How I Automated $102,000 in Deals in 9 Weeks. Make.com + Cold Email Icebreakers & More

Hey guys, how are you doing uk? My name is Michael Elliot and today I'm going to be showing you how to scrape any website using make.com use that information in a way which will allow you to be able to extract key and core points. Things such as website updated, one line, icebreaker, reach out cold email, about US sections, etc. Different individual things. You can play with the variables later, but this is going to be something that showcases to you exactly how you can do it and how you can improve efficiency when you're talking about things like cold outreach. Sl so let's dive right in. When we're looking at this, the first module that we need is a HTTP request. I have it down here because it's a common tool that I use. But if you don't simply click Add module, go to HTTP and hit Make a request. This Make a request sends a HTTP request to the to the specified URL and processes the response. Very simply put, it pulls the information from scrapes a website URL, which is exactly what I just said we're going to do. That's some JSON. The business that I'm going to use is the Reporting Hub. Now the Reporting Hub are a company that I actually built out all of their ticket in triage. I improved. They were manually handling tickets, inbound tickets, I reduced that down to 8% in handle manually, 92% are currently handled by knowledge bases or by historical contents. In terms of chats with real agents, it pulls all that information, goes through the loop, finds the correct one, but sticking on task. The method is get because we're trying to get information and receive it. We're not trying to patch anything, we're not trying to post nothing, forget about all of those. Get is the process, we don't is the method. Sorry, you don't need any other changes in here. You literally save it. But one thing that I'm going to say to you right now, off the bat, I don't know how experienced you are in make, but running modules individually is absolutely critical when you're building out large flows. In this particular use case, this wouldn't have harmed you that much. But if you've got 258 modules inside of a flow for some massive, huge flow that you've got for something, right? Whatever it is, crazy amount and you need to run a module, you need to make sure that you understand, you understand the value of being able to run said module on its own. It is such a basic feature, 90% of you probably already know it. But just in Case you don't, I'm going to show you. You right click this now and run the module only and it will give you. It tells you the the input the report and hub. It'll give you a Status 200 code which means it was a success and it'll give you the data. This is. This data is what makes up the whole of the reporting hubs landing page, the URL we just put in the home page that what we just put in now. You can be using any site for this. Don't know what site you're using Disclaimer Some sites do have blockers. There are ways around specific things. But that's not what this video is about. If you're scraping any normal site, 99.9% of sites will be absolutely fine to scrape. You can use my example if you want just to use this. But this is what you see. Ugly but contains specific valuable information that we want. No code Reporting portal developed for Power BI using embedded technology. That is important, that is critical. You can see how already the value of scraping a site and just receiving this information could be used. Competitors let's say you're trying to reach out to SEO agencies in Texas and you want to. You find the biggest agency in Texas, you scrape their site, you find something that they are good at and then you use that information inside of your cold out cold outreach icebreaker in the email and you reference their biggest competitor. You're going to put ego and fear into into that individual who reads the email which should be founders high up senior. This is a whole different video on how to scrape and I will produce it for you and link it in the description below. But I just wanted to showcase to you how valuable that can be because there's so many use cases. But that's just one of them and it's one that I've used very well historically. If you haven't tried that, try that in your cold email outreach. It's a little bit more granular but if you choose large cities, London, etc. You go to these places, you. You'll find plenty. New York, you'll find plenty to go after. So this pulls it. The reason that the website looks like this is simply because of a rendering engine. You don't need to bother yourself with what a rendering engine is. But it draws this makes it look, it look very cute. It makes it look sexy. They've actually got quite a nice site which shows their interface. I think that it's. You know, I do like the site. It's load speeds are good and you know, it's well built, but it's all built off of this. There's no magic behind it. This HTML is what it is, what's required. Now I believe, I don't care unless it's completely excessive. I don't care about additional operations. I don't care about using an additional operation in my flow and the spend. So I will always do the most logical way format in terms of the way that I look at the flow so that then in future, if I ever have to come back and edit it, it's instant in my head. I know exactly what is happening and where. You can obviously rename them, which gives you that ability. But regardless, we now have that data and I'm going to show you why