GOD Prompt Protocol for ChatGPT: From Overwhelm to 6x More Effective Research & SaaS Marketing #009

Viktor: 0:16

Welcome to Disruptor Digest, the top disruption business show. We dig up the secret playbooks used by first movers, featuring the latest tools, technologies, and science, ensuring you won't fall behind or succumb to FOMO. To singularity and

Mihaly: 0:30

beyond. Hi Disruptors! Hi Viktor! Hello everyone! Today we are gonna talk about prompt engineering and prompting. And when we started to plan this episode, I wanted to give you 10 prompts to 100 prompts, and on internet I found more than 700 and 1000 prompts, and I thought it might be overwhelming. So let's go and... Just do it. The 10 most important, the few critical, but after that you had some, an even better idea. Can you tell us what was that? In

Viktor: 1:01

the last 15 years I'm working in marketing and even though I'm Working in marketing in 18, 21 different industries, lots of languages. I don't always have the perfect and exact keywords I need to use to get the perfect results I'm looking for. So I'll give you an example. If you ask ChatGPT or any kind of large language model to write an email for you. It's not. Going to produce the same results as if you write, give me a cold outreach email. So if you need a cold outreach email, you get better results. And that's actually happened with me when I tried to make social media titles and posts from the top of my mind, I should use curiosity gap. And that's prompted me. Okay. If I want to get results in any kind of field, we need a god prompt. So because ChatGPT and these large language models know everything, or mostly every, everything which humanity came up with so far. So we can use it to actually mine and understand first. What I should include in the prompt and then actually executing the prompt itself. So that's like the base structure of the god prompt to teach you fish instead of just giving you the fish that first you define what you want to do and then. You tell the model to do an exhaustive research on a topic to which kind of mental models and which kind of terminology you need to use to get the best results. And then you say, okay, please execute it and give me the results and execute the task itself. So that's Gist of it. And that's like the high level overview of the God prompt. Yeah. All

Mihaly: 2:39

So today we are going to give you the ultimate prompt, the one prompt, the rule them all and it's version 3. 0, which means we have worked on this for a while and also it's work in progress. Yeah, but let's get to it. But before that, just a few sentence about our podcast development. So Viktor, in the past week, you have been setting up the, backend when somebody goes to disruptordigest.com. Can you tell us about that?

Viktor: 3:07

Yeah, sure. So we're using to send out emails, we use beehive and we set up cloudflare email worker. So if someone is writing an email, we can write actually easy JavaScript code to intercept the emails and see what is the content of it. And what it allows us is. If somebody is signing up and sending us an email, we can automate the process of sending them the god prompt with examples. So if you go to disruptordigest. com, you sign up with an email, then if you go through the flow at the end, you get the result of the god prompt with read examples and it should work already.

Mihaly: 3:47

And also we started to get our first feedbacks. For example, one of our feedback came from one of my best friends, Miro. And he said that the what and why is very good, but also we need to more talk about the how. So actually examples. And we totally agree with that. And we want to really give you a Practical protocols. So this is our aim with our podcast. Can we move to the god

Viktor: 4:14

prompt? All right. Yes. So as I said, it consists of different steps and seven steps to be exact. Start with the goal. The goal you are specifically say the goal semicolon is get a superhuman word class level result in. And you can fill in the blank. And in my case, it was finding a name for a new product. I try to launch that is helping people to get from not being sure what makes them happy to exactly know what type of work is good match for them and what are the top three explicit values we should be met for our work to be fulfilling, like a tool for business model you. So that was the first segment of the prompt. I just. Prompted it, okay, what is my goal? And basically I tried to come up with a name. So the fictional

Mihaly: 5:00

example is here that you are developing an application, which is helping people to find their work or life goals, right? And you want to create a landing page for this,

Viktor: 5:10

right? Yeah. So it's semi fictional because back in the days, like six years ago, we spent half a year on finding our purpose and it was a lot of work and we came up with a good. Workflow, but we never productize it. So it's yeah, at this stage is semi fictional because maybe we're going to launch it because it actually helped us from getting, we are not really sure what we want to get done in life or what, makes us happy. To the point where we were crystal clear and having the top three values for us. For example, for me, it's doing things on a scale and also doing things which are practical and getting results and actually learning from working with smart people. So that's like my calling and. This whole process helped me and we never productize it and just coming up with a name, coming up with the framing and coming up with a target audience and coming up with a landing page. It's like daunting because this is like a passion project maximum. So this is the example for it. So that's like the first half. And shall I share my screen for this? Sure. So yeah this is the prompt. The goal, as I said, was as I was reading before, and then I prompted ChatGPT to follow these steps. It's consists of seven steps and actually has a zero step, which is always starts to answer with this prompt and always indicate which step are we at so we can move forward or backward. And that's basically just. Being able to navigate these steps and if we are stuck, we can move back or we can move forward if I want to, that's like for having a meaningful back and forth conversation with ChatGPT. Okay,

Mihaly: 6:50

so for me, it looks a little bit like a code that you define logic and you define successive steps,

Viktor: 6:55

right? Yeah, sure. So it's like defining, playbook and having someone who is extremely knowledgeable to walk you through the steps and walk you through the path. So yes, exactly. It's like a code and executing a code with having a back and forth with a, with an extremely smart person. So yes, that's what the zero step. And the first step is clarification steps. So if. Clarification step. If my goal is not clear enough for you to move forward, ask me one question at time and wait for my response and repeat it until you feel confident that you can move forward. So this step is about making it sure that you provided. The end goal clearly, and we're going to see it, that it's actually helpful because sometimes we think it's quite clear what we want to achieve. And then just be realized, Oh, it wasn't that clear at all. So then the second step is mental model research, do an exhaustive research on the most important mental models and terminology that is needed to get word class results. Include a one sentence, short summary and example and limitation for each. So this is a quite complex step because we are asking. To do an exhaustive research like this word exhaustive. I use it a lot because otherwise it's just quite arbitrary. If you ask for five mental models or five terminology, but if you ask for exhaustive and you say that you want to get the best results, then it's going to collect all the most important models. And also I'm asking for one sentence, short summary example, a limitation as well, because obviously. Each mental model has a limitation. And then, the third step is inversion research. Now do an exhaustive research on inversion using similar structure. And in parentheses, summary example, limitation, meaning what we should avoid. So that's. Again, doing like we define what is helping us to, get to the end state. And also we've tried to define, okay, what should we avoid? And that's is a good framing for ChatGPT to, solve our problem.

Mihaly: 9:06

It's very similar if you would work with a person, right? With

Viktor: 9:09

a coworker. Yeah, exactly. That's how you prompt ChatGPT is quite similar how you manage people. So if you. Properly define okay, what is the end goal? Are you properly define what should be avoided that helps? That would be helpful for me as well. That's like how we actually managed people in our marketing agency. So the fourth step is book research to do an exhaustive list on the most important and practical books on the topic and their main takeaways. So that's, again, it's like Spending the research to books and why is it good? Because it's also like a coaching step for you. So maybe you are not aware what kind of books could help you, but it's collecting them for that for you. So either way you are exploring them to the chat interface, or you can go and read them, or you can go and find summaries for them. And so on but that's like also collecting all the information.

