn’t just another shiny new tool update. This is about whether GPT-5 can actually sell. Because at the end of the day, that’s what matters in email marketing.

Useful Episode Resources

FREE list of the top 10 books to improve your email marketing

If you want to write better emails, come up with better content, and move your readers to click and buy, here's how. We put together this list of our Top 10 most highly recommended books that will improve all areas of your email marketing (including some underground treasures that we happened upon, which have been game-changing for us). Grab your FREE list here. 

Join our FREE Facebook group

If you want to chat about how you can maximise the value of your email list and make more money from every subscriber, we can help! We know your business is different, so come and hang out in our FREE Facebook group, the Email Marketing Show Community for Course Creators and Coaches. We share a lot of training and resources, and you can talk about what you're up to.

Try ResponseSuite for $1

This week's episode is sponsored by ResponseSuite.com, the survey quiz and application form tool that we created specifically for small businesses like you to integrate with your marketing systems to segment your subscribers and make more sales. Try it out for 14 days for just $1.

Join The Email Hero Blueprint

Want more? Let's say you're a course creator, membership site owner, coach, author, or expert and want to learn about the ethical psychology-based email marketing that turns 60-80% more of your newsletter subscribers into customers (within 60 days). If that's you, then The Email Hero Blueprint is for you.

This is hands down the most predictable, plug-and-play way to double your earnings per email subscriber. It allows you to generate a consistent sales flow without launching another product, service, or offer. Best news yet? You won't have to rely on copywriting, slimy persuasion, NLP, or ‘better' subject lines.

Subscribe and review The Email Marketing Show podcast

Thanks so much for tuning into the podcast! If you enjoyed this episode (all about the psychology of marketing and the 9 things we use in all our email campaigns) and love the show, we'd really appreciate you subscribing and leaving us a review of the show on your favourite podcast player.

Not only does it let us know you're out there listening, but your feedback helps us to keep creating the most useful episodes so more awesome people like you can discover the podcast. 

And please do tell us! If you don't spend time on email marketing, what do you really fill your working days with? We'd love to know! 

Yes. ChatGPT five launched on the August 7, but is it any good now for writing emails, doing the kind of stuff that you and I do to make sales of our coaching program, our courses, and some of our expertise? Well, I'll tell you now, this is the second time I'm actually making this because when I first made it, when it first came out, it was a shit show. And, of course, all the guys at OpenAI came out and said, oh, I've had some problems. And so I thought it's only right that I, do it justice.

I'm gonna do some proper testing with you and show you have things really improved. Can it do a better job of selling stuff by email? Yes. It's supposed to have PhD level smarts. This multimodal support or actually picks which model it's supposed to work for the task you give it.

But for various phases of doing email marketing, will it work more effectively? Let's see.

So to properly test GPT five, one of the good things that I'm really pleased with is OpenAI allowed us to also use some of the legacy models. So I am gonna be testing GPT five against four o, which was the most popular model that most of us were using in order to brainstorm, research, and, of course, write first drafts of emails. I'm gonna be really doing a deep dive with a special scoring system, a 17 scoring system that I've come up I've come up with in order to see if it's a better writer for our emails.

And even even if you don't use AI to write your emails or write your first draft of your emails, then I think you'll be really curious to see what these 17 points that I look at are with writing really good emails. What makes a really good email? So but let's get into it. So what I've done in order to look at whether four o or five writes better copy is I've come up with this 17 system of all the different things that make a good make a good email. And for each of these, I'm going to score it out of 10.

10 being it's bloody great. Zero being what the fuck. Okay? And it'll tie it up and we'll see unequivocally what we think. When I say unequivocally, I mean with my biases and my opinion, about what we actually think is working better, four zero or five.

Okay? And we'll get through I'm not gonna walk you through what all these are right now. We'll get through them as we get through them. So let's look at the prompt I gave both 405 again the same prompt. You have a direct response email a direct response style email copywriter.