running that module was so important. So what we're looking for next is a text parser. The next step of this flow is to convert the HTML that looks horrible, has golden information in it, but looks horrible, into plain text or text that we can use. The text parser offers a HTML to text transformer. I select that and it's asking me for the HTML. You don't want to copy and paste the HTML. You want this HTML to be a variable. The reason why is because you're not just going to be coming in here and running it one business at a time. You're going to be using this flow at scale. That's what you should be using it for. And I will do that in this video and show you how to. So you select the data module. You see, you see how that is blinking? Did you see. Well, before that happened, did you see how. Did you see how that module, how that module is blinking there like that means that the data is coming from that module in a large flow. Even the most seasoned of no code experts with development backgrounds, etc. Can get lost trying to find them. You will learn how to code manually eventually. So you'll learn how to. How to code manually in terms of variables. So you know, like 18 data curly brackets and it will pull the variable together itself. But for right now, you can just use the list. They're not advised that you try and get stuck into anything like that. You just do that. But that's why we had to run this module on its own, because now we have the variable. So now I hit run. It's going to give me this error message. This error message is baloney. It's just because the transformer is the last module in the route. That is good practice. You don't want a transformer as the last module in a route, it generally speaking means that you've not finished the flow. But for this case, when you're building flows out, you need this to happen to check data. You don't want to go and build your whole flow and then come back and realize there was an issue here. So now we click the little bubble. Ignore this because it's just because it's the transformer. The input was the HTML ugly looking. The output was text format, readable to a human, much more digestible. For me I can see that it's a plug and play reporting solution that deploys to your as your environment provides instant customizable delivery platforms. All of this I can read it now in a clear context. And obviously if I can read it in a clear context, then so can AI. So that's the reason why I do it this way. AI could do a pretty good job. In all fairness, AI could do a pretty good job of taking HTML and breaking it down itself and having that understanding. But I like to condense my information. It's why I mentioned about operations and do it in the proper route which gives me the correct sort of output based off of what I desire in my own eyes and brain. As you become more seasoned and developed, you'll learn there's areas where I cut corners and stuff. All this will come together. But in this particular case, I think that it's important that we're using this now we have the clean output. What do we want? We want to use AI. And in this particular use case, as I explained, I'm going to choose it to find a specific piece of content that is about the business. I'm going to use it to find when the website was last updated and I'm going to use it to create a one or two line icebreaker as an intro. So I'm going to use OpenAI, you can use Claude, I prefer OpenAI, any of these modules. One thing, if you haven't got your connection, you need to add the connection to it already. Simply click add and connect. It's very, very simple now. So many models to choose from. If you're a real nerd and you really go into the back end of this stuff, you might find that seasoned Make.com users say to use a specific type of model to get the best output. For cold email outreach, icebreakers or for individual tasks, they can be funny. I'm not saying that they're wrong, I'm just saying that I don't care and I'll use ChatGPT 4.1 for the majority of them because it just keeps it simple. If you are price conscious, though, of how much it's going to cost you whilst running things like this and testing these flows, something like the 04 Mini would be absolutely perfect. Like it would cost you less for the outputs and obviously as you get bigger systems built in, in your testing, you can use them and then just either change it later if you, if you want to see that improvement or not. That is more than reasonable. There isn't, there is no issue. So now we get into the, the sexy sort of implementation of, of prompt engineering, right? And I'm going to do this off the dome. I'm not going to. I've got the, the JSON format in, but I'm going to do this off the dome. So this is not going to be perfect, but it's going to be a good use case for you. Now you have four individual options. User Assistant, Tool and developer and System. The two that we're going to be using today are user prompts and developer. The developer and system prompt, you'll use one developer and system prompt per individual chatgpt model. And if any of you have seen the film Big Hero 6, you will know who Baymax is. Baymax is the AI robot that wakes up and goes and can tell you what's wrong with you. It scans you, it'll help you, it tells you issues. You've got, you know, an arm that's scratch. You've got this, you've got that. This developer and system prompt that I'm about to fill in is the chip that they put inside of Baymax. If you haven't seen the film, that's unlucky for you tonight. That's your homework because it is