Mihaly: 10:07

And also, as far as I know, GPT was trained on a lot of books, but also there are a lot of summaries about almost all the books, all the important books, such as GPT know the essence.

Viktor: 10:23

Sure, So it's since it's containing the common troll, so like the text content of websites and Reddit and those kinds of sources, since people are posting about books. It knows about the essence and it knows what it is about. Yes, sure. These were the research steps. And then there's a post research clarification step as well. If you think you need further clarification from me, ask me one question at a time and wait for the answer and repeat it until you have all the information to move forward. So this is similar to the. Beginning clarification steps. So maybe from the research itself, some additional questions came up, which are preventing the model to know how to move forward exactly. Why does it work? It works because how these large language models work behind the scene, they use the transformer model and the transformer model, you provide a text. In this case, this prompt, and it starts to generate one word at a time. One token at a time, but essentially it's just one word at a time, but as it's generating the words. It, they become part of the original context as well. So what does it mean if it is like starting to do the research first, it is the same as I already provided the research for it. So if I get the original prompt, I provide the research myself and I say, okay, now solve the problem. It's like the same technically. If I just provide the prompt itself, let it do the research. It starts to become the part of the context itself. And as soon as it starts to work on the problem, original problem itself, it already has all the information, the prompt, all the research, the clarification steps and everything in the context itself. So that's like something what goes in behind the curtain. So it maybe may helps you to understand why does it work. And obviously the last. Step is suggestion, use the research to suggest the most effective, exhaustive ways to solve the original goal. So that's like a blanket description of prompting the model itself to come up with a solution.

Mihaly: 12:30

Let's see an example, how ChatGPT would reply to these

Viktor: 12:34

prompts. Sure. So let's go through these. In this case, I was prompting, as I said before, to come up with a name for this tool, to find your calling and your main values in life. And it's, the clarification step is quite good because it's, it asks me, okay, to make sure I understand your request correctly, you are looking for a name for a product you are launching, a tool that helps people understand what type of work makes them happy, and identifies their top three values for fulfilling work. Is this tool a book, a software, an online course, a coaching program, or something else? Is it targeted towards a specific age group, professionals in certain fields, Or any other specific demographics.

Mihaly: 13:12

ChatGPT started to clarify about what is the actual thing that you define. And it was very great, right? So it got the essence what you want to build here.

Viktor: 13:22

Yeah sure, So it summarized how it works. And let me diverge for two minutes here. That's like the secret behind ChatGPT. Not too many people notices it. But how ChatGPT works is usually they summarizes what you are trying to do in the first paragraph. And then providing you the answer. And then you clarify okay, that's not really what I meant. In the next first paragraph, it's changing the goal or changing the summary of what should be done to reflect what it understood from you. And then answering and why is it good? The first reason why it works is because you, it's very explicit. Of what the model is understanding from your request. So it's a good feedback to you, but because maybe you didn't, you weren't clear at all. So that's one. And the second thing is since it has a context window, so it has in GPT force case, it's 12 pages of text, which can fit in the context window, maybe some back and forth. Is missing or getting out of the context window, but the previous message always can fit into the context window. And since the previous message, first paragraph is summarizing the current task, that's how it can keep consistent for a long time, even beyond the 12th page, you can keep on discussing and stay coherent basically. Okay, so the

Mihaly: 14:44

next thing that GPT asked back is, it a, tool, a book, software, online course, right? And you, mentioned that you haven't defined yet, but it's very good to clarify.

Viktor: 14:55

What you realize with these language models is you are not clear enough. So when you ask for something, you are not clear. It's easy to misunderstand and it's actually teaching us to be More specific and one of the target audience who is using ChatGPT to its fullest is project managers, because project managers and product owners, their task is to prompt, to write prompts to actual engineers, to write prompt to business people. So that's their day to day job to be extremely clear and extremely specific about what they try to achieve. So if you are not yet got into. large language models, you're missing out a lot because this is something worth considering to pivoting towards.

Mihaly: 15:41

And the last thing that chatGPT wanted to clarify, who is the target audience? And it's, I think it's also very important because if you write an email and you think about who is it targeted to, who do you want to write it to, it becomes easier to write a great email or write a great Google ad and in this case write a good landing page, right? Absolutely

Viktor: 16:00

right. So yeah, and then I just answer that online software that helps anyone, but mostly relevant. Those who are not completely fulfilled at their work, get people from the state of, I'm not completely sure to, I'm crystal clear of what are the top three things that are needed for me to be happy. So that's what's my answer. And that it proceeded to do a mental model research. It's step two. So mental model research, understanding that the product is online software aimed at individuals seeking fulfillment in their work. I now explore relevant mental models that may be useful in finding a compelling name for it. So that's again, as we discussed, starting with a summary of what should be done now. And it's collecting basically the Pareto principle, Maslow's hierarchy of needs, job characteristic model, SWOT analysis and Ikigai. And it's including limitations. So for example, in Ikigai, it's explaining that the Japanese concept that's combines the four elements of what you love, what the world needs, what you can be paid for, and what you are good at. An example, an individual may find fulfillment where these elements intersect and limitation is that this model may oversimplify complex real world scenarios and personal motivations. That's what's the answer for collecting the different mental model and then I just write continue and then we are continuing to step three, which is the inversion research. Okay, what is it? In this case, it also states in the first paragraph, inversion is a mental model where one thinks of a solution in reverse. In this case, let's consider what we need to avoid. Choosing a name for this product and it's collecting vagueness, common names or cliches, how to pronounce or spell culturally insensitive or offense and limited to current state. And in this case, just limited to current state is the name should be flexible enough to encompass potential future growth or. Change in direction example, if you name it, job matter, it may be restrictive. If you expand to offer a broader career guidance, a limitation, a too generic name may not communicate the product's current value proposition effectively. So what it basically says that if you are too focused, and I can give you an example, like in Estonia, toxify, which is a. Uber competitors started out basically competing with taxis, and then they realized they want to get into mobility and different modes of mobility, e scooters, car sharing, and so on so that's why they changed their name to Bolt. The original naming is, was too too narrow and too restrictive. And that's like the. Trade off between being too generic, which is not talking to anyone because you just say, okay, we are doing something just for matching jobs. But if you want to try to move to coaching and going down that route, then it's too restrictive. So being too generic versus being too specific, a balance have to be found. And that's why it's extremely good because it's not just. Collecting. Okay. What mental models and terminology we should be aware of, but also it's avoid it's put on the map. What should be aware of? What should we try to

Mihaly: 19:15

avoid? I really like these five mental models that GPT came up with to avoid like cliche

Viktor: 19:22

and Bigness, hard to pronounce. Yeah, it's,

Mihaly: 19:26

it's very good to keep them in mind. It's easy to forget them because they're so simple. And so Yeah, of course, I think about that, but no, at the end of the day, a lot of brand names are cliches and too vague, right?

Viktor: 19:38

Yeah, sure. So if you come up with a new brand name... It's easier to have like a checklist, similar to how you manage people. If you get a task to come up with a name versus you get a task to come up with a name, but avoid these five steps, it's much easier to get the five steps as a checklist, because you can basically go through them and see whether a new name is checking all the boxes. So once again, you should, and you can. From these models, quite similarly, how you manage people and vice versa to talking to these models, also helping you to manage people better because you learn to be more explicit. You learn to provide not just examples, but counterexamples as well. The thing

Mihaly: 20:19

that I like about this approach that it has GPT provides the summary, what is it? And also examples so I can understand and it would be easier for me to judge the end product because I will know what to focus on.

