Proven Podcast - Mastering AI to Increase Sales - Andy Crestodina

Episode Date: May 7, 2024

In this episode, Charles sits down with Andy Crestodina of Orbit Media Studios, a renowned content strategist who shares his groundbreaking approach to using AI for content creation and optimization. Andy reveals how to train AI to understand your target audience, identify content gaps, and generate compelling ideas that drive engagement and conversions. Discover the power of having a conversation with AI, rather than simply commanding it to perform tasks. Andy emphasizes the importance of training AI on your audience's pain points, desires, and information needs, enabling it to provide highly targeted and valuable insights. Throughout the episode, Charles and Andy explore the various applications of AI in content marketing, from analyzing transcripts and identifying the most compelling soundbites to auditing e-commerce product pages and creating persona-driven content strategies. Andy shares his proven prompts and techniques, demonstrating how to leverage AI's capabilities to supercharge your content creation process. Gain valuable insights into the future of AI in marketing and learn how to stay ahead of the curve by experimenting, collaborating, and continuously refining your prompts. Andy stresses the importance of setting realistic expectations and understanding that AI is a powerful tool that still requires human input and guidance to achieve optimal results. Whether you're a small business owner looking to scale your content marketing efforts or a seasoned marketer seeking to harness the power of AI, this episode is packed with actionable tips and real-world examples that will help you take your content strategy to the next level. Key Takeaways: Discover a technique in AI that can help you create content that resonates deeply with your ideal customers, as if you could read their minds Uncover an approach to content creation that will make you feel like you have a genius ghostwriter at your fingertips Get a glimpse into the future of AI marketing and learn how to position yourself at the forefront of innovation by embracing new technologies and strategies Head over to https://provenpodcast.com/ to download your exclusive companion guide, designed to guide you step-by-step in implementing the strategies revealed in this episode. Key Points: 0:38 Secrets of Content Creation 2:49 The Magic of Gap Analysis 5:00 Teaching AI for Targeting 6:50 Emotional triggers in AI 8:45 Improving AI accuracy 13:51 Marketing emotions matter 15:22 Thought leadership creates tension 18:03 Triggering conversation for visibility 21:00 Ask questions, fill blanks 25:00 Learning from others 28:20 AI data insights 30:07 AI for marketing 32:05 Conversion optimization prompts 35:30 Training AI on audience 37:09 Collaborative experimentation

Welcome to the Proving Podcast, where it doesn't matter what you think, only what you could prove. Andy proves today that AI doesn't mean artificial intelligent. It means always incorrect or average information, unless you use very specific formulas and prompts, all of which he gives away. He reveals his systematic approach that delivers 10 to a thousand times better results. The show starts now. On this episode, we're going to have someone who absolutely kicks my eyes. There's just no way around this. Individuals who are light.

Hears more intelligent than I am. Andy is one of those people. Welcome to the show. I'm glad to be here. I love the topic. I'm a long, long time digital marketer, and I'm going to show you now, I mean, in the next few minutes here, just the best, most effective, most useful things I've done with AI. And, you know, just a bit from now you'll have all the prompts and all the methods right in front of. Before we go into, I break out all the prompts, I kind of let us share with the audience what I ran into. You sent over kind of your PowerPoint of what you teach off of. And it was a bit like being a caveman walking out of the cave and seeing a space show. It wasn't. seen fire. It was seen a space show. For so many of us, AI means, you know, artificial intelligence. And the reality is it's not artificial intelligence. It's average information. And what I've learned from you in a very short period of time is, that's amazing. I didn't train it in a specific way.

I don't know how to use prompts. It's like walking into Home Depot and saying, I'm going to buy a hammer. And that's all that good that Home Depot is for, because I can hang a picture versus I can build an entire community. And you've mastered that. And you're going to kind of show us exactly how to do that step by step. If there was one part of AI that was just the most magical and you would use it just for this, like this is the one thing that's good.

If it's email marketing or funnels or any of that, what's the one thing that's like, this. This is what changes the ballgame for all your clients and what you've done. Yeah, AI can do something that human brains really can't do well or in certain cases can't do it all, which is gap analysis.

Very, very difficult for a human to look at an asset, any marketing asset and say what's missing from that thing. AI could do that beautifully. But the trick is, and Charles we're about to jump in, you have to train up first on your audience. You have to teach it. It has to know what your audience's information needs are before you can write that prompt that says,

to what extent does this landing page meet or not meet the information needs of this audience? So I'm going to do persona-driven gap analysis on virtually anything, any landing page, any blog post. And then we can do it basically on the entire internet. It's amazing that it works. You can say, like, what's missing from the internet? What's not covered on the big blogs?