excellent. Me and my daughters love it. But inside of here, you could be tempted to write a stupendous amount of information that goes through, you know, copious amounts. You're this, you do that, you're an excellent this, you're an excellent that. Less is more, less is always more when it comes to the AI integration. So I don't know. You are an intelligent website scraper. Done. That is all I'm going to give it as an initial prompt. That's you are an intelligent website scraper. So it knows that its job is to scrape the website. Let's add another message and this time we're going to make it a user prompt. User prompts can be used for an abundance of things. It can be used for so many individual things that I'm not Going to list them all, but giving it examples of cold outreach, giving it examples of, you know, good about sections, examples of competitors, you know, pages, whatever. You can literally use user prompts and prompt engineering to its core, as deep as you want to go. But there's a few rules on that and I'm going to go through that with you in a minute and then that is golden, right? Because this is so important. I'm also going to feed it the text so it has the plain text output, not the HTML output, so it can read that page quite clearly. And what I'm going to do is I'm going to add a message which is another user prompt. And this user prompt is going to say, tell me about this website. What would we look for specifically achievements or case studies, let's say. But searching it for specifically achievements or case studies and then I want the output to be JSON format. So user strictly JSON formatting format as per below. Now I have a copy of a JSON on my other screen that I'm going to put inside of here. But what I'm going to do is I'm going to show you how easy this is. No code. Don't learn to code. Forget that you don't need it. This is powerful enough. Create me a JSON prompt that contains the following variables. About 1, 1 line. Ice breaker trying to camel code at the same time. And what was the final piece? Website. Website updated. So website updated. Okay. Website updated. If I fire this off, this is going to chuck me back. Something May math examples. May not. No, it doesn't. That is perfect. That's exactly what I've written. You have to write one prompt. You can literally ask it. You don't need it. You copy and paste that. Some gold that I'm about to give you here actually in regards to JSON is this is incredibly important. If you struggle with JSON prompts and it's all a bit overwhelming. It was for me a couple of years ago, right at the start, you know, when you get into that sort of process because you will learn naturally how to be able to edit things and spot certain errors. It just sort of comes naturally. It's part of the project. It's like running anything. You do it more. You get better at it. I'm going to showcase to you now a tool which allows you to validate Jason to make sure the output from ChatGPT because AI can make mistakes is correct. Now in this particular case, it's very short, so you could spot if it's well formatted you could spot it. Jsonlint.com if you don't have this bookmark, bookmark it. Now go to it and bookmark it if you're going to be a regular user of ChatGPT and specifically JSON outputs. If I validate, it's already validated. But if I go in here, let's just say that I remove this comma and break it and I hit validate, it'll tell me exactly what the error is. It'll tell me exactly what it's expecting. It's expecting a comma and it got a string. If I put that comma back in validate. Done. Now this is a really important tool that I use when I'm prompt engineering. You could just chuck it back in chat GPT and say that this was the error or you know, this didn't work. But you save yourself the headache because when you have a lot large JSON you won't really use that many large JSON outputs. But you know, if you do then and the times when you do, you want to make sure that the format is on and I'm not a developer and I don't have any interest in learning how to be one. So that's. That will save you a stupendous amount of time. Highly recommend. Jason Lynn so we tell it to do that. Tell me about this website specifically achievements or case studies. Use strictly the JSON format below. Keep the about information relatively condensed. Condensed strictly no more than 50 words. Let's just say that. Okay, so we now have. We now have a few prompts and we're in a position where, well, we have a couple of prompts, we have a few, but one of them is the developer and system. Let's save that and hit run. Don't do this to me. We're going to hit this and we're going to hit this and run it. And when we run it, what we're going to do is we're going to gauge the output. Remember we're scraping the Reporting Hub so this is going to give us this information. So the output was about. The Reporting Hub is a no code white label business intelligence platform. Gives you a good overview. One line Icebreaker. I noticed the Reporting Hub is featured by Microsoft as a power BI embedded accelerator, highlighting its success in rapid, secure and customizable BI deployments globally. You know that Icebreaker's pretty ass. You won't use that in any of your flows but you can you get into now the understanding of how we're doing this and you can See that the website is updated. So website updated 2024. Now we have the output. That's. That's correct. I mean, do we want to, you know, do we want to give it an example? Let's maybe just tweak this ever so slightly. I'm not looking