Viktor: 20:35

Right? Absolutely. Yes. And also we're going to see that it's not just coming up with a name because yeah, you can just prompt ChatGPT to come up with a name and it can do it quite well because this is, this got prompt is actually a process. Where it really shines, it's when the end product becomes quite complex. So the end product is not just the name, but from the name, it's also like a landing page, target audience, Facebook ads, Google ads, and so on and this is exactly what I'm showing you here. So we are going through this exact example, but. Once again, you are going to share Mihaly also some examples with the dear listeners. Let's move to step four. Yeah, sure. So I'm just writing continue and it's replying step four, book research. And it's once, again, it's summarizing the task first. So here's a list of impactful books on the topic of job satisfaction, fulfillment and finding one's purpose and their main takeaways. It's just collecting Drive by Daniel H. Pink and it says this book argues that, that autonomy, mastery and purpose are the key elements for true motivation in work. It's, once again, collecting Man's Search for Meaning by Viktor Frankl. What color is your parachute, a practical manual for job hunters and career changers by Richard Bowles. And so good. They can't ignore you. The element how finding your passion changes everything. Ikigai, the Japanese secret to a long and happy life, and also business model, you, a one page method for reinventing your career. So it's basically collecting the relevant books and it's once again, prompting. The model itself to, to think about the end goal with these books in mind, it's helpful for you as well, because if you try to dig deep, maybe you get some inspiration to where you can learn more about reaching your, end goal, which you have in mind. So step five is post research clarification. Basically, once again, summarizing the first paragraph, what has to be done. And it says, before moving on to suggesting a name for your product, could you please provide a little more details on the following aspects? And it has two questions. First, what is the unique selling proposition of your software? How is it different from existing tools and services in the market? And second, is there any key philosophy or principle That underpins that way your tool works or the approach it takes to help users. And understanding these factors can help generate a name that accurately and compellingly represents your product. This is quite intriguing, right? Because I never thought of what could be my selling proposition. That makes sense, right? It's it's, like a good question. And also the second question is something I couldn't come up with because he has USP something, which is top of my mind. And I could have come up with that. But the second question of having a key philosophy or principle that underpins the way my tool works, that's like intriguing, right? I would have never. Guess that, okay, that's something I can provide the consultant like GPT, right? So my answer was USP is basically it works with a smiley face. Really we use the tool to get unstuck. Most other tools are not really practical. This on the other hand is extremely throughout, structured, goes through everything to dig deep into who you are and what makes you happy and what doesn't make you happy. And it's not just extremely simple step by step, but also ask for feedback from those that know you the best, for example, friends, family, and coworkers. So that was my answer to it. Step six, basically suggesting again. And it's once again, summarizing that based on your USP and the research we've done here, a few suggestions for your product name, TruePod, Deep Dive, Fulfillment, Virtual Navigator, Clarity Compass, You, Revealed, Inner Compass. And it's like. Explaining carry a clarity compass that suggests the software as a tool guiding towards clarity about one's career and direction and values. So it's like summarizing each name suggestion and also giving me basically six different

Mihaly: 24:51

answers. There are six answers and I think two or three I really like. For example, InnerCompass and TruePass. TruePass is very short and, it's very clear. So yeah, there, I think it's better names that I could come up with even thinking about it for one or

Viktor: 25:09

two hours. And the thing is, it's like this whole naming, what we are talking about, and we just did talk through in the last 15 minutes, is something. Which someone can charge you starting from 15, 000 to maybe hundreds of thousands of dollars. So this whole process of naming is quite painful and this is quite excruciating to go through yourself alone. And this god prompt is just rubbing your hand and walking through you, this whole process. But the beauty is that actually you can use the god prompt to, to use. To solve any kind of problem and we're going to see it, you're going to provide more examples. So not just this and where it's really going to shine is the next step. So I'm just writing now, using the research you've done, give me an exhaustive list of potential target audience. And after that, write a short testimonial for each, but don't mention the specific product name. Just

Mihaly: 26:01

to clarify, this is the end of the original GAT prompt. And now you want to use the research and the answers of ChatGPT. To develop something

Viktor: 26:12

more. Yes. So it's, like expanding from the god prompt, original god prompt itself to, all right, we got the name or this soft names. And I just want to explore a little bit more, right? So I want to get something more tangible. And I'm asking for target audiences here. And I'm asking for testimonial as well. So I'm want to have a feel for what a target audience would say. So that's like a trick of not just asking for names, but asking for testimonial, because that's even better than having a persona because the personal is just quite. Dry. It's quite quantitative. It's not lively. So even though, if you put a picture to it, even you, say, make it believable and so on I found that if you just ask for a testimonial from the target audience, that's makes you believe that. Okay. And makes you easier to understand what is the target audience. And in this case, it came up with six different target audiences. First, recent graduates. Second, career changers. Third, retirees. Fourth, freelancers and gig economy workers. Fifth, disgruntled employees. And sixth, return to work parents. And it's writing testimonials, like return to work parents. It's returning to work after a long break was daunting, especially when I wasn't sure what I wanted. Now I have a clear direction and know what will bring me fulfillment in work.

Mihaly: 27:41

We had a client a few years ago and we did a similar research, an actual research, talk with HR people, talk with career changers. And so we organized focus groups in super professional settings. We recorded it, analyzed it. And after, I don't know, dozens of hours, maybe hundreds of hours of research, we came up with a similar result.

Viktor: 28:02

Yeah. That's crazy actually, because what's happening here is like you are making something which takes. Months and lots of people to work on something and lots of logistics to, to actually find people to do research on, then do the research and analyze the research and then just distill the research itself. It's insane that, yeah, it's and in this case, I'm getting back like recent graduates, fresh out of college, I was lost in the sea of potential careers. But now I have a clear understanding of what truly brings me joy and satisfaction in my work. Wow. I'm getting back a testimony, which is real, right? Which in this context, so in this whole context of what we went through with, god prompt, right? It's giving me a believable. And something I can resonate with, right? A testimonial and the same for career changers. It's like after a decade in a job that didn't fulfill me, I now have the clarity and confidence to pursue a career that aligns with my deepest values or retirees. Retirement no longer feels like an end. Instead, I found a new path that align with my interests and values, making this phase of life more enriching. Wow. It's just, I never would guess like retirees would be a good fit for this tool.

Mihaly: 29:24

This kind of testimonials, I know it's made up, but also I feel that they are real. And also this is something that you can brief to creative people. So if you want to ask your creative employees, like the copywriters or designers to please come up with something for career changers. You don't give them a lot of statistics, but you give something in one sentence that they can easily understand and

Viktor: 29:52

create. Yeah, in this case, it's just a short paragraph, right? It's three sentences and it's conveying the person much better than the person could. And that's why I said, I love it. This is one of the best hacks ever to ask for. Testimonials, because it's relatable much more so than any kind of persona, like describing retiree named, I don't know, John Martin, age 67. So it's like, very dry. You cannot relate to that, but with this short. Example is just insane. So these results are saving thousands of hours and hundreds of thousands of dollars.