It's important to my audience. Super fun. I love that you're getting inch wide mile deep. There's so many of us like, oh, my grandmother makes these amazing cookies and my whole family likes them. Therefore, I'm going to start a company. You need to know exactly who you're selling to, exactly what the information is, what pain they're in.

And Nepsos are doing that with AI. So we're getting average information. We're not getting exceptionally targeted because we have no idea what the prompts are. We have no idea how to set it up. For those of you who are listening to this, you're going to need to go to the website because he's literally going to show us. So there's only so much I can say, this is what's going on.

It's kind of like when you're listening to a baseball game or the Super Bowl and the radio. Any of you have ever done that? Probably had it. You're missing out. So you're really cool songs, at least the chewers. So this is kind of exactly in detail how to do that. So before I take away too much of your time,

let's get into it. Rock and roll. Please, by all means, continue to kick my butt because it's brilliant. Okay. Here is our little thesis. And I'm going to start with a sample business. Right now I'm looking at

on my screen. I've got like a sample company. I pick this at random. We're going to be a spaceport for the next few minutes. We launch satellites and our target audience is commercial satellite operators. Very useful in marketing to know like the typical job title of the person who kind of signs the check or hires you or approves your proposals. So what I don't like by AI and I see these like just terribly written articles that say this is how to use it.

Write 10 headlines for articles about launching satellites. I'm calling it now. That's a lazy prompt. doesn't work. It's really boring. The responses are undifferentiated. As you said, we joke, like AI might as well stand for average information. It read the internet. It read the internet. It came back and it's going to be the average of the internet. It would be weird if that was somehow on target. It's not, uh, it's sort of like, it's face like water. You know what I mean? So what we're going to do instead of that is to create it, to teach it about our target audience. and if you have a documented ideal client profile,

if you have like battle tested, you know, sales-tested personas. As long as you have, I'm just going to use chat chbtee here. As long as you have a Chatt-T plus account, you can just upload that file and now it has it. And you can use that, right? That's, but if you don't have that, I've got a prompt for you that will help you do that. Here's my persona prompt, we'll call it.

Build me a persona of a job title, commercial satellite operator, at an industry and company size, billion-dollar global telecom, like a mid-market telecom. With the goal and responsibilities, they're responsible for launching satellites, planning to launch missions. And what they need, they need help deploying communication satellites. They're considering new large services providers.

Now, the next sentence is kind of important. List their hopes and dreams, their fears and concerns, their emotional triggers, and their decision criteria for hiring spaceport and satellite launch services companies. How they buy? what they care about. I want to end up here with a persona that lists their prioritized information needs.

A little bit different. What I love about this is, so many of us are using AI, and it's a bit like going and buying a pair of Air Jordans, and they're just expecting to walk out on a basketball for it and being able to duck. I mean, I'm six foot, and I have size 13 feet. It's an take an act of God for me to touch the rim,

but we have this expectation with AI that we're just going to go in, we're going to type in, write a perfect sales funnel for me, or give me the copy or do this. you haven't done the work. Again, we talk about this. You don't want to save the 10 minute. Don't worry about that.

Training to do this. And one of the things we talk about all the time is people don't buy products and services. They buy stories, identities, and ways out of pain. So you literally went in and the bottom says list their hopes, their dreams, their fears, their concerns, and the emotional triggers because they're not going to draw what it's actually happen.

You're teaching the actual AI to start thinking differently versus very detailed and averaging. If you want the average from the internet, you're going to get average results. And that's, as you've said, a million time. The goal is into 10. The goal is that your goal is the 100 exit. So again, just by this prompt alone,

which I'm curious, how did you learn these props? You just got bored or how'd you do it? Well, this is probably the eighth version of this prompt. I keep hammering on this. You just highlighted one of the things that I added most recently, which is the emotional triggers. My early versions of the prop didn't have this in it.

I've been teaching AI now for more than a year and so I can make presentations and feeling the pressure to present something. Just keep improving. keep adding to it. I think this will probably continue to evolve. But it's just trial and error. Combiding hopes and dreams, combining fears and concerns.

Basically, marketing is effective when you do things like answer questions, address objections. How on earth would it answer questions and address objections without knowing the audience? So the magic is just kind of having a persona up front first. When you put it in, weirdly, it's almost always named Alex. Chachapitia. So don't be surprised if your persona is named Alex,

you'll add that. So next thing, do not trust this. It will be wrong. It's inaccurate. It's obviously going to be inaccurate. It doesn't know you specifically yet, right? So you have to improve it.

To confirm that, I contacted one of my old friends and persona marketing expert, Ardeth Albi. And I did that in front of her. And I said, Ardith, rip this to shreds. I want to improve it.

And she said, Andy, what are you doing? Like, how do you know that's accurate? You didn't validate that code. Is that your person? Is it your competitor's person? Go check it with your customers. I don't trust this.

