114 | AI Lead Generation: Transforming LinkedIn into Business Opportunities with Jeremy Grandillon

0:00Hello, and welcome to another live episode of the Leveraging AI podcast. I am so excited to have you today. We're here with a very important topic and with an amazing guest. And so I will start with why is this an important topic? So the first thing we need as a business is clients. Like without clients, you don't have a business. You can have a nice hobby. It could be fun and you can do a lot of interesting things, but you cannot make money. To make money, you actually need clients who are going to pay you. To have clients, you need to find your target audience. You need to find the relevant people.

0:34You need to have a conversation with them. And the more effective you are in doing that, regardless of what your industry and how big your company is, the more money you're going to make. And that's why it's a very critical topic, because literally the livelihood of your business depends on this as the very first thing. Like you can run a company without legal. You can run a company without marketing. You can run a company without finance. You can do it. It's not great, but you can do it. You cannot run a company without clients. And I guess today, Jeremy, is an absolute ninja when it comes to figuring out how to find the relevant people, specifically on LinkedIn, and I'm sure on other platforms, but we're going to focus today on LinkedIn.

1:17How to use AI to automate the process of finding the right target audience and finding the relevant people and having meaningful conversations with them so you can turn them into clients. Now, if that's not important to you, you probably, again, don't agree with what I said before, which would be surprising. And so I'm really excited about the topic. I'm really excited about our guest. Jeremy, welcome to Leveraging AI.

1:48In the next few years, AI technology will change our world dramatically. Whether you are a business executive trying to catapult your business forward, or just somebody who refuses to be left behind and want to advance your career, This is the show for you. I'm your host, Isar Maitis, a serial entrepreneur and an AI enthusiast. You'll hear invaluable practical tips from innovative business leaders, AI practitioners, and some of the brightest AI minds in our world today on how you can leverage AI in ethical ways to advance your career and grow your business.

2:28Thank you very much. I'm very excited to be here with you. I'm not in Hawaii, but I'm in Paris, which is quite cool too. Not bad, not bad. Getting ready for the Olympics? Yeah, yeah, yeah. We are afraid of the mess, but yeah, that would be fun. I'm sure it's going to be hectic, but it's going to be probably fun, fun hectic. I've never been to the Olympics. So first of all, for all of you who are joining us, whether you're with us on Zoom or LinkedIn, I'm really happy to see all of you. As a first thing, wherever you are, please let us know where you're joining us from. in the world. That will be very cool because we have now two people, one in Hawaii and one in Paris.

3:07So it'll be interesting to see where everybody else is. But then Jeremy, let's really get started with a very high level question and then we'll dive into this. I know you're doing this for yourself as well as for multiple businesses. How did you figure it out? Like how long, like AI started like showing up about a year ago, a year and a half if you were really early into the game. And then now you're already, again like an online ninja teaching people how to do this how did you figure it all out i failed a lot like a lot that's that's the secret that's the secret actually i i own an agency right now so yeah as you said i'm doing this for me and for my clients and we'd be more than happy to develop on what we do but in the last five years i was a sales person myself in the last three years i was a head of sales at a software company and actually i started this entire sales department from scratch in this software company.

4:02And yeah, I had to find a way to get clients. Like you just said, it's quite crucial for a business. So I tried almost everything and failed a lot. That's not a joke. And based on that, I've learned and I've seen some things that works better than others. And at the end of 2022, Chagipity came, it's a revolution in my face like everybody else. and I thought that it might be something interesting to implement in my sales processes. So again, I tried a lot of things and this is how I've learned. And at the same time, I've started to share what I've learned on LinkedIn. And here we are doing this.

4:44Okay. Sounds amazing. I'll ask a quick follow-up question. Or you know what? I won't ask a question. I will state something that is very important. That's the way to learn, right? So a lot of people are afraid getting started with AI. what if it's not going to work? What am I going to do? The only way you're going to figure this out, even if you follow people like Jeremy, you follow people like me, you follow other people, the only way you're going to actually learn is getting your hands dirty and trying stuff out. And it's not going to work the first time or the second or the third. But on the 20th time, you're going to create magic, which is going to save you hours and weeks and going to throw your business into an upward efficiency curve that you've never seen before.