Mihaly: 30:29

So just to clarify, we don't want to put these testimonials on our website because they are not real, they are fake, but we, it's much easier for us as human beings to understand these sentences than a

Viktor: 30:42

statistic. Yeah, sure. So yeah, absolutely. So this is something which we can relate to easier. Yes. That's like the first step to getting a target audience. And now I'm prompting and then quoting now do an exhaustive research on what makes a good landing page. The most important elements, the most important mental models and terminologies to make it a word class in conversion and also limitations and also what to avoid. So see and do a mini. Got prompt here again regarding landing page, and it comes up with having a clear value proposition strong headline and sub-headline. High quality visuals, social proof, which means testimonials, reviews, or case studies can increase credibility and be it trust. Call to action. Simplicity and clarity. Mobile optimization. So these are the seven things we should keep in mind. Also mental models and terminologies, AIDA, attention, interest, desire, action, FAB, feature advantage benefits, cognitive load, which is this term refers to the amount of mental effort required to process information.

Mihaly: 31:50

Cognitive load is my number one mental model when we design a survey and also when we design Anything that people has to understand. And also when I'm writing the people have

Viktor: 32:01

to understand. Yeah, also like it's a products of choice. I guess it was in Ciardini's book, one of his books that if someone could choose between 26 different honeys, then obviously people are stopping because yeah, having 26 honeys, it's like intriguing. It's, unusual. So more people stop much less people buy on aggregate. Then if you just sell only two or three types of honeys. So maybe less people stop, but more by, because it's easier to choice, you have less cognitive load. And also you are less concerned about making the bad choice. Or yeah, not having the perfect choice. Viktor, what I see

Mihaly: 32:42

here, that there are two type of mental models that GPT defines. So first one for the landing page format, like call to action and headlines, and also the underlying strategy to create. The cognitive load and the

Viktor: 32:57

IDA model. Yes, sure. And it's also like the F patterns. Eye tracking studies found that people often scan websites in F shape pattern and placing key information following this pattern can increase its visibility. So it's also coming up with these things which if you are not into web design, it's extremely useful. And once again, it's it's thousands of dollars. Which is it's saving for you each and every field you are trying to get results because it's collecting the most important models and telling you why you should be able to concern about that. And obviously you can go down the rabbit path and a rabbit hole and get into and, dig deep to understand like the F pattern or cognitive load or ADA and those models. But yeah, generally just having. Bird eye view of what you should consider. That's, pretty awesome. It's also called that's what to avoid. So it's in this case, overloading information generic stock photos, weak or unclear CTAs or call to actions and. Slow page load times. So that's quite neat. It's once again, slow page load times wouldn't be something which I would come up for the first. So it wouldn't make the shortest for me, but it's absolutely true. As it says, like they can frustrate users and lead to high bounce rates if the page is slow. And the limitation is one side does not fit all. So what works well on one landing page may not work well on other. It's essential. Continue test and optimize and also changing user behavior as technology and user behavior change. So two must landing page design strategies. So that's once again, it's extremely actionable because like just a few years back, mobile wasn't prominent that much in website design, but now on, Most websites have 80 to 90% of the traffic through mobile. So that's can radically shift. And also what happened with ChatGPT, like in a few months, they just reached 100 million users. So these chat interfaced. AI's. Just suddenly became a thing. And now suddenly they, lots of people use them, right? So this is a new interface, which are quite common now. You have to keep in mind. Once again, it's collecting what to avoid and what are the limitations extremely well, and this is extremely valuable.

Mihaly: 35:17

Okay. So let's

Viktor: 35:18

see the results. That was like the. Once again, I just prompted it to provide me an exact structure for a landing page. So what I do now is I'm prompting it and quote, now write all the copy for landing page for freelance and gig economy workers. When you get the testimonial and pain problem section, make sure to include at least three elements. Also make sure to include something specific, tangible, each testimonial and give proper persona. So once again, I'm just spinning up a landing page here and. It's once again summarizing the first paragraph. Sure, here's the draft copy for a landing page targeted at freelance and gig economy workers. Headline, discover work that truly resonates with you. Sub headline, navigate the gig economy with confidence and fulfillment in your work. Benefits, uncover your core values and interests for a fulfilling work life. Get clarity on your unique career path with our in depth structured process and leverage the insight of those who know you best and comprehensive self understanding and social proof testimonials. There's three, I'm just reading one, Jake, freelance graphic designer. Before using this tool, I was just taking any graphic design job I could find, even if they didn't really excite me. Now I only focus on projects related to sustainable brands, a value that is very important to me. And also there is Rose gig worker and also Lion freelance writer and pain point is also. Asking three questions of do you find yourself accepting jobs that don't align with your skills and passions? Are you feeling overwhelmed by the wide range of opportunities in the gig economy? And are you struggling to find fulfillment in your freelance work? And there's also a call to action So that's like it And what I did next was prompting it, all right, I've got this landing page and can you write a whole HTML using all of these information using tailwind CSS and make sure that the website is modern and easy to scan. And I just writing that you can use images. When appropriate with this format, but make sure the text is readable. And I'm just basically copying in source dot unsplash dot com slash random slash vid times height slash and question mark comma separated search keywords. So what unsplash is, basically a free. Stock photo website, and you can define the dimension of the images you want to get back. And you can define the keywords you are looking for. So that's like a neat way to spin up quickly. Website, and it's just writing me a website, a landing page. So if

Mihaly: 37:57

somebody's listening, it wrote the code in HTML, right? In chatGPT, and you... Yes. Put it in where, and what can we see now? I

Viktor: 38:07

basically put it on gsfiddle. net and I just copy paste the HTML. So it's basically a preview for HTML code. And what we can see is just like the landing page, what it generated. That's I just spin it up in basically five. Minutes, considering everything and that just one. Okay, so

Mihaly: 38:27

this is a real landing page with starting with a title, then a sub headline, then it actually have pictures of people who look like gig economy workers, right? So this is something that you can get feedback

Viktor: 38:40

on, right? Yeah, sure. So this is something you can already send to someone like, okay, this is the tool I'm working on. What do you think? And I can just spin up like six of these, right? So getting one specific landing page to each target audience is extremely key. And why is it good? The main advantage is not that, okay, this is something which is going to be final. And this is the best final landing page that could be the best thing about this is you can test extremely quickly. So you can create a landing page quickly. You can judge yourself, whether it makes sense or not. If it makes sense, you can ask for feedback. If you get feedback, you can use the feedback to prompt ChatGPT to make changes. And you can iterate extremely quickly. And you can explore what the different target audience is. You can, if you see the landing page, you can even see okay, is this something I'm passionate about? To do a solution for, right? So maybe I realize, okay, what resonates with me is actually doing this tool and tailor this tool for retirees to give them new meaning. And you would have never guessed it if you don't try, right? This is like for exploration, for getting started faster. Yes. And