Artith, thank you. Huge love. I agree. I don't trust it either. Don't trust it. Don't trust AI. Trust yourself. You know your audience.

AI does not yet know your audience. So whatever's wrong with it, you have to fix it. Tell it to improve it. Add the following, right? In minutes of research, I found out that commercial settle that operators care about risk, insurance, right? Like that, okay, add the following decision criteria and I just put those things in. Okay, geopolitical considerations matter to global telecom companies.

Put that in, right? Give it those things. And then it adds that to the persona. Now, work hard on this because it's garbage in and garbage out. Again, I'm not really trying to become that much more efficient here. You'll see in the end I actually become, you know, you'll get superpowers following this process. But the goal is not here to save time. I'm going to spend as much time as it takes to make this thing good.

When I trust it or when it's good enough, you can ask it anything day or night. It's breathtakingly useful to have a synthetic version of your target audience that you can just talk to whenever you want. What information would help you do your job? What social posts do you like to click on? What queries do you type into Google? What research studies could I publish that would help you do your job better?

What do you hate? This is a good one, Charles. What do you hate about looking for commercial stuff for launch companies? We should all know what our audience doesn't like about looking for our services, right? they have pain even looking for us let's know that right

we like it it's working save it keep it keep it forever you can name that prompt on the left hand panel if you're in

chat chippy teap you could copy and paste it out into a PDF and share it with your team that they can use every time here's our

you know final approved uh official AI friendly persona right load this every time

yeah before you ask AI to write you a headline before you ask AI to evaluate your email copy or whatever you're doing, right? Because you can share it with your team. I think there's also a teens version of chat QPT where you can maybe, you know, that becomes a prompt that other people can use. Everything now that I do, I'm going to do in that conversation. I'm going to do it below. I'm not going to start a new one. I'm going to keep this one and have all of my subsequent prompts happen there because now this is the conversation that knows who I'm trying to reach. So you've made a filter that's based software persona. Everything that comes through it now

works. So for those of your phone we're trying to figure this out, it's kind of like, hey, I want to be phenomenal at sports. Well, what sports start? They're talking about tennis. They're talking about swimming, about a hockey player, because it's a huge difference between I can get the best hockey players in the world, giving the best pads and the best skates, put them out there and say, congratulations for swimming today.

They're going to lose. They just won't survive. So teaching AI to get it through this, you now have a persona that everything's going to filter through, and the fact that you can save it and share it with your team, I didn't know you could do that. It just kind of makes scaling so much again. Yeah, I love that metaphor.

I might steal that. It's single. It doesn't know. It doesn't know until you train it, right? It wouldn't be weird if it worked, right? If the data, if the outputs were very good, until you know this. Okay. The next step, this is basically, I'm taking what people do with one simple,

foolish, lazy prompt and breaking it up into several steps to get them a far better. a result. So next, I'm going to ask it about the persona's information needs. I'm going to do some ideation, right? Brainstorming. It's actually the most survey through this. This is the most common use for AI right now in marketing is to like, you know, brainstorm topics and ideas and just ideation stuff. So, but before I ask it for topics, I'm actually going to go one intermediate step first and just ask it about its information needs.

It's called chat. You're going to have a conversation with it. Just start talking to your persona. Here I'm going to start with the kind of a classic from start with a skill. You're an expert content strategist skilled in selecting topics that build awareness and trust with an audience. What information? Does this commercial satellite operator persona need to do their job well? And it comes back with some ideas for topics. Some of these may fit perfectly into my content strategy. Some of these may be don't at all. So I'm going to apply that filter again, right? Because I trust myself. I'm an experienced strategist. I've had a million calls with my

prospects. I know them. So I'm not going to, uh, I'm not just going to take what it gives me I'm going to get a list and then consider, and I'm going to pick a good one down here. Risk management and satellite launches. Assessing and managing risks associated with launches. Strategies for backup or recovery in case of failed launches. Maybe I know for sure. Bingo, that's it. Put my finger on it.

That's what my audience needs. Now I'm ready for my, the prompt. Suggest 10 articles on the topic. These will cast the ideas of Fisota. Yep, yep. Now we're finally at the point where you're going to ask it for topics. And I like these little words to add, you know, suggest topics, look after the interest,

they provide practical utility, make them compelling and memorable. Everyone's got their own favorite words. The words matter a lot. Those are words that I like. Words matter. They matter. Compelling and memorable, right?

I'm trying to save enough of mind. I'm not doing this for fun. So so hard, you didn't just say, hey, give me a bunch of articles. You created a persona. You then asked the persona, hey, what do you need? And then you narrowed it down because you already had the information from the answer that chat.