5:25So it's really get started in into your process and I will let you just walk us through step by step on what do you do how do you do it which tool you use so the stage is yours I'll just ask hopefully good questions okay sounds great yeah by the way totally agree you have to try and what we do here what we are doing here like showing you stuff will just make it happen more quickly you will succeed more quickly because you will have some information but you will still have to fail yourself so so yeah totally agree with you so yeah let's put some context here how to get clients big question um let's start with what you shouldn't do i love to say it you shouldn't use the spray and pray approach it worked like i don't know six or seven years ago when nobody was sending emails when you received an email at this time you were like oh interesting a generic wall text oh interesting now somebody want me an email yeah exactly and it's not like someone i know wow amazing today you receive another generic email you're like oh come on and you just delete it at best or you put it in spam so it doesn't work anymore so what you need to do is to be absolutely relevant straight to the point and if you can you personalize your approach And to do that, it requires actually a lot of time.

6:54You have to do your research. You have to, if I want to reach out to you, I need to know what you are doing, where you are based, what your business is doing, what's your offer, what kind of problem you're solving for your clients, what kind of problem you're facing, et cetera, et cetera. A lot of information. And also, I'm adding to this huge amount of information that I need to find on you. I'm adding some intense signals because not only you need to be relevant, personalized, but you also need to be here at the right, in the right timing, at the right moment. Because let's say I'm, let's say if I'm selling some talent acquisition services and I reach out to you and you don't hire, you don't have to hire anybody at the moment.

7:37You will just say, hey, interesting, but not right now. And you waste your time basically. So it's not wasted because you have a contact. That's positive. but you won't have a deal from this email, for example. So we had more information in 10 signals. So that's a lot of things that you need to find. And if you do it manually, it requires an entire team of, I don't know, 10 % doing this, which is basically not the best activity to do as a professional. I've done that, I know. Like looking, searching on LinkedIn or Google all day long, it's a lot. It's a bit too much. At the end of the day, you are like, okay, let's have a drink of wine, of course, in Paris.

8:18Of course. So the good news is that today, so first, the first level of things was automation tools. So that came five or six years ago. Automation came to us, helping us doing more work without investing more effort. And the next step, which is today, was AI. And AI can not only automate things, but you can add some intelligence in your processes. And this is very interesting because let's say if I want to, I'm taking the example I was giving you just before. If I want to reach out to someone who is hiring right now, I need to search on LinkedIn. Is there any open position for this company, for example, or another job board to see if there is a position.

9:01The good news is that AI agents can do that for us now. So you can have high quality information at scale without hiring dozens of salesperson. I want to add one small thing to what you said as far as the intelligence and old school automation versus automation with AI. And the way I like to describe it when I teach my courses is there are two, there's old school automation, right? Zapier, Make, NA10, Pick Your Poison, right? There's a lot of... Absolutely.

9:35ways to do that. And when you do old automation, then it helps you build processes if it's a tedious process that is the same every time, meaning it does not really require any human input. I want to search through this list, find anybody who has four stars or more, save their name, copy this, put it in Excel table. Yes, it will work. You can do this every time, all day long. It was possible again, six, seven years ago. Every time you want to add a human input, meaning I want to filter this by some kind of a thinking process. I want to send them a message. I want to create a new text. I want to respond to something.

10:13I want to do anything that is not a simple what if based on the existing data, which means it requires some human input. AI knows how to do very well. Meaning if you're not asking it to rewrite your strategy, then that would be probably too much input you're asking for it. But if it's small human inputs that are required, then AI does a perfect job of complementing the old school automation stuff. Absolutely. I totally agree. And what's interesting in the tools that we have today is that you can split the process in many small tasks. And by doing this, you will only ask one very simple question to the AI.

10:58which will reduce the risk of DI just hallucinating or doing the wrong thing. If you do your prompt quite well, I'm not a prompt engineer expert, but I'm okay and I understand how to use prompt engineering. So the result would be quite consistent every time. So that's also very interesting and very important. So I have prepared a small use case. so everything that I've described it applies to cold outreach for example cold emailing and it still just to say it, it still works you can see a lot of people on LinkedIn say cold emailing doesn't work anymore, I think they just don't do it right they suck at it, that's why they are saying this if you do it right and if you use AI to do your research it will work that said, there is another level of things that you can do and this will involve LinkedIn and content, as you mentioned earlier, which is warm outreach, which is even better.