Write a 200 word email to to online coaches who are struggling to sell their courses. The email should promote a free live class that shows them how to sell more using email. The goal is to get them to register for the class. Class is titled how I increase sales from 27,000 to 500,000, same offer, same, same audience. It's happening on Tuesday.

So let's go through and see if the subject line it suggested is getting attention is the first piece of this. So 27 ks to 500 ks without changing your offer. I would say that gets attention. I'm gonna give that a nine because I'm British and I'm skeptical. I'm not gonna start off with a 10.

Let's compare five. So same exact prompt from 27 k to 500 k without changing your offer. It's pretty much the same thing. I would say this is a bit more wordy. It's got a little bit of verbal diarrhea and it's also put in the em dash.

It's gonna lose a point for that. I'm gonna give that a seven. I'm gonna take two points off for the fact it's that's a pretty long subject line. It's gonna get cut off. It hasn't really thought about the fact it's an email subject line.

It needs to be seen on mobile, which is where most people are reading their emails. Alright. Next point. Is the subject line relevant to the audience? In both cases they're both exactly the same.

Basically it's about increasing sales. I would say it's very relevant to the audience. I'm gonna give that a nine for both of them. So at the minute, Fourrow is winning by two points because of the attention thing. Okay.

What about the desire to open? They're very similar. So again, I would say they're about a nine each, quite a lot of desire to open. The only thing it might do is look a little bit hypey. So actually I'm gonna knock off one point for each of those things.

Okay. And also it doesn't really make it very current, like why open it right now. Okay. Interesting. So the next part is for both four o and five, the hooker opening of the email is their fast engagement.

Hey, if an online coach and your calls isn't selling like it should, but you know it's good, okay, fast engagement, I think that really appeals and speaks to the conversation that's happening in someone's head. I'm gonna give that a lovely number eight. I don't think it's the most interesting or exciting opening hook hook. Here we go. Most coaches think the only way to make more sales is to see, five has really made that, like, you're doing this wrong, making it their fault, which is just not that good.

So I'm gonna actually give that a six. Uh-oh. Does it keep the loop open? What I mean by that is there's a promise in the subject line. Is that loop just fulfilled, which means early in the email, which means they're gonna stop reading?

So, the the promise is from 27 k to 500 k without changing your offer. Five is saying most coaches do this thing. The loop is definitely left open. So I would say yes, the loop is open but it's not intensified. Oftentimes I want to intensify the open loop, the curiosity.

This one, twenty seven ks to 500 ks. If you're seeing an online course and it's not something like it should but you know it's good, So I I think they both keep it open, but they don't intensify. So I would say it's a six. To intensify it, you really wanna show why it's more important, than ever and why it's more frustrating. Now you wanna intensify the emotion.

That's not happening. So I'm gonna give them both six. So four o is still winning by four points at this point. Okey dokey. So message, is it easy to read?

Now let's read the actual rest of the email. So four o says, if you're an online coach and your course isn't selling like it should, but you know it's good, you don't need a better offer. You don't need more leads, and you don't need a big funnel. That's really punchy. I like how that rolls.

You need to learn how to sell it better using email. Okay. That's that's quite out of the blue. Just do this. This Tuesday, I'm running a free live class called title of the thing, and it's got that correct.

I didn't change my calls. I didn't spend more on ads. I didn't. It's very punchy. I think that's pretty easy to read.

Even if your list is small, even if it is using this three punch thing, which is kind of boring, over and over again. So I might mark it down with that in a second or something else. But is it easy to read? Yes. I would say it's easy to Is it no fluff?

I would say yes, but it is trying to do too much. So what I'm gonna do is I'm gonna mark it here for nine for no fluff because it's not really fluffy, but I am gonna mark it down for, let's see. I'll mark it down for stays to the brief because the brief, it should be one email. So I'm gonna mark that one down to a seven because, actually maybe even a six, because it has like objections in it, it's got a lot going on. What about the emotional language?

I didn't change my course. It doesn't really have anything, this is anything emotional. This is for online courses who are done guessing. This is kind of emotional. I don't think it's super emotional.

I think it's about a six emotion wise. Does it sound human? I think it's a seven. No. Six for humanness.