CPT gave you and it's like now create miracles that are most likely going to have a better engagement based on because you're doing it based off of if it's compelling, if it's memorable. Again, in the marketing, feelings matter, the emotions matter, what you're doing. Don't be, I tell people all the time, don't be vitamins, be ad lit them. If someone has a headache, don't offer them vitamin. Eliminate their pain and then enroll them in. You have to eliminate their pain first. That's, it's so important. So when you're talking about compelling and memorable, trust me, eliminate someone's pain, they will love you for a very, very long time. that's another good one.

So in that process, I took, I did the persona, I got the information needs, now I took one of those information needs and put it in here and asked for topics on that specific information need. And all the responses, assuming that this is what my company cares about and the value we provide in the world, did it quite well, right? On my screen here is a list of topics that might align perfectly because I kept filtering, I didn't trust it, I kept editing, I kept improving the answers it gave me, and narrowing down to that specific thing.

So these, in theory, are like, way more on target than you would have gotten if you'd just written that lazy prompt, you know, write me to the headlines. So weird idea. Start with your persona. Talk to it about your persona's information needs. And then for topics that align with those information needs. And obviously, you know, those will also align with your content strategy. You know, the things that you care about, the things where you have good examples or research or strong opinion. Which is another fun segue.

the strong opinion topic is one that it doesn't fit everyone's content strategy, but something tells me, Charles, I think you're going to like this one. I'm going to jump to, yeah, it's good, a content strategy where I'm going to identify the topics that are most likely

to be, have an emotional response from my audience. Seth Godin once said that her thought leadership creates tension. You're making assertions, you're willing to be wrong, and you can be certain that some people will disagree.

In other words, if people can't disagree with you, you're really not doing thought leadership. You might be a good content marketer doing purely educational content, which is mostly my strategy. But thought leadership is a different thing, right? You're actually making some assertions. Now, you don't have to stick your neck out and talk about social justice or, you know, geopolitics to be doing thought leadership. There are lots of relatively mundane topics that people care deeply about. I'll give you an example for my Settle that operator.

Here's the prompt. What are some relatively mundane, almost trivial space industry topics that professionals have very strong opinions about. The answers are actually charming. Units of measurement? I know doubt, right? There's barfights in NASA about this, right?

Naming conventions? What are people names, you know, spacecraft and celestial bodies? It's so interesting. Like, it quickly shows you the things... I mean, what a great outreach. output. So you can actually figure your conversation without kind of, you know, taking

on the world. I've been the JPL. I don't think those guys are having bar fights. They're adorable, but JPL's not having to do that over and that. These are brilliant individuals. They're not out on barfights. We'll talk about, you know, because talking about scaling and making this important that most of us aren't selling satellites.

How do you do this? How do you, how do you scale in an effective way? And the two examples I would always give is, when I'm a stage and I'm talking about, you know, who changes everything, I always tell people I'm like, why is bullshit. It just doesn't work. If you're trying to find your why, it's an ineffective thing.

It does not work for you. And there's a lot of people who think finding your why is the most important thing. That was an idea that was polarizing. The other thing I did was I sat down and I said, hey, at the time, I was a 42-year-old guy without chesticles. How am I going to grow 100,000 followers on Instagram in less than 60 days? So now of a sudden, I'm going to be outside of the normal one now that's going to get their attention. It doesn't. To your point, it doesn't have to be pliable.

I don't have to say, right, right, right. Right, right. That's right. You just have to not, you have to be the lighthouse amongst all the fog. You just have to stand out a little bit. Be able to signal a noise. And using the ability to say relatively money, almost trivial ways, but have very, have very strong opinions about that allows you to be that signal against the noise. And you don't have to always piss people off or you don't have to show a bunch of skin. That's not how. There's ways to do this that are practical, that are affected and in theory. So there's ways to do that. Yeah. And in the end, you can actually trigger kind of. conversation and conversation triggers algorithms. I mean, if you ever see something go bonkers on social media, it was probably checking

this exact box. It just people feel strongly about it. People start talking about it. People start reacting to it. It starts a conversation with comments and comments on comments. That's what the platforms want. So the examples in my career where I've done this, it's led to just breathtaking visibility, right?

I did a video and an article about why you should, you know, you should remove dates from your blog because it makes it kind of look older, faster. You know, if you publish evergreen content, you don't really need dates on there. And if you want to put a date and you can put it in the headline or the body of the article. Holy, you know, we have hundreds of pretty angry comments on YouTube. And YouTube decided, hey, world likes this one and showed it to hundreds of thousands of people. Why? Because that just was a bit more assertive.

A pretty mundane topic. You get the idea. The subsequent prompts are also similar, but definitely worth trying. These are magic. Totally worth experimenting with. What questions are people in my industry afraid to answer? What false things do people in this industry believe to be true?