for perfection here in terms of this Add message. User, please ensure a break. Please produce a two line and a cold email Icebreaker that has two individual sentences. That has two individual sentences. That is spartan yet friendly in nature. Be quirky. Be. No, be. Quirky is not the right word. Be punchy and eye catching. Save. Let's run it again. Let's see what the prompt puts out. So now we see. Loved your global approach to power bi delivery. Impressed. The reporting was handpicked by Microsoft as a solution accelerator. Boom. There's nothing wrong with that. That's a decent enough. You want to tailor it if you were going to do a high scale amount. There is absolutely nothing wrong with that. So now we have the information, we've got the information from the website. How do we use this? Well, I'm first going to show you how we would add this to a Google sheet. You can use airtable, anything, exactly the same. You simply add a row. When you add this row, I'm going to just pause one second, right? And I'm going to go back and I'm just going to cover something. And what I'm going to cover is something that is extremely important about that prompt of what I've just done. Okay. Because there is a strategic element to, there's a strategic element to this. When you're prompt engineering like we've just done. You saw the quality increase in the one line icebreaker, two line icebreaker, whatever. Like as it, as it, as it happened. Now the thing is, is that this is a rule to live by when you're using prompt engineering in OpenAI 1 few, many. Inside of that module, we had two individual prompts that we use the prompts. We have the developer system telling him that it's a website scraper, an intelligent website scraper, because you don't want the robots coming after you. An intelligent website scraper. We had a user prompt that explained about the JSON and we had the other user prompt which was the data that came from the text parser. Now, I want you to live by this rule because I do. And a lot of people who are successful in prompt engineering do this. When you think of a graph, okay. And you think of, you know, sort of the amounts of, you know, the quality of the outcome and the, and the, the prompts. The number of prompts. Just really rough graph, right. If you have no prompts at all, the success. That's it. Like that obviously. Well, this needs to be quality. Whatever you get in the picture, the quality of said response would be like this. Okay. Or even come from the. Come from the top. But this is a good use case. An example. I'll do it this way. I don't need to be an expert in drawing graphs to showcase to you what I'm. What I'm about to do. That is zero with zero prompts. It will just. It wouldn't even increase. You know, it wouldn't even. It really wouldn't even increase. It would just be a flat line. It would be. We saw that when we did it. Like, you know, when we were running the different things. We saw the quality of the output. If you use few, the increases here. Sorry. The difference between. The difference between one, the difference between one and none is massive. Okay. It's quite large in nature. Maybe the graph won't do it justice. Maybe I'll have to extend it to show my, my point here. But when you're looking at this, the gap is massive between one prompt and zero is huge. If we look at few, try and color over this. It's been a while since I did any coloring in few has a big difference between that and one prompt. So the gap again gets larger. But for many of. I run out of colors. Kind of run out of colors. Yeah, let's just use green again. For many, the difference between a few prompts and many is like this. So many prompts is better, but the difference in size between these here, that there and these is enormous. So the sweet spot as a prompt engineer is right in there. When prompts engineering, you need to live by the one fewer many rule. If you're using OpenAI module inside of make.com and you are trying to engineer a prompt to do a specific task, cold email output, search, scraping of websites, information. One output is a significantly, statistically significantly better output than if you had none at all. A few outputs has a huge difference and increase again in comparison to one prompt. However, many will be the best output you can possibly get. But many versus time that it takes for you to prompt engineer. The gap between a few good high quality prompts and many high quality prompts is relatively marginal. And you should stick with using a few prompts at a time. Between three and six, seven prompts maximum. When you're creating. You should be. You should be. You should have the ability to be able to produce prompts an efficient rate that allows you to be efficient in flow building. Imagine if you had to go out there and build 50 examples for every single AI agent prompt that you had in a large flow. It would be absolutely impossible. So I hope that you understand that. Let's get straight back into the Google Sheets module. So what you want to do is you want to add a Google sheet. You select Google Sheets, you we want to add a row because we want to take this information and add it to a Google sheet. Just let's say now you can search by path for a sheet that you've created. If you don't have a sheet created, literally just hit sheets new and it will create you a new sheet and you can populate that from there. You're going to want to give it the headers that I put in, but I have open the wrong tab then close the other one. I have this created already and my the headers are