11:58And which is the reason I've called my agency outbound, which is outbound, like cold, but also outbound, warm, and inbound because we are creating content on LinkedIn to generate this awareness, to generate this authority, this credibility, and also inbound leads. That's one thing. But also when you reach out to someone that has engaged with you in some way, it's warm leads at this moment. So we will emphasize on this today so I can show you something that I've prepared for you. Okay. And let's set the problem that I have. Let's say I have created content. That's the first step. I've had quite a successful post, like 100 likes and comments.

12:45That's pretty good. Cool. probably because i'm talking about something that is relevant to my audience probably in this people like in these people liking my my post some of them would be interested in my offer but i i can't i don't want to manually go through 100 people liking the thing open their profile look at if they are matching my icp because remember i said you have to be you have to target these people. You can't spray and pray. We want to be relevant. So if I'm offering talent acquisition services, I will target maybe people who are hiring and I don't want to do the research manually.

13:24So they are wrong, but maybe not my target. So the use case that I've prepared is using AI and automation to just scrape a list, like the Android likes, put this in. The tool will do the filtering for me, meaning it will go step by step, verify all my criteria and say, okay, this person is matching your ICP. You can now reach out to him or to her. And I will just show you this. So let me show you my screen. Let's do that. Okay. So you should be able to see my LinkedIn post of today where I'm actually describing this exact process. Can you see it? Yeah. Yeah. Okay. By the way, for those of you who are listening to the podcast and not watching this, A, come join us.

14:15We do this almost every Thursday at noon Eastern time. So you're welcome to join us either on LinkedIn or on the Zoom with all the really cool people. I'm not taking anything away from those who are joining us on LinkedIn. I appreciate every one of you, but if you want, all of you, if you want to be able to chat in the chat and see what's happening, come join us on the Zoom calls as well. And if you are driving your car right now and it's not Thursday and you can't join us, then we're going to describe everything that's on the screen. You'll be able to understand what we're talking about. So sorry, let's continue.

14:45No problem. Of course. I have created this four step process and I will describe every one of them. So the first one is data scraping. As I mentioned, we want to extract this profile, the person that are liking this post because I don't want to be stuck inside LinkedIn and do it manually. So I need to extract this into an Excel file, for example. Most of the time, it will be a CSV file. And this file, I will be able to import it in the next tool. Note that you can automate this. You can, for example, create a Zap with Zapier to use, I don't know, Phantom Buster is one of the data scraping tool.

15:24And you can directly push it into a clay table. So that can be done on autopilot. In that case, it's an example, so I have done it manually. So for the data scraping, I've listed four tools. Some are advanced, some are free. So that's for you to have the choice to match your needs. So Fontombuster is one, Person.ai is one, Epify is another. And there is this little Chrome extension by Folk, which does it for free. It doesn't work all the time, but it's free. So it's pretty good to start. Second step. Quick question. Because I'm sure you've tried all of them. Which one works best for this use case of just scraping data from LinkedIn consistently in an effective way?

16:09I would pick Phantom Buster. Yeah, me too. Okay, we're on the same page. Yeah. By the way, for those of you who don't know Phantom Buster, it's an amazing automation tool with really advanced capabilities. Only disadvantage is cost. I assume it's more expensive than all the other ones. but if you want to do stuff at scale and with very detailed processes there's nothing that competes with it yeah absolutely and as i mentioned it offers like api keys to put it into automation and push all of the process and the workflow automatically so yeah it's more advanced it's more expensive but it's also better people are asking for the chrome extension can you say the name again for the people who are listening to us you can go in the chrome store on google and search for the folk look f-o-l-k which is a crm and they have developed this additional extension that allow you to extract leads from from linkedin directly and let me show you if it works oh yeah okay it doesn't matter it's okay just people wanted a specific name yeah yeah so So you will have this here.

17:20OK. And if you click on this, you will be able to scrape. OK, so next step, now you have your list of people who have liked or commented your posts. So they might be interested in you. This is a signal. Next, you want to filter them to make sure they are matching your ICP because you don't want to reach out to anybody. So to do that, we will today use Clay, which I think is one of the most advanced tool. Same thing. is a bit expensive, but it's also the best. So if you want to do it at scale, at scale, I recommend using this one. Persona is doing quite well. Redevents has a cheaper offer at$20 per month, which will allow you to do this also.