We're plugging in the industry in each example. One of the most common assertions in this industry that are least likely to be supported with evidence. Hey, maybe I can produce a piece of research on that, right? That would be just catnip for journalists and trade publications. What are the most important, oh, here's the one that I love. This is almost my favorite, Charles. This is where I'm going to do gap analysis on the internet in my industry, for my topics.

What are the most important topics in this industry that are the least likely to be covered by the popular blogs? Literally, I don't think the human can do that. You can't read the whole internet and come back and say what topics are least that can be covered by the big blog in your category. And then this is also a really fun word to put a new prompt. What counter-narrative opinions on this topic are least likely to be discovered by bloggers and thought leaders? In other words, what assertions could you make that would literally be different from what the audience expects, that what others are not doing? I don't think there's anything except AI for which you can do persona-driven, industry-specific, gap analysis on the body of work for your whole category. worry. This is unique.

I also think what's unique about this is the way you're approaching it. You know, everyone asks like, what is GPT sample? And I just realized in this call, like, who cares about what with GPT staff? It's the chat. Have a conversation. Every single one of your questions starts with the same word, which is what? You're giving the conversation where most people, me, have just gone to AI and say, do this, do that, rewrite this.

And they're giving command. That's not a conversation. That's not chatting. You're starting this is actually viewing this as an entity. going, hey, you're going to help me out. We're going to have this conversation. You're going to be how. Please don't lock me outside and open the door.

But you're going to speak everything with this one. And you're going to ask these questions and have it fill in the blanks for you versus telling it what to do, which is a completely different way to use AI that I'd ever seen anyone do. I remember what I was reviewing this. I sat around. I kept circling what over and over and over. And I was like, oh, my God, he's having conversations with it. So these are, if you just did these, it would change things magically.

if nothing else, just try these prompts. I think that the fact that we're all sort of accustomed to the U.S. of search and we query things, we see a little box and we type it a query, but this is completely different. It's a different paradigm. Probs and queries, not at all the same thing. In Chhapit, you're really going deeper, right? It gives you a thing, it's not right.

You can, you tell it to correct it. No. Or you ask it how it got there. Or you tell it to show its thinking. You know, you can, continue, it's not, it's hard to teach in a way to make slides for it or talk about it at a conference because, uh, the proper way to do it is a lot of back and forth. And, and then once you get it, and then interesting, here's a pro tip. It's not in this deck, but, uh, if you get into where it

gives you something useful and you want to go repeat that method again later, you can do a final prompt that says, write a prompt that will give me a similar response in the future, uh, that I could use to quickly get to this, you know, have you do this analysis. And it will write the prompt to have you repeat that entire exercise you did. And then you saved that as you prompt in your shared prompt library. A lot of people like us now have shared prompt libraries, Google folder or someplace where everyone can go grab the best prompts. So don't delete the thing.

Don't quit. Don't leave it until you've gotten the value from it in a way that makes that analysis or that method repeatable. This is Sulkis question. Is there a place where all the best prompts are stored from like all the companies where everybody comes together and say, hey, you know, kind of, you know, work with the geeks we would get together like, hey, this is the best thing we did here, the best for forage trading or for this, hey, we figured out this thing.

There was a knowledge base where we as IT guys used to go and share all of our stuff in these knowledge bases and then those ended up being monetized in a very interesting way. Is there one that has kind of the best, most proven tried and true prompts that are working or is that that doesn't exist? I know of some private Slack groups for AI practitioners who share some things in there. But Charles, not only is there not like a public place for this, but companies are not asking new employees to sign employment agreements that say, you will not share any prompts that you learn to create on this job.

Prompts the outputs of AI are not copyrightable because there's no significant human involvement. But the prompts themselves are copyrightable. And I think that there's people who write these prompts, they're almost squeamish about sharing them. There's a joke now. They say like English is the hot new programming language. these prompts are basically doing what expensive software would do. Two years ago, for only 200 bucks a month,

we've got a tool that will analyze the internet and tell you where there's content gaps across your entire category at 200 bucks. Here's my money. That's only one of the million things I could do in here now. So I think that there's, it's so powerful, there aren't that many people teaching it in this way.

People are starting to hold back. Weird. I'm just saying as, you know, because as entrepreneurs and I love scaling and the worst thing I would ever tell anyone to do would be to start a business

that's like try and get people to never do that. Don't start. The amount of friction to actually get launched is a nightmare, but the entrepreneur in me is like,

I'm going to make a vault and charge people 70 to 100 bucks to love to get into this. Keep going. No, no idea that pops in mind. Cursting in scaling.