Mihaly: 39:52

also this is a landing page that you can use to buy traffic. So you can get Facebook traffic, Google ads, Facebook ads. And. I'm sure that the conversion rate won't be great, but you can compare six versions, right? You can see that some of the versions doing 100 or 200% better, and okay, I got the qualitative feedback, which means somebody from the target audience give me feedback that, okay, this is something that I'm interested in. And also you have the quantitative feedback. So you know, which one of the six versions are doing better, even it's the results would be. Not that good at this point okay, this is a landing page that I can give to a professional designer, UX designer, UI designer, website designer, and I think creating the first version of your startup website is like 5% or 10% of the cost than it was before,

Viktor: 40:44

right? Yeah, it's even less. It's like almost like a fraction of a percent. And if you see like now it's actually responsive. So it looks good on a desktop and it looks good on a mobile as well. And it costed. Nothing, it's actually costed nothing for me, only five minutes of my work. So that's insane. If you think about that, that I did just like thousands of dollars worth of naming consulting. And also it did for me hundreds of thousands of dollars worth of research. And also just create a big landing page. And I can just rotate it for each and every single target audience and getting a responsive design. That's insane. This is the best time to be

Mihaly: 41:24

alive. Any more conclusion here, or can we move to another two examples? Yeah,

Viktor: 41:28

sure. On my screen, what you can do also, you can rewrite your messages on on ChatGPT and this last step of creating the HTML, actually, I just rewrote it to now first define five Facebook target audiences within the niche of freelance and gig economy workers, make sure I can specifically target them on Facebook and write three, three Facebook ads for them. So this is insane again, because I get the landing page and now I have the different target audiences within this target audience, right? So it's like freelance designers, gig economy delivery workers, freelance writers and content creators, freelance developers and IT professionals, freelance consultants. And then it's coming up with, for example, for designers, like an ad copy of lost in the sea of design project, this could have a truly sparse or creativity and aligns with

Mihaly: 42:20

your values. What's the bag that when you create a Facebook ad one of the challenge is to, how can you target these people? And it actually gives. Useful options. How can you target a freelance gig economy worker? So for example, people who work at Uber or DoorDash, so we're already gig economy workers. So this is it's not junior level or maybe at least junior level Facebook ad manager.

Viktor: 42:50

Advice. Yeah. And once again, it's like the end goal here is not that whether it's replacing someone who is extremely proficient in ad management, because obviously someone who is extremely proficient using the same tools is going to get further ahead than you or me, who is not doing with that advertisement all the time, right? The goal here is that it can speed up discovery by thousand times. It's almost like the speed of light. It's like you can generate different target audiences. You can generate different copies and you can test them. You can quickly iterate. And if something is seeming to work, now you can double down. You can think of, okay, how to improve it. How can I be more creative, right? You can even ask for feedback from the target audience, right? And incorporate that feedback and prompt ChatGPT. Okay. So this is the feedback, right? Menu advertisement. So that's like exploration is just. Spell out by a thousand times here. And also it's insane that how well it resonates, what it writes, because for IT professionals who freelance, it's right. Developing great software is one thing, developing fulfilling careers is other. Start your journey towards Cori's satisfaction today. So that's yeah, developing great software is one thing. It's holy shit, it's actually talking to the target audience. That's like crazy. I don't know what's your opinion about that, but for me it's, insane. It's unbelievably

Mihaly: 44:17

fast with the process. Yeah. And also if somebody want to read the whole conversation that you had with JG pt, we will put it on disruptor digest.com.

Viktor: 44:28

Yeah, sure. And it's going to be inside the god prompt protocol. So if you go to disruptor digest.com and sign up, it's free. It takes one or two minutes it takes nothing and you get all the

Mihaly: 44:42

examples. Okay, so I would give you another example. The two examples will be the first is an email and the second one is a LinkedIn post. And I will just quickly go through and you can be also. Read this whole conversation, this letter digest. com. So the thing is here, the purpose of this email would be to persuade free users of Grammarly, which is one of the biggest grammar checking applications to upgrade to paid versions. So I think this is a very common problem and this is a very common thing in software, the service companies that there is a free trial. And when it ends, 95% of users don't convert to a paid version. So I'm sure that there are always at least one email to persuade these users. So this is what we want to create here, an email for free users after the free trial, and we want to convince them to. Upgrade to paid versions. Also, we will give them a 50% discount. And it's important to note that it's a unique opportunity and not

Viktor: 45:42

usual. What you try to avoid here, and that's quite clever, is you don't want to train the market to wait for a discount, right? So it's present it as it is okay, this is something. It's like a one time opportunity to upgrade and please write me a text, the exact email, right?

Mihaly: 46:04

Correct. And also the last part is additionally, we would like to highlight that new features are on the way for Grammarly Premium users. I think this is a very real brief that I have seen several times for our clients. And I just put it as it is without the god prompt to ChatGPT and ask him, okay, please generate this email. And. On the screen you can see that it's very long email, like seven paragraphs, and it's very long sentences, and if you go into this, it's not concrete, it's a little abstract. Let's go to our GAT prompt. This is the version 3. 1 of our GAT prompt, and change two things, because I know this is for marketing communications, there are two mental models that we have been using in the last few years. 15 years for every client and every kind of marketing communication. This is jobs to be done framework that we extensively talked about in the marketing research episode, and also Chialdini's principles that I think we mentioned it all the episodes in the past. Adding to the second step that please make sure that you include jobs to be done framework and Chialdini's framework. And also a little bit, even go further to please define the jobs, pains and games for target audiences. The second thing I added is the format, because after a few iterations, I found out that the formatting is extremely important. So for example, the first thing about the first generated email I mentioned to you is that it's very long. So I think one of the most important things about cold emails and all kinds of marketing emails to make it short, because we. used to a few seconds on TikTok videos and so on. We don't want to read long emails. So this is the two things I added to the GAT prompt, but everything else is the same. So first for the clarification step, ChatGPT asked me clarification. Is it the typical users who completed the seven day trial? And I said, yes. And also what are the main reasons you believe these users have not upgraded to the premium version. And this is something that I think. SaaS companies know, or at least have an opinion, and this creates output that I think the biggest reason that these users don't convert is that they are unsure about the value of premium features. And the third thing is, what have been the most effective perion taxis or messages from this group in the past? Okay. I really love this question. I always ask when we start to work with a new client, what worked you in the past and what didn't? And so you don't have to. Repeat things that didn't work and What work in the past and I, and put limited discounts because in my experience, limited discounts, increased conversion rate in all field, every field that we have worked. Okay. Step two, mental model research and also I included the, please make sure to use Chaldean's principles and jobs to be done. And they start with them reciprocity, scarcity, consistency. We talked about them in previous episode and also the jobs to be done, but also it mentions a few other. Mental models like the decoy effect. So people tend to change their preference between two options when presenting with the third

Viktor: 49:19

option. The standard example is like you have a small soda, which costs like two dollars. You have the medium one, which costs like five dollars, right? If you add like a big one, which costs like 5. 5 dollars, then it's suddenly increasing the average order value by a lot because the difference between five and five and a half dollars It's so small that people are quite intrigued by the bigger option because it seems like a good deal and that's like the decoy effect. So the average order value just increases. Correct.