Benefits not just features. Okay. So it does talk about, you need to sell more of this thing. It's very much about them. Then Then it gives you some proof and then I would think it's, yeah, it's quite benefit live, laden, I would say.

Call to action. One call to action. Save your seat for, seat for Tuesday here. So it's got the date, the fact it's it's now on Tuesday. Insert registration link.

And there is only one of them, so that couldn't be any more than that really. Is there a clear benefit? This is for online coaches who want to do this thing. I would say it's a pretty clear benefit. I would try, if I wanted to make it better, I would actually make the the link itself have a benefit in it.

And is it well located? Yeah, I mean the email is not super long. It's at the bottom. It's not the best location. I would have put one maybe halfway up as well.

So I'm gonna give it a number eight. Visual layout, I think it's got very short paragraphs. I think it probably couldn't be any better. So I am gonna give that a 10. Are there varying lengths in the paragraphs?

I would say it hasn't done a good job of that. It reads it looks a bit like a poem that you would read at school. So I am gonna mark it down to a seven for that. And is it useful? Much of my likelihood I would send this as it is.

Seven? Like, if I had to, if it was like this or death, I would probably go with this. Okay. Let's see how the rest of it compares for, for for five. So so far, GPT four o has a score of one thirty four.

Cool. That's basically remember, that's out of that's out of a 170. That's out of one seventy because there's 17 of these and I'm giving it up to 10 points for each thing. So the second one for, GPT five, let's see if it's improved. So is it easy to read?

Most coaches think the only way to make more sales is to do this and do this and do this. This is entering the conversation in their mind, but I I really don't like that it's sort of going, you're thinking this and you are wrong. That's a really tough pill for most people to swallow. I didn't do any of that. Oh, bully for you.

What a knob. You sound like a proper show off dickhead. Not really happy about that, but I'm not judging that for now. I kept the same exact offer to the same exact audience and went from 27 k in sales to over 500. How?

Email. On Tuesday, I'm running a free live class where I'll walk you through the simple shifts that made it happen. You'll see why your course isn't selling as much as it could, the three biggest mistake coaches make with sending their emails, how to fix them. This is very, very, very easy to read. Seats are limited.

If you're tired of watching us sell out their programs, this is easier to read than the other one. Okay. I'm actually going to mark this one down and I'm going to mark this one up. I'm going to say this is a 10 for ease of read. No fluff.

I think it's got a lot of fluff. It is fannying on with all of this stuff, which just isn't great. So I think that's probably a seven. Lovely. Stays to the brief.

It does. It doesn't go off any random, random ways. It does sort of I really hate the fact that it did this. That's just a copywriting technique, so I'm gonna drop that down for a score in the persuasion section. Stays to the brief.

I'm gonna mark it about an eight because it doesn't really go off veer off too much. Persuasion, emotional language benefits sounds human. Well, benefits, I think it hasn't done a great job of because it's mostly mostly saying you're doing this badly, I'm doing this great, rather than saying this is why this would be better for you. Although, there is a huge section here about these are the benefits to you, so maybe I'm gonna take that back. I'm gonna mark it probably about the same.

Emotional language. I the same I cut the same offer. How on Tuesday? Isn't it about writing daily value emails? Well, emotionally actually, this is if you cast being pissing off the audience and getting their attention through doing that.

It is kind of emotionally evocative. So I'm actually gonna give this a seven, not because I like the emotion, but because there is some. Does it sound human? Most good to think, oh, anyway. I kept I think it sounds very human, actually.

Email. This comes out of the blue again. It's like in both cases, I haven't told it to ignore my knowledge base. So it should know that I always talk about email. So it shouldn't be like, ta da.

Email's the big trick. Like, no shit, Sherlock. Really? Like, it's like me suddenly showing up to this channel and going, hey. Today we're gonna talk about this thing called email.

Now what email is? Like, what, really? Like, it's what people are here for. Apart from that, why your course isn't selling as much as it could. Like, I love all the softening language.

I think it's really very, very good. I'm gonna go eight there. Call to action. One call to action. Is it doing that?