Yep, yep. Right now, I think the results are so bad that if someone said like, hey, here's 500 prompts for $20, it's probably garbage. And honestly, you know,

prompt visionary is probably overrated and AI can help you write prompts itself. But this little framework, right, I think we'll get you a lot of where you want to go. Also, I learned a lot of what I learned about AI by experimentation, but also by contacting people I knew that we're using it and doing little knowledge swaps. Like, can we talk for 10 minutes once a month

and I'll show you what I'm doing and you show me what you're doing? I learned a lot very quickly by talking to people who I knew were using it a lot. like, yeah, that guy uses it to audit landing pages and here's his prompt that has all the best practices listed at the top and it comes back and tells you like all the ways in which that landing page does not align with his best practices. Dude, so grateful.

Let me show you what I got. And just kind of back and forth really helped, you know, took months off my learning curve. Again, maybe on what you know, it's who you know. Yeah, it helps a lot. I mean, just the former mastermind group are like, get on a month like, this is four people. I have, I mean, two mastermind groups, one AI, one agency owners. And there's four other people separate where I do a 15-minute calls once a month.

And we learn tons from each other. I'm going to come back in and do another prompt, to make the most of this time. This is something that is very, very hard to do as a human, which is to take a transcript. I'm looking at a YouTube video. To take a transcript and turn it into a piece to reformat it as another piece of content. So if you go to YouTube and you expand the description and you scroll down, to the bottom. There's a little button that says show transcript, then you copy and paste that transcript out into a text file. It'll be way too long.

Fairfifty you will choke on it unless you upload it as an external file. Upload that as a text file. And then you can say, you know, suggest article titles based on that or write a summary. Summurization is one of AI superpowers. Very, very helpful. What it just did there is actually a huge productivity game because it's super tedious to try to turn it. long, you know, these like super long monologues and giant blocks of text into anything useful. But the next prompt is even more fun. Listen, this is similar to the gap analysis, but it's going to find for me the juiciest nuggets, right? Those 10x or 100x little quotes, they just flowed past.

You've dropped several here, Charles. If we did this on this conversation, I know exactly what it would suggest. Very difficult for a human to pull out the most compelling sound bites. Or the counter-narrative opinions. or the surprising facts, right, or the most emotionally triggering, you know, bits of language from a long transcript.

How do you do that, right? You're going to read the whole thing and, like, try to say objective the whole time? So here's the problem. Which statements in this video are people most likely going to be surprised by? Which soundbite in this video would make the best clip,

the best video clip for social media? And it will come, and basically the response is sort of amazing. It's like, okay, here are the top three surprising statements. These ones were on lines 44 through 4th 48, right? Because you've got the transcript. These are potential sound bites for social media. Now, just go back to your video and just chop out that bit of that video. That's your, that's the, you know, throw gas in the fire of social media content, right?

That's the, that's the, so really fast, really effective, a smart way to do it, aligns perfectly with strategy. All the good content marketing strategies, right? Repurposing, upgrading formats, you know, uh, an X promotion, So that's one where, again, you're using, and this one, by the way, you didn't need the persona because you could do it in that same conversation, but mostly this is an example of another use for AI, more or less, where you're giving it data and asking it for insights. Similar to this, I upload a lot of GA4 reports. If I could show, it's kind of fun. It's very nerdy, but if you give it a GA4 report showing all of your content, yeah, you want me to jump into that one? I was going to say, yeah, I'm curious two things. One is when you're like, what are the best social media?

It's like maybe would I prompt it on chatting with it? Would I be able to ask it, hey, what is it better? Should I use the prompt best or should I use the most viral? You know, so when you go into that, when you play with those, is it just trial and error? Because with the podcast, with what we're doing here, I can literally take the entire, you know, 40, 50 minutes. Because some of these are longer than not. As we're doing this, how then just do this? And it cuts all of my time for my team because week later, we a lab report every one of these where we give people, you know, here are the prompts, here are the steps. Go do this. Here's the homework. Go get this done. This would help the team out creating those lab reports are, like 50, 60 pages. This will help me take out the information, expand on it. Now we're taking like 200 page lab reports for people that so they can have that level of value. This alone changes the business because there's so many people that need to write link out of orders and they need to get information out there where they're like, oh, well,

fired, I don't really feel like writing. I'm guessing even if you transcribed, because there's both of transcripts tools, your latest meeting or your meeting with a client. And you did that. You could run this through there. And then give an email back to them saying, hey, this is a recap of what happened. This is how we did it. These are the most important things. These are your next step.