18:00It's not exactly the same way, but it provides the same results. So I wanted to offer you different possibilities here. For those of you who do not know, Clay is also a good scraper, just not for the LinkedIn use case. Like you can scrape data with clay very effectively. On LinkedIn, it's just not working that great. Yeah, exactly. You can scrape stuff from the web, which works quite well. But from LinkedIn, it's not that good. Yeah. Absolutely. And then after the clay filtering, we will use, you can use your favorite outreach automation tool. I'm using Lemlist. I will show you how it looks. Reply is quite good too.

18:40And there are others, of course. and my personal touch is the is break cold which is my crm and this allows to continue the conversation because the whole goal of this strategy is to start conversations but you don't want to at least for now i don't want chat gpt to answer and make the conversation entirely i just want to start conversation at scale and then i'm driving them and i'm continuing the conversation from Breckhold, which allows me to manage the CRM part of things too. Yeah. Okay. Awesome. I'll show you. Let's dive in. Let's see step by step. Okay. So I've prepared this and we will try with your profile, Isa, if you want.

19:23So for those who don't know Clay, it looks like this. It's like a table, like an Excel table, but inside the website, inside your browser. And you can add rows and you can add columns. And in the columns, so basically the first column here will be the trigger. There will be the starting point of everything that comes after on the right. So my source is the LinkedIn profile as this is exactly what I will scrape from the people who are liking the post. And so this is basically the LinkedIn profile URL from the people who liked. Let's do all the columns one by one. So the first thing that I need is to extract the name of the person, full name, name and last name, first name and last name.

20:12So there are different columns that you can do with Clay. Some are built inside the tool. Some, for example, this one, and it will extract some information based with their own technology. And some others, you will, for example, this one, you will just connect an external tool to your workflow. So this one, enrich person. So basically what it does, it's based on the URL. It will extract from LinkedIn directly the name and the first name and the last name. So that's my very first thing, my very first step. The second step is I have created this file that I've named. So it's for those who can't see it's a Google sheet that I've named already script because I don't want, and I think I made a mistake here.

21:02There is only one P right. oh maybe I won't do this because it will it will it will no I think it's fine okay it will trigger the thing to run again yeah okay so let's keep this mistake no problem so what why I've created this is because the way it works when you pay for you're paying for clay you will have credits credits that you can spend to do tasks and using AI and to use AI so to save some credits you can use some steps that are free. And for example, Google connection to Google Sheet, it's free. So you can look up to a file. You can add rows. And I think you can also delete rows. So the first thing that is I have put these two columns.

21:50The first one is look up to this file and tell me if this name, full name, is already in the file. The reason is that if it is, that That means that I've already scraped this person. This person liked at least two posts that I have wanted to use for my warm outreach. So I won't put this person in the entire process again. So I will just ignore everyone that has been already scraped. And if it's not in the file, then I will add it. So next time it will be recognized as, okay, do not continue the process for this person. We have already done this. then I'm doing the same thing with another file which I named the blacklist and that can be inside this this file that can be like my clients I don't want to reach out to my clients that would be weird that can be my own team that can be my friends that I know that I'm not interested etc you can note also if you want to do it more professionally than this example you can connect actually your CRM.

22:57So I think there are connectors already existing with the most common CRMs like UpSpot, Salesforce, etc. If your CRM doesn't have a connector, you can use the APIs. Most of the tools, modern tools offers API and you can just build it. It requires a bit more skills, but as we said earlier, you will fail at five times and the six would be the good one. So you can replace these files with your CRM, which could be more streamlined. OK, so next step, based on that, yeah, I want to show you something here. So in this column, for example, this step will only happen if it matches this weird formula. So obviously, I didn't write this myself.

23:49I used AI here. And what you can do is describe the condition that you want to do. So in my case here, I want this column to run only if the lookup is not empty and false and the blacklist is false too, which means that you will run, I'm telling to the tool, you will run this column only if this person has not been scraped before and is not in the blacklist. So this is also one way to save credits. So again, to explain to people what we're doing from the beginning, We started with scraping people's profiles based on some criteria on LinkedIn using Phantom Buster. This populates the first column of a table.