horrendous. But about Icebreaker site updated that's the way that we that is integrated. Now the reason that I used the because I did that information there about the step I've missed a module. I'm going to delete this just to showcase this to you in the correct order. We need a module here that parses JSON because the format is JSON output that we've just received. Now technically and for operational purposes, all of you more experienced make.com engineers are going to say hey Mike, we know that you can go inside of the OpenAI module and you can go to advanced settings and you can and you can have the output is it even in advanced sentence where is it there response format and have it as a JSON object and have it automatically output JSON. Yes, I'm aware of but I like to keep my flows consistent and smooth and I just do them this way. And for example, in learning purposes of a use case of this video, I think that adding the parse response is a good way for people to learn and understand. So we're going to parse the JSON. Now the JSON string comes from the result of the OpenAI module that we've just ran because it has all the information in it and we formatted it for JSON purposes. So now we've done that, it's going to give me the error again, but we're going to run it anyway and we're going to have a look at the output. About the Reporting Hub, one line Icebreaker impressed that the reporting hub is A Microsoft recommended Power BI trusted across six continents. Love how you simplify analytics deployment with no user limits or coding required. That's pretty good. Tells you the website hasn't been updated since June last year. That's over a year. Yeah, definitely over a year now as of today. So we're in a position where we have extracted this information into adjacent format and now we can add the Google Sheet. So we go into here, search for Google Sheets, go into Google Sheets, add a row as I previously just showed you. Now you can find your sheet here. I'm. I'm fussy. I don't know. I have a thing. I have to enter it manually. So I'm going to go to my integration sheet and I'm going to get the id. The ID is everything in between edit and the. And the letter D there. The slash in between the slashes before those two things. You copy that. You put that inside the id. Importantly here though, it asks for the sheet name. You often rename this. This is sheet one because it's an example. But you'll need to put the exact name with the correct capitalization inside of it. There's and column range. We want A to Z. Okay. So very clearly now, all of a sudden all of the values have become available and visible to us. We're now in a position where. We're now in a position where we can update these values and map them out. And what we're going to use is we're going to use the JSON that we just passed in the previous module in order to do so. That's why I want to take you through that step so you understand the data flow and how it's actually going to proceed. When we look at this individual, I believe it was about Icebreaker website updated. That was my columns. This is the way that it's going to be. Remember, if you hover over it and it flashes, you can see it sort of blinks. You can see the module that the information is coming from. Just trying to drive that home for any new to make.comdevelopers/. You know, a developer, no coders and AI specialists. This really isn't rocket science. Trust me. I am no superhuman. I'm an entrepreneur. I own six businesses. I'm an AI specialist and deploy this consistently. But when I started, I had no idea about any of this. So I will, I'll go into. I can go as in depth as you like and build you the most complex flow ever. But I will also point out obvious and you know, things that I think that Newbies would. Would need support things that I needed when I originally started. So if you think I'm repeating myself, it's intentional. So now we've linked this Google sheet and put that together. We've now mapped out them in the correct order that my list is given the headers in. So we're going to run this module. It's going to spit this out and it's going to give the updated sheets. It updated three columns. We go into the. To the data. Let's make it unbold. The reporting up gives you the about section. That gives you the icebreaker. Didn't pull through. The site updated for some reason. Interesting. Okay, very weird. Anyway, doesn't really matter in that individual run. It didn't pull it through. What's the. What's the variable? What's the JSON format? Yeah, I mean to pull that from that, it's going to be. It's going to be tough. It'll be the formatting of it. It doesn't make any difference. Using the cold email line break is. Is. Is. Is the most important thing. That's exactly what I was trying to showcase to you here. I've got bigger fish to fry than worrying about that why that variable didn't get pulled through. We'll see if it does it for the. For the other businesses as we go through and if not, maybe I'll debug it at the end. But what you can see now, what you. What you should be able to see, what you should be able to clearly see as a. As a business owner is the value in that said flow. No, you might not want the about the. About information. You might not want the icebreaker. You might not want the date the site is updated. It's a good job. Cause we didn't get it anyway. But you might not want the site updated. You might not want these things, but you can format this to bring anything, any single piece of information that you physically possibly want in your