And it's a immense amount of time. So there's a solely different ways to use this. Yep. Yep. The AI meeting assistants are mostly doing that part, the summarization, which is mostly what I'm doing here. The slightly different part My twist on that here is that I'm using it for marketing So I'm trying to get it to find me something that's provocative

Basically, right? I'm doing social media marketing at the moment So I'm trying to find something provocative There are, I don't know if the meeting assistants really do that as well You can certainly get the transcript out of there They put a little summary at the top usually They're good and they become pretty standard But this is, you might like this

Looking back I now wish I had The last 50 or 100 sales falls I'd ever done recorded because I could put all those into transcripts, and then I could give those transcripts to the AI and say, what are the top questions people ask me during sales calls? What are the aha moments that clients have when I do sales calls? You could actually, if you had all the transcripts from all of your last sales calls,

so start recording these days. In the future, yeah, you're going to be able to query, like basically talk to the AI about your last 100 sales calls and figure out what was the thing that, they always asked, what was your best answer to that top question? What are the questions where you didn't have a great answer and what could be a better answer for that? The value of that, I mean, probably as more greater than anything else we've discussed so far. I just, I don't have all that data yet, but going forward, yeah, we should be saving more of our

language, chunks of language so that you've got a big enough body of, of, short language, you can start getting insights from it. I'll give you another example of like, you know, are these the words that I use? Is this like, what are the best words to use in these? Are they specific to, you know, are these just my techniques or do I need to adapt those for myself? Big picture. I'll show you what one of my friends who, Justin Rondo, conversion optimization master, is like done literally thousands of baby testing his career.

Former head of growth for digital marketer.com. Like, badass. Super humble. So, so I got to give a presentation. an e-commerce conference. And I suggest that I want to try something. Can you send me your best practices for product detail pages? Like, oh, yeah, I got a deck right here.

Sends me his deck on how to make the perfect product detail page. It's 148 pages. This is not a prompt. But at 90 minutes later, sitting there right before breakfast and my kids wake up, I went through the whole thing and turned his 148 page deck into a for auditing e-commerce product detail pages. This is basically everything from his deck. And it's on screen right here.

This is the prompt. It's got everything. You're a conversion optimization expert. You're still in maximizing sales from e-commerce websites. You can evaluate your product detail pages against trust and conversion. Here's a 15-point checklist. The checklist is broken down. So what the prompt is is basically best practices for, from my perspective, you know, or from

the experts. and then at the bottom it says I'm giving you a screenshot of a product detail page Provide a detailed audit screenshots you can take out of I'm using like a Chrome extension as a bunch of them to scrape a page

and then give it the image of that page and then say provide a detailed audit and then rate the extent to which this page meets or does not meet the criteria on this checklist on a scale of zero to five visualize it on a colorful matrix, which it does. Or I'm going to give it

product information and say create a draft of a product of an e-commerce product page for the product. So basically, you take... Some of these great prompts are combinations of your best practices that you convert into a prompt for auditing an existing thing and finding gaps, making

recommendations, or for creating a new thing and creating a draft, which you then obviously have to edit because you don't trust AI, you've got to, you know, use your own brain. So that that approach, you know, these pumps look, I show these to people are like, wow, that's a crazy detailed prompt. Why don't you write a shorter pump?

Because I need to teach it what quality looks like before it will come back and, you know, meet my standards. I might be different from other people. So that idea is, uh, just so maybe... At the time that we've had so far, we've already identified that you're different than most people. Let's get that out of the way really quick. So I know we're really close on time and, you know,

you don't have a ton of time here because you've got a hard stop coming up. When you're going through here, and I would love to give all this information out and steal hours of your day to do this. As you go through this, what are some of the things that if you're at home right now, if you're sitting down and you're a relatively small business owner, you're making, you know, you're,

you cut to a thing and you've made your seven feet and you're stuck and you're trying to scale that next one. You're trying to figure this out. What are some of the tools? What are some of the resources? Because you've got to try and be respectful over your time that you would say, hey, go do this. Or, hey, do you?

this. I know you've got to, you know, you're redesigning your book and that's going to be coming out and the podcast will be out beforehand. And when your book comes out, we'll just bring you back and we'll have more time to do this. What are some of the things that, if you want, what are some of the things that you just woke up going, oh my God, this is chat. I'm actually having a conversation. Oden the prompts that you've given and changed some of that. And we'll put us all in a loud report. Some of the things you would do, okay, this is what you need to do right now. This is vitally. Yeah. Pay your expectations close to zero. for quality outputs until you have trained AI on your audience or your best practices.

There's, we do not live in a world where you can type 10 words and it does your job. I don't actually want to live in that world anyway. That's when people start to freak out about like labor market impact. AI still needs, you know, it's a tool like many other tools and it's a really powerful, useful thing if you use it in a way that is strategic. And we're going to begin with the added mind. As you said, and Charles said that one more time. I love that.

People don't buy products and services. They buy stories, identities, and ways out of pain. So how could you possibly to sell that unless you know this person, right? What's their pain? What do they care about? What's the true story? We sometimes say there's a true story in the life of every visitor to every web page. So when you know that, when you know their context, now you can help them.