24:34And then all the following columns, the first two columns were filters. Don't do this. Go through this file, go through this file. And then the next column is saying, okay, don't continue to the next column, which is how this tool works, before you check if the other two columns are, the lookups with these lists are false. And this, again, is done A, to save time and B, to save a lot of money. Absolutely. It can be expensive if you don't optimize your workflow. So yeah, it's important to do it since the beginning. So you save money. Exactly. Okay. Let's just a quick warning. We have about 10 minutes, so we need to move a little faster.

25:08Okay. Let's do it faster then. So the next step, if it's much, so it's not already scraped and not in the blacklist, I will start enriching and ask questions. So the questions will be done by the AI. and if the answer is the one that I want, then it continues. If not, it stops. And at this moment... Give me an example for a question. Yeah, so the first one is I want to filter people based on their location. Let's say, so in this example, I will show you the prompt. So quick note, you can, when you use AI in Clay, you can select the model that you want to use. So in this case, I use GPT-4O, which costs credits.

25:50You can select 3.5 Turbo, for example, or Cloud, or their own model. The price changed, but I would recommend for very simple tasks, you can use 3.5 Turbo for more advanced, for always better, because you don't want the tool to say, I don't know. So yeah, the prompt here is you are a geography expert. I will provide locations and you will reply with true or false only. you will try you will reply through if the provided location is in north america europe blah blah blah so i'm adding this that i'm that i want to target this location and nothing else and here is the location and here you can see the location is actually a variable which is my previous column based on the enrichment that i've done on the with the linkedin profile so this location comes from linkedin and if it's much my question that the tool with chat GPT will do, will answer, then it continues.

26:49And then I've done that with multiple criteria because, okay, I want to target these locations only. So if it's in France, for example, I've excluded France because I don't want to work in France with this offer. So I'm excluding France. So if someone comes from France, it will stop at this step. Yes, at this step. Then it will continue and it will make a summary of the profile. and their current activity. And their current activity is the next filter that I'm adding. I want to know if their activity, their company is B2B. So same thing here. I have used this time not ChatGPT directly, but a Clay agent, which can visit a website, in that case, the LinkedIn company page.

27:37And based on the information that there is on this page, you can ask a question. In that case, I said, read and analyze the about section. And based on what you learned from it, tell me if this is B2B or B2C. So if it's B2C, I will stop the process here because I don't want to work with companies that are working in B2C because my offer is for B2B companies. Same, I've added this one. So I hope it works because I've added this just one hour before. So same here. I want to know if they are working in an industry, which is not the same as mine, because I don't want to send an offer to a competitor or a partner.

28:19I don't want them to be in the lead generation or automation industry. So exclude this and tell me if not. So in that case, for example, Brigitta is working as an agency like mine. So it's false. Cool. It stops here. then i will do the same on the employee headcount because i know that my targets are smbs so if it's like 10 000 it will be too much and yeah also i want to talk only to the c level of the companies i don't want to talk to anybody else because i need to target the decision makers so yeah same thing based on the information step by step i will say okay this time you are an hr expert based on the title, you will tell me if this is a C-level or not.

29:07And the final step is adding a row in a filtered another file, which and I will show you with your profile, as I will do a test with your profile. So this is the live part of the demo, so of course it won't work. Just a disclaimer. That's what we do live, so things happen. So let's try it. So let me just confirm that you are not. Yeah, you are here. I've tried it before Okay, so you are not in the list and it's saved so it should work just a quick note also you can so for example This one here. No, not this one. Sorry this one I think Yeah, no Second Okay, we'll do it. You will see. Okay, let's try it.

30:06So you can see here it's running. So it's, yeah. So again, for those of you listening, all that he did is, in this case, manually grab my LinkedIn profile, put it on the very first column, which starts the process. And then it's like a waterfall, right? With all these gates that he has defined. And what we see is that it's getting filled up with information as we go to the right. Exactly. So it has extract your name. So far it's good. So far it's good. It says your name. Yeah, I wanted to show you this. Here you can define if it will run automatically or not. So I've stopped it just for, so I have the time to explain what's happening.