industry. You can use these sorts of flows to scrape YouTube URLs for updates and give you information. It's limitless. The, the output with this sort of flow and scraping sizes is limitless. And it's. Especially if you've done cold email before and you're watching this video, you know exactly what I'm talking about. Because you will use something like App Fight to scrape Apollo because it's cheaper, which is a whole separate another video. But jumping back into this now, this is fantastic. We have it for the reporting hub, but that's not going to make you any money, is it? The flow can be sold. Don't get me wrong, because this is not, this is not something that everybody is capable of doing. But we have the flow now. We have a, we make a request to a website, it converts that text. We've created, we've prompt engineered it, we've passed it and we've passed it into the sheet. That's been done. Now what we're actually going to do is we're going to use some test data which you saw a snippet of earlier when I accidentally opened the wrong tab and then closed deleted the other one and had to reopen it. But we're going to have a look at implementing this and pulling data from a Google sheet. So we're going to have it search rows. You could have it watch new rows. It sees when a row is added, but we want it to search rows because they're already there. Okay, now you've got to move the starting point over because our starting point is now going to be the Google sheet and now we need it connecting. Okay, so you, we do that. We move this trigger over. We, we get into this position and now we need to map this, this sheet. Okay, now I've just explained to you how, how I sort of approach it. So I will get the spreadsheet id. Okay, so I will, I will map every spreadsheet that I, that I use. This particular URL, this is a scrape of like marketing agencies, I think. And it was, it was, I think it was international. Yeah, I think it was international. Not 100% sure. But you take the ID number that was in between the, the, the letter D after the slash and the edit. You go into here, you add that, you add the sheet name. I'm going to type sheet one because I'm pretty sure it's, I left it as sheet one. Yeah. Sheet one to match up, it now pulls. That table contains headers. Yes, it does. Okay. Yes it does. And the column range save. Now we're now watching this, this row, but there are a few individual things that we need to do because if you think back to the start of this flow, as we've started out in here, we put the reporting hub. If I run this flow right now, regardless of the fact that I've got this sheet connected, it will spit out exactly this. Well, it won't spit out exactly the same, but it will spit out the same information around the reporting hub in this module with the HTTP request. What we now want is to run this module only and pull all of that information. Now, just looking at that screen, which I'll let you look through for a moment, that is the screen of what's been scraped. This contains a stupendous amount of value. Names, email, obviously I've scraped this from elsewhere. Apollo, I've scraped this from. So if you wanted to go and have a look about Apollo, use apify to scrape Apollo titles, business names, position marketing, advertising company LinkedIn's personal LinkedIn. If you want to scrape for posts using App Fight, you could scrape for his latest post and comment on it in your online icebreaker. You literally can do any, anything that you want. I love this section though, because if you're trying to sell AI automation solutions or anything like that, you can see they use Active Campaign, you can see that they use HubSpot, you can see that they use Microsoft 365. This, I'm not going to use it in this particular flow, but you know, you could say your tech stack includes this, this, this and this. You know, I've built. You can go into that and you can sort of tap into, into that. The point I'm making is that if you cannot see the value of what I have just shown you in terms of how you can manipulate that data in any avenue, not just how I'm showing you, but in any flow that you want to create, then you know, make.com is not for you. Like no code isn't for you. The value inside of this is absolutely critical and I'm going to show you how we implement it into the flow right now in a few, in a few different areas. That noise is coming from in a few different areas. So we have this. So website is, is column E. So we look here, you can read it website, they're well named because of the scrape. What that now means is instead of pulling the reporting hub, we don't need to update anything else, but it's going to pull the, the website from column E. That's what it's going to do. So we save that as a variable. So maybe now you're watching and you're thinking, wow, well, actually we've got a huge amount more data than what we had originally. How else can we implement that? Well, what we can do is we can do simple things inside of here. Say, please produce a cold email icebreaker that has two individual sentences that is spartan and yet friendly in nature. Be punchy and eye catching. In the second sentence, please start, please ensure you use first name. So now in the second sentence of every icebreaker, it's going to put in. It's going to use the individual user's first name. We can also founded congratulations on specific years. I mean Colorado. You know, let's just say this. I'll do this for you. Okay, let's just do this. In the second sentence, please ensure you use first name. In the first sentence, please