Mihaly: 49:53

And the example that GTP presented here is, I will read, present the monthly, quarterly and yearly plans. With the yearly plan looking like the best deal. Also two other mental models. We don't have to go in very deep. And I said, okay, great. Proceed. Next one, jobs to be done. And okay. It defines the customer jobs, the customer pains and customer grains. And I just want to cherry pick a few. For example, customer gains, increased confidence in their writing abilities. Improved grains on written assignments. Yeah, I think great. And also I asked for objections. For example, one objection of this target audience can be, I can get these services for free elsewhere. And then I also asked JGP to handle this objection, and it wrote, emphasize the uniqueness and superiority of Grammarly's features and user experience. The convenience of having all writing tools in one place, and the potential time savings compared to using multiplied tools. And I think, yeah, that's great. I was a Grammarly user in the past, and I think, yeah, there are... A lot of tools in one place. So if somebody wants to craft very high quality text, then Grammarly is a great tool. And I said, okay, great. Proceed. Next format, length, tone, and other considerations. This is my favorite part. I think for example okay, basic formatting of an email, start with a greeting, nothing new, but also one of the last step is to put a call to action, and I think this is something. A lot of marketing people forget, at least if I check my inbox, that there is not a clear call to action in the emails, but chatGPT will

Viktor: 51:37

know that. So for those who are just only listening, so what is the structure in his mind? So what ChatGPT is telling

Mihaly: 51:44

you. So the structure is first, start with a personalized greeting. Second step, a reminder of a completed trial and the value they found in Grammarly. Yeah, it's great. So it, gives them a context and why are we writing this email? That's

Viktor: 51:58

Like genius because I remember that Alex Hormozy was like giving the same template for how to raise prices. So if you raise prices, you always have to remind people what value they got. So that's genius. So that's like the same thing here. It's yeah, your trial is completed and this is the value you got. And that's something that lots of people forget that you have to highlight, okay, what is the value they already got?

Mihaly: 52:25

Okay. So step introduction of the offer. Great. Straight to the point. No fluff. Fourth, highlight the benefits of upgrading. Fifth, addressing potential objections. Six, call to action with urgency, and seven, a cordial sign off. I wouldn't be able to come up with a better

Viktor: 52:46

structure. Yeah, this is quite neat.

Mihaly: 52:47

And here comes the next part of this answer, the length. And it says it should be around two or three hundred words. It's... It's short. And the first example was at least 10 times of that, like 3000 words. So GTP knows this, but you have to give them the context.

Viktor: 53:05

Yes. Okay. Go up a little bit because it also had other points like bulleted lists. So that's correct. It's it has to be easy to scan. We're also using emojis, so it's going to be less dry. And also the tool should be friendly, direct and professional. Yeah. It's yeah, and it says it should be clear and efficient. So that's once again, nothing fancy, but it works. Yeah. I checked

Mihaly: 53:29

several hundred prompts that I found on the internet and the blank. Part of most of them was the tone, to define the tone, but I don't know what kind of tone exists and what kind of tone should

Viktor: 53:43

I use. Yeah, it's overwhelming. I saw this list like shitload of tones and you don't, it's like even, it's more confusing than, even don't having that. Yeah. So

Mihaly: 53:52

should I be quirky? Should I be joyful? Should I be friendly? What are the differences? But ChatGPT can give me a general direction. And I think it will most of the time, or almost 100% of the time, it would be better than what I would give. Yeah, sure.

Viktor: 54:08

And in this case, it's already doing the research, but the got prompt is giving the research, right? All the background information, all the context. And it's, since it's doing the context for you It's, already giving the tone and it can give the tone because it understands the context. So it's genius. I love this example. Okay.

Mihaly: 54:26

In this case, Chajipit already generated an email. But there is one more step or a few more steps. Next is inversion research. What should we avoid? Yes. Overly aggressive sales pitch. Totally agree. Long and complicated email. Thankfully, we have covered it with asking for the ideal length. Lack of personalization. Great. Neglecting to address potential objections. This is one of the top rules of sales. And last one, lack of clear call to action. And this is, I think, one of the most common problem with marketing communication, not created by

Viktor: 55:04

experts.

Mihaly: 55:05

Yes. Great. Proceed. Book research influenced by Cialdini. We have already covered it. Made to Stick, my second favorite book on marketing communications with a few examples. Make a message simple, unexpected, concrete, a great one. We don't have to go deep inside here. And the third book that he mentioned is Predictably Irrational by Dan Ariely, which is a very practical book on behavioral economics. We already mentioned the decoy effect and the anchoring in the past and happy that HIGPT will use these models. Proceed. And now it will write the email and it comes with like this subject, your chance to unlock Grammarly's full potential, 50% off. Okay, I think that's a great headline. Why? It starts with you or your, and this is a important rule in copywriting to use the word you in English as much as you can, because the user will feel that the message is written to him or her, and also a chance to unlock grammar's full potential. This is the classical example of the curiosity hook, so something is You don't know something, but here is your chance to know and, experience and also some nice deadline and

Viktor: 56:22

50% off. What I like about this is, one of the biggest mistakes I see that okay, there's a chance for you to buy and, buying and spending money is pain. So you should never communicate from the perspective or from the angle of someone is Spending money, you should communicate of what they get. And this is they can unlock and full potential. This is something which is, it sounds exciting to be honest. It's yeah, okay. Let's see, what does it mean exactly

Mihaly: 56:48

for me? I think for the first version, I wouldn't be able to come up with a better subject line, even after sending 1. 6 million emails, although I would AB test some shorter version of this one, but this is something that you cannot judge. You have to test. Yes. Okay. Here comes the email. Dear username, we hope you've been enjoying your experience with Grammarly. We noticed you recently completed your seven day trial and wanted to thank you for giving us a try. Okay. As we mentioned, it gives context to the user. Second paragraph. In appreciation of your interest, we are extending an exclusive one time offer for you. For a limited time, you can upgrade Grammarly Premium at 50% discount on yearly subscriptions. Okay. In two sentences. It defined the offer for the user.

Viktor: 57:36

For those who are only listening, also doing something which is extremely important to make the email easy to scan. So it's bolding the most important points and you never prompted it to do that. And it's doing it on its own, that exclusive one time offer is bolded and also 50% discount on our yearly subscription is bolded as well. So the main points that it's time sensitive and the exact offer is bolded and also But it gets into the what are the values. It's also highlighting the values as well. So it's just nice that since you gave it proper context with the got prompt. It's also doing the editing of making it easy to scan on its own, without you saying anything about that.

Mihaly: 58:18

Next paragraph. Grammar in premium goes above and beyond to provide, and it comes a bulleted list. Advanced grammar, punctuation and style corrections to help you write with confidence. Okay. This is a very great structure because it starts with the features, but also define the benefits. As far as I see for marketing beginners. They always list the features and they are sure that the user will understand. And no, because without defining the benefit, it's. It's a bigger cognitive load, and people don't know how this feature works.

Viktor: 58:52

Yes, obviously, if I I'm a user, I want to know how, what it means exactly. Yeah, there's a feature, but I cannot imagine without any help to, okay, how exactly it's going to happen. In this case write with confidence, holy shit. I want to write with confidence, and I understand that, okay, that makes sense. That if it's correcting my style, I can be confident that, okay, what I'm writing is professional.