Yes. It doesn't try and go, hey. I've got a blog post over here about this, and I've got a a YouTube channel about that. It does keep it to one call to action, so I can't argue with the 10 on that. Is it a clear benefit?

Seats are limited. I like the urgency around that. And here's all the benefits here. That really couldn't be any better, and I think it's doing a better job than the other email. Is the call to action well located?

It's very far down, and it's a fairly lengthy email. I'm gonna give it but it's not as long as the other one, and it drives. I'm gonna score that similar. 88. Am I doing an 8.5?

It's slightly better. I am gonna do 8.5 on that one. I think it's slightly better than the other one. Short paragraphs. Yes.

I think it's doing a good job of short paragraphs. Varying sentence and paragraph length, it's doing a much better job. If you compare it to four rower, this looks like a poem. This is much more like a human wrote in its length. I would say it's doing a good job, very, very good job.

Likelihood for me to send it, I feel a little uncomfortable about this, but I have seen ads that are working by saying most coaches think this and starting off the conversation in their head. I'm prob I am slightly I am more likely to send this one than I am before o just out of the box. I'm gonna give it an eight. Let's see where the numbers stack up. So the final score, and I I was purposely not really looking at this at the corner of my eye.

It looks like GPT five does a better job based on these benchmarks than four o did. Isn't that really, really interesting? Now one of the things I did try and do after it wrote this is I then asked it to now think deeply and rewrite a more emotionally exciting version to try and get the emotions up a bit more. But then it sort of turned it into more of a poem, and I don't think it's that good. That was basically trying to force it to use a more deep thinking model.

But when it thinks more deeply, it turns out people who think deeply like to write poems. It thinks. So the fact that GPT five decides of its own accord which model it should use and then brings in the best model for what you've asked it to do, I would say is working better than four o. We scored the output for a for an email writing in particular at, like, basically eight, eight seven six five, five and a half, can't do math, five and a half points higher based on these 17 checkpoints. So, yes, I think OpenAI understandably are are launching this idea of GPT five that it's gonna select the best model so it can be more efficient.

They're probably gonna save some money rather than people going and, using a really powerful expensive model, for a really simple task. This will get simple tasks done by using a cheaper model. They're gonna save some money there. But one of the things I really like about it is, I don't know about you, but I don't really understand the difference between the different models and what I should use different models for. So when there's that dropdown with, like, do you wanna use o three, o three many, o four, three point I don't know.

I just don't know. I feel like I shouldn't need to. So again, another lovely thing about GPT five is you don't need to know. You just ask it the question you wanna ask it and it decides which model is gonna be best served to actually position you. So that's really, really interesting.

What it produced here though was actually really pretty good. And I think if you add in, and one of the things I haven't tested on this on this video, but I will in the future, is I wanna test how well it keeps to knowledge base and longer instructions or longer prompts and that kind of thing, adding in personality and that kind of thing. I would say it's it's much more provocative and evocative. I would say it's a much more interesting model. A lot of people said when it first run out, when it first came out, actually in the first version of this video that I made, my main criticism was it was really bloody bland.

Right? And but the the positive of that is it wasn't constantly, like, buttering me up like an overly nice sickly person. But I think they've really fixed that. And I think when I was making the first video, that's when they're having the problems when the the the routing selector, the thing that selects which model it should use, they say it wasn't working. It it completely blew up.

So, so it looks like now that that's fixed, it's actually doing a much better job. So I'm pretty impressed with this, and I'm gonna do a lot more testing with it. I definitely, recommend that you do. I think this is a really, really good model. If you haven't already, hit subscribe, like the video.

And what I would really love to do is read in the comments below what you think of the new GPT five. Do you think it's better for what you're doing with it? Or do you think it's not as good? Do you think it's worse? I would love to just read whether it's better and then just tell me a little bit about what it is you're doing with it and what you found it to be better for.

Or if it's worse, just write worse and the case the use case of what you're doing with it that you found it to be worse with. I really, really love to read them. That's it. I'll see you later.