ensure. Please find a five star. Find the best five star restaurant in in city and comment on a specific item on their menu. Okay, so you can see now the power that we've got in this. Hopefully this works. If it doesn't, we'll fix it. But we're now going to run this flow. I hope I didn't mess up with the columns in that search row. See how it's running once. It's ran once. But now it's because there was three data sets inside of this example data lead sheet. It's ran through three individual times. It will run that. So it'll increase the operations. That's why some people are fussy about the operation numbers. If you were to run this 10,000, 100,000 times, you'd be using, you know, a lot of additional operations for the passing module that is complete. Let's check the output. We go into here. Updates. Yeah, yeah. It's not going to give me the values, is it? A perfect, perfect search media is an award winning digital agency. And then his black truffle explosion is a revelation. Trent, you've led your team to an impressive two impressive digital wins. Let's talk about doubling down. Updated 2004 so let's have a look in the. In the sheet. If you look at these three and you go through that individually, I hear Chamberlain Steak and Chop House in Carrollton as affiliate Oscar. That's out of this world. Ryan, congrats on your strong digital results. I'm impressed by by what your team is accomplishing. Sketching in soho as a legendary truffle risotto. That's making waves for a reason. Shanto, your team's bold digital. These aren't ideal. These aren't excellent. I wanted to show you that as a use case as to how you would use the data into the icebreaker and what you can pull out. So if we do this, I'll do this one more time inside of here and see what the output is. It doesn't really matter at this stage. We. I've taught you how to build the exact fly but what is important is getting something reasonable put out. So tell me about this website. So if. Let's just see if I stick in here. Never use hyphens in outputs. Just stick it in there. Hopefully that works. It's not the ideal way of doing it. You'd put it in here. What was, what were we doing? We were editing this, weren't we? So in the first sentence, please bet. Let's just take this out. Let's just take this out. Okay. And let's just run it now. Let me go to the sheet. Actually no, I'll keep it because I want it to be stack up against one another to be able to see it. So it's now running. It will use the first name so it personalizes the email. You will have a cold email template. If you're cold emailing that will have hey variable instantly IO is fantastic for that. Instantly. Instantly AI is literally like probably one of the best tools I've ever seen. So if you haven't checked that out, you should. But when you look in, when you're looking into this and you're sort of looking at the outputs, what we're looking for, you see, you see how much of an improvement that is just taking that out. This is critical for you to understand because it showcases you the power of how you would actually use things like restaurants. I just chose restaurants and it did it. It picked different things off the menus. It searched them out there. If you went and had a look for Chamberlain's, I know what sketches in Soho from London already but if you went to have a look at Chamberlain steak and shop house in Carlton, I bet the reviews are excellent. And it's a, it's a place that is popular which means that individual that's based in said location from the data that we extracted because obviously from extraction there's a lot more data that's available to us than what we started with at the flow. Started with at the start with the flow. If you look at those, they are relatively poor in terms of output because it starts with the, the the restaurant which was by design from the way that I told it to. But if you look at this impressive to see your client results hit 280% of their marketing goals. Ryan, that leap is the kind of growth every marketer wants to brag about. Impressive case studies like UX design for Tubus and SMM projects caught my attention. Shanto, your team's results driven spirit stands out. Impressed by a 47% lift in paid media conversions and those 30 plus recent awards. Trent, your team's track record is exactly what caught my eye. Those sorts of outputs are things that you can use in terms of your in terms of your. Your cold email marketing. Now, as I wrap up this video, I want to thank you for hearing me out. I hope that you've learned about the individual flows. The template will be in the, in the description below so that you can take this away and run with it. Make sure you document the tools that I have used throughout this process because it is extremely easy. And what I have shown you is the skin of the surface of what can be done with those variables. If you go into detail and you practice your prompt engineering and you understand we've just had an output in terms of the last 40 minutes, 45 minutes. An output that actually gives you credible, usable, like decent response rate icebreakers. In terms of emails from the flow that we've just built in what I could have built and what you will now be able to build after this in the space of 5, 10 minutes, you can have those pumped out. What you need to then connect is the way that you scrape leads and the way that you accomplish that to get the information like the examples that I had together. My name is Michael Elliot. Thank you for watching this video. Catch up with you soon. Cheers. Bye.