Mihaly: 59:14

Okay, another bullet point is pluggear is detection. Across 16 billion web pages on ProCast databases, protecting your academic integrity. Same thing starts with the feature, even put some numbers for credibility and protecting your academic writing is the benefit. The thing here is that I think Grammarly is a big enough website that GPT was trained on it. So it knows. Some actual benefits that they state, but the good thing is even if your web page is new or the low traffic and GPT training didn't include it. Now there are plugins, the browsing plugin, so you can put your website and ask the ChatGPT to scan your website and put it in the context and it can come up with the actual features and benefits that you mentioned on

Viktor: 1:00:03

your website. One of the good friends of mine just did this that. They have more than 2000 clients and he wanted to have a better understanding. So he just basically wrote an extremely short script. We are talking about like 10 lines maximum in Python. So it was, what it did is was scraping the websites. It's a beautiful soup in, Python. It's just, you scrape the website content and then it provided to chatGPT through API. And it just asked chatGPT to summarize. What this company is doing based on its website. And even though the websites were in non English in this case, they were mostly. In, Hungary and Romanian it gave perfect summary of the website. And what he did after he just clustered the clients based on who are similar and get like different clusters. And then you can basically get a better understanding. Okay. Who are the different target audiences I'm already serving. And if you layer on top usage data and, lifetime value data and customer satisfaction data, you can have a nice understanding of who are the best customers of mine. In terms of revenue or their happiness, and also who are the worst customers. And that's again, giving you a better context. And then you can specifically use ChatGPT to write improvement plans, how I can improve my software in this case to serve my clients better. What kind of offers can I craft for these target audiences to make them happier or more successful and those kinds of things. So this is quite genius that even without plugins. If you can code and you don't have to code well it's, not even complicated then then yeah, you can get a lot of insights, which previously would take once again, like thousands and thousands of

Mihaly: 1:01:55

hours. The last paragraphs of this email is this is a unique opportunity tailored specifically to dedicated users like you. I think this is based on Chialdini's principle called liking, that you're more likely to buy if people like you

Viktor: 1:02:09

are buying it. This is extremely important because if in the previous example, what I provided, if I'm targeting retirees, then obviously the testimony should be from retirees, right? If I'm targeting gig economy workers. They should be able to gig economy work, especially if it's more niche, it's for delivery food delivery workers, then it should be from them. So it's yeah, if you're a small and medium sized business in, any kind of niche renewable energy, it's more convincing for you if other. People in the same field are talking about the product and how it's helping them.

Mihaly: 1:02:48

Next paragraph, think of this as a small daily investment, less than a cost of a cup of coffee towards boosting your grades, communication skill and future career prospects. Okay, I think this is great copywriting as well. Remember this is a limited time offer and upgrade Grammarly Premium today by clicking here and unlock the full potential at Grammarly at half the price. So this is the clear call to action that we have been waiting for.

Viktor: 1:03:14

Dr. Mihaly, I guess you just did with with GodPrompt and and and your work with Grammarly. It's like you basically did a masterclass. And show the masterclass for people, how to handle subscription, how to increase revenue for SaaS companies. So it's extremely valuable. And you lost an opportunity here to invoice thousands of dollars.

Mihaly: 1:03:35

Okay. Just a few advanced things here. So I think this is right. I would send this email. Maybe check the benefits are actually the features are actually correct and things like this, but I would send out this email and compared to the current version. Next one. Advanced level with this code prompt. I really want to create advanced level marketing messages. And the next thing I asked it, how could I improve this email further to make it more convincing? Give me 10 ideas. To this point, it's about copywriting and, the benefits and the, features and the psychological levers like the Children's Principle. But now it says, okay, you can add testimonials. Great example. Use visuals. So it's not just copywriting, but it's like the concept, conceptual copywriting. Also use social proof and of course you have to get real data for this. And send a follow up email. So again, it's not just copywriting, but the email strategy

Viktor: 1:04:38

itself. Yeah, it is a whole process of not just sending one email, but actually it's like a sequence of emails. And yeah, I like it. It's quite neat that... All these terminologies, all these actual copies it produced with basically zero inputs on your side. So you just basically provided a god prompt and the goal that, okay, you are grammarly and you want them to upgrade and some minimal amount of extra information was provided, but otherwise. All the results came from ChatGPT. Okay.

Mihaly: 1:05:09

Last example, Viktor, just in five minutes, we won't go that details. Just, we want to show you that you can use it for LinkedIn posts as well. So what we did here is we get the first. Episode of our podcast called email marketing and asked a very great software called decipher that AI to create a summary, create timestamps and create a key takeaways based on the transcript that it generated from the audio. The first prompt was without god prompting, just as this is a podcast episode summary, please read it and say, okay, when you are done and I put about five, five pages long summary and key points of the podcast episode. The goal for us is here to create LinkedIn posts based on this episode and persuade people to listen to this episode. So judges PC said, okay. And I started with a god prompt clarification step. It asks for who is the exact target audience for this LinkedIn posts. I think great question. And what level of expertise do they have in cold email marketing? And this is something I haven't thought about because I have the so called curse of knowledge. So it's very hard for me to think about the mindset and the level of knowledge of novices. Of a subject that I'm an expert of, but I replied that level of expertise in cold email marketing is zero or close to zero for this target audience. And also what goals do they hope to achieve through cold email marketing? This is part of jobs to be done framework, but it asks if I know anything. And I think the most important goal for coordination, they hope to generate new leads for their companies. Okay. Mental models, research, jobs to be done, Cialdini some other relevant the IDA model, which you already mentioned it's important for a social media message grab the attention. And okay. I as proceed it elaborated on the jobs to be done framework. What can be the pains gains of this audience formatting style. Okay. Here comes interesting thing for format. ChatGPT replied, LinkedIn posts allow for up to 1, 300 characters, including spaces. To stand out, structure the content into clear and concise paragraphs, three or four

Viktor: 1:07:23

sentences each. Basically easy to scan, right? So if, that's quite, once again, goal of the god prompt is to get you up to speed in any kind of field. And in this case, if I'm not well versed in LinkedIn, this is extremely... Helpful for me, but obviously if someone is bread and butter is linked in post and linked in marketing and they do it the whole day And they know the ins and outs then obviously for them It's they can already prompt and give all this information. But in this case what is happening is The god prompt and GenGPT is generating the research, the generating the outline, generating, okay, the character limits in this case, or the length or how it should be structured and emojis and the tones.