Before that, I saw it. It's almost a joke. but there's a tool out there that will build you a website from a prompt the entire site. It makes images or pick stock photos and writes all the headlines. They just type in like, you know, we are marketing automation company. And it makes a thing. And it's garbage. Why is it garbage?

Because it doesn't know your audience. That's why we don't have the same website. It doesn't know their pain. So step one, brain it on the audience. Or if you're doing analysis, trade it out of your best practices. Once that's in place, keep experiment. and keep trying different things. And when you find something that works,

save it, copy that prompt. It is like code. Our shared prompt library is called the codex. It's like, it is like programming language, right? I like that. Keep it, share it with your team. Do little lunche learns or make sure everyone in your company knows that these are, I mean,

I've got 55 employees who have wildly different skill sets from programming to project management to, you know, to sales. everyone has different use cases and AI has infinite use cases but if you have a way to collaborate and keep experimenting, keep trying new things and keep teaching each other

and maybe you're a solo pro great, you know, jump on calls people like me or Charles or anyone, right? Like, hey, do you have 10 minutes or do, you know, can we do a lovely 30 minute call to trade all of our best notes? The people who do this now, right? This is early.

2024, early days. We just got up and had breakfast right? What happens in the next five years? Well, you're going to have a far better position for it. Yeah. There's a, have you ever, Amara's law? Roy Amara was a computer scientist at Stanford like in the 70s. And he said, people overestimate the impact of technology in the short run, and they underestimate the impact in the long realm. Vastly. And I think people also underestimate understanding the persona. People walk away, they're like, oh, this is the greatest new tool. This is the addition to this. And we talk to people, and again, if you live in the United States,

it makes it really easy for you. If Saul has a pickup truck, who do they vote for? You know the answer? If someone has a previous, what do you vote? You're not going to market the pickup truck guy, the same end of art to the Prius. It's a completely different thing. But if you're going into chat GPT, you're going to AI going, give me a website or a funnel or an email or whatever it is without preening it on this and having those conversations, you know, you might get luck, but it's kind of like fishing with dynamite. So it's just not going to work well for you in any way, shape, or more.

All right, before you take off and disappear, I know you get off here, how do people get a hold of you? Where do they get more resources? I know you're redoing the books. I can't plug your book. Where do people find you? How do they get more information about it? Where do they go?

Yeah, I've read a book about content strategy. It's in its sixth edition, and it's got tons in it. But it doesn't have the prompts because the last time I rewrote it, it was just before the chat, GPT, blow up. So that's in the process of being rewritten now. So don't buy the book if you're looking for AI tips, but still useful for all-purpose content strategists. My best one from the book is LinkedIn.

I read an article on my website every two weeks. That's orbitmedia.com. But you find me anywhere. Ask me anything. And if anyone wants to really connect with me on LinkedIn, what do they call it? It's like that blue button is set to follow. There's also a three-dot button there that you can click to connect with me

if you want to drop me a line and like, you know, let me a DM. So, I'm going to go over a lot of noise on social media. If I was about to go back to chat, you can say, what's the best way to connect with Andy? And they have it a most effective way that will get his attention that he responds to the most. And they get it and just keep going through that

before you even reach out to you. It's kind of like how I'm thinking. Well, um, that one here you got on. You can take a LinkedIn profile. Expand all the sections on a LinkedIn profile. Take a full-page screenshot of it. Upload at the chat GPT. influencer outreach is something that you can use it for.

Something else, cool. One last thing, we can't resist. If you're in transition and looking for a new gig, or hiring someone, let's say you're hiring, you're in growth mode, you can take the job description and their LinkedIn profiler resume, upload both of those to chat GPT and say, do a swap analysis for this person in this role. It will, it's a fantastic recruiting assistant because it will list for you like the possible risks of hiring. It's going to be wrong. You've got to check it. It's just a point of view.

AI should say it for another input, right? This is another input. But you can actually give it a LinkedIn profile and a job description and start to talk to it about that person in that job, assuming the LinkedIn profile is accurate. You might get some insights. Yeah, I think, as you first came into this, you know, we've talked about AI meaning all these things, but the way we're doing it, it's AI stand for always incorrect because you haven't trained it in a way. By default, it's going to be always incorrect. So, you need the person. teach us those things and then go then kind of go from there i can get more for you so much thank you thank you so much for being on this one we will definitely have to have you back and i'll talk to

very much super fun thanks charles thanks everyone and you just proved that ai without a strategy is expensive guessing stop asking for miracles start building systems create your persona teach the machine iterate relentlessly company's mastering this now will dominate tomorrow while everyone else drowns and lazy prompts