Mihaly: 1:08:09

For the tone, it says professional, informative, engaging, totally agree. And if you remember that for the email, it says friendly and for here, it doesn't say friendly. So yeah, it

Viktor: 1:08:20

It understands the nuances of the medium where the communication is happening. Okay. Again, it's already

Mihaly: 1:08:26

started to give me a few examples, but okay, let's go to Inversion Research. Here, the same things, what should be avoided? Long posts, too much information, lack of personal touch, also absence of call to action, same as email. But it's good to remind us. And

Viktor: 1:08:42

also overuse of jargon. And ignoring engagement. Yeah, so if people are commenting on the post, then you have to, you should reply, because otherwise, what's the point of your doing LinkedIn marketing? So this is quite neat as well, that it's very explicit on even obvious things, which would, experts would be blind to. Because maybe if I'm a LinkedIn expert, I'm just doing something naturally and it doesn't occur to me that, okay, I should prompt, or I should tell you that you should answer each comment, right? So that's quite neat that it's throughout, it's doing an exhaustive research indeed. Yes. And the step

Mihaly: 1:09:19

six is book research. And I think it will be pretty helpful for me. For example, mentioning Jab, Right Vaynerchuk, which is, I think, a very practical book on Social

Viktor: 1:09:32

media. But also like email persuasion is in there as an example. Yeah, and LinkedIn Riches. So it's like having directly and indirectly related books as well. Yeah,

Mihaly: 1:09:45

and in the past six years, I didn't post anything on LinkedIn. But for this podcast, I plan to have a better LinkedIn game. So I know that these are the books that I should looking for first reading their summaries and they are great and insightful that I can read or listen to the whole book. So I want to be better in LinkedIn and it gives me a great direction. Okay, and proceed. Next step please write the actual emails and I will read one for you. LinkedIn post one, cold email marketing introduction. Are you new to cold email marketing? Here's a nugget of wisdom from Dr. Mihaly Kertész, an expert who sent over 6 million emails. Cold email marketing can help reach dream clients, scale inbound marketing, and stand out from the crowd of low quality email spammers. Dive into the podcast to learn more. Hashtag cold email marketing, hashtag out email outreach, hashtag inbound marketing. What

Viktor: 1:10:44

do you think, Viktor? For my taste, I would include some specific value from the podcast. So that's how I would improve it. Otherwise, it's just yeah it's, much better than someone randomly trying to write a LinkedIn post. And it's also about surprising once again, it's it's quite insane that the summary was generated right from the podcast episode. And now the LinkedIn is generated as well, so it's like almost our job is, taken by AI completely. And so we won't be sitting here and, discuss.

Mihaly: 1:11:18

You are lacking actual practical advice from this post, right? But I think the second one states that Templates can easily become burned out, recognized by spam filters and will not be effective. Oh, wow. So this is one of our key insights of our podcast and Chajipiti currently identified and put it in the LinkedIn post. And also the key personalization and targeted outreach. Yeah. So I think. Some of these posts are actually

Viktor: 1:11:45

good. Okay. This is a neat thing. Is it like, and even if that's like the formatting of the God prompt is using the last step is to use an exhaustive way of solving this problem in a world class manner. So it's not just providing one, but providing you several different examples, and that's quite neat that yeah, it's, you can pick and choose basically, and you can even tell okay, these are two that I like. Please provide new. Variations for them. Okay.

Mihaly: 1:12:13

And the last step is I ask, how can I improve further these LinkedIn posts? And it said, add statistics. Yeah, they usually get use more hashtags. It's interesting for me. Incorporating more relevant hashtags can increase the visibility of your posts. So sometimes too much hashtags doesn't help, but in this case, it would be something that I would test. Yeah, exactly. Tech people or companies. Great example. We will do that for sure. Include a call to action. I will check them, but all the posts should be include a call to action. I listened to this episode. Okay. I asked to try to find more LinkedIn posts based on this recommendation. So I don't have to think about this. Just come up with examples and put it in and LinkedIn.

Viktor: 1:12:55

One thing, but I noticed that the second iteration after you ask for improvement. Formatting the post better. So there's several paragraphs, right? So it's, more split into paragraphs. It's easier to scan. It's like the formatting better now. I will read

Mihaly: 1:13:12

one. Okay. It starts with a gigaphone emoji. Attention, marketing managers. Are you struggling with your cold email outreach? Dr. Mihaly Kertész shares some enlightening insights. One of the latest podcast episodes, he recommends avoiding templates. as they can lead to poor results. Instead, he emphasizes the strategy behind the email and using cold outreach for targeted research. Have you tried this approach? Let's discuss. Okay this is, it is way more engaging style, right? So asking questions, call to action

Viktor: 1:13:46

to discussion. Yeah it's, insane. Now it's it's, properly calling for yeah, engagement on the under the, post itself. So yeah, it's, it indeed became much better. I love it. Okay.

Mihaly: 1:13:59

For conclusion. So we wanted to create one god prompt and we show you a few examples and here comes the best part. Please go to destructor digest. com and try this. Because you can see an examples here but it's even more mind blowing if you try it with your company, with your offer, with your product.

Viktor: 1:14:19

Right, Viktor? Please, share. So please, whatever you find, because as we said, and as Dr. Mihaly also shared, this is his version is 3. 1 already. So it's like a work in progress. So please share your results. And also just three quick hacks I want to share with you. To get even more value out of it, it's one is if you want to get more creative ideas, you can say, give me ideas with temperature equals one. Because that's like the temperature setting is between zero and one. Is. Something that which a transformer is using to get more creative ideas. But you can also just state in English that please give me more creative ideas. If you don't like the line of thought it produces. And also you can say that avoid hallucination. So don't make up new things. So for example, in this case, if. It would use information which wasn't present in the podcast itself that you can say, okay, please don't make up new things. Just use the notes from the podcast I provided. So that's if you are explicit about that, he will use that and one cannot and almost nobody's talking about this. Okay, so this is something new that you should always say because. Okay, once again, hallucination is not good, right? Because it's coming up with information which is not true. You can say that, okay, only use the information I provided, but that assumes that you have the right information. So you also have to say that if you think that the information I provided is wrong, Please give me a suggestion, please notify me that you found something potentially wrong in the data itself. So just to give you a really example, if data was used to generate advertisement for accommodation booking site, that it produced content like New York. central pork without an hour. It's not central. It's sent our park and it was because the model got the data and the data was not right. The central and in this case, if you just from the model that if you find, if you think you found a problem in the data or an issue in the data or data is not right. Please notify me about that. So that's again, you have to handle the edge cases. And that's like the essence of programming, that you're always handling the edge cases of, okay, this is the task, if you cannot solve it, please don't make information up, or just say, don't know, or ask questions, full of questions and those. I think, and also a third leg is as a must again, that, okay, if you think what I provided is faulty, please notify me. So that's like a quick hack. I don't really see anyone talking about that, but that's very important. And also a few short learning is important once again. So the quality of the output can. Improved by six time, if you provide examples and that happened with our intro. So that exactly what we did was providing the summary of our podcast and trying to prompt to create a podcast in intro. And once we supplied with two real intros from the Huberman podcast lab okay, these are the exact intros which we think are valuable, which we think are good, then it's produced much better. Output. So if you can find, if you're the end result you're looking for is like a Facebook ad and the end result is like an intro, the end result is an email. And if you can have some kind of swipe files or some kind of examples, which you think are good. Providing them as okay, here is an output, which I think is good. So get inspiration from that ChatGPT and these language models can handle it very well. So these are the three hacks and this can even skyrocket your results, even above what the God prompt is already doing for you. Yeah, they are great. Thank you very much, Viktor. All right that's a wrap. See you around. And if you found this useful, please rate, subscribe, and also go to disruptordigest. com and sign up. We're gonna send a god prompt to you and also please feedback. So give us feedback about what you find and whether you like or not. What we are doing. Bye. Bye.

Mihaly: 1:18:33

Did you enjoy this episode? Don't forget to subscribe and rate us, it really makes a difference. For all the insider info, including show notes and transcript, check out disruptordigest. com. Are you a part of a software as a service or a cutting edge tech company with more than 1 million in revenue? We're offering a limited opportunity to a select few to accelerate growth. Visit artisan. marketing to learn more. See you next time.