88 | Boost business growth with AI infused LinkedIn strategy w/ Celeste Yamile

0:00Hello, and welcome to a live episode of Leveraging AI, the podcast that shares practical ethical ways to leverage AI, to improve efficiency, grow your business, and advance your career. This is Isar Maitis, your host, and we have a special topic for you today. As you probably know, LinkedIn is a business goldmine, but it's only a goldmine like a real goldmine, meaning you need to know how to mine the gold in order to actually benefit from it. Now, when AI started coming out and content creation suddenly became easier for people who are struggling with it, many people jumped in and started creating AI content.

0:36The problem is they're just creating more content instead of developing a strategy based on actual real data from LinkedIn itself to help them create content that will actually convert and drive business. So the problem is the questions you're probably asking is, okay, so what do I post? What days do I post it? What time of the day? What topics will drive the most engagement for my audience? What drives sales from my audience? what people actually click to go to my website, and so on and so forth. All these questions, you don't have to guess. This data actually exists in combination of LinkedIn data plus some other sources.

1:17And if you know how to mine that data, going back to my example, so pun intended, if you know how to mine that data, if you know how to analyze that data, it will tell you everything you need to know. And then you can use AI, not just to analyze the data, but also to create the right topics and post it at the right time, which will drive engagement. Our guest today, Celeste Jamile, is a LinkedIn content guru. Her company, Temporal Labs, helps businesses and individuals create LinkedIn content that converts. And she does it at scale because it's based on actual data. So she takes away the guessing game by taking data from LinkedIn and other sources and using AI to analyze them and then create the content versus let's just create content, brush it on the wall and see what sticks.

2:03Example for what I just said is her latest post from today. So thank you for that. We didn't plan that, but great post today is called AI strategy I used to go from 248 to 27 ,000 views on my videos. So that's obviously doesn't happen by luck. It doesn't happen by just posting videos. You get to these kind of numbers when you have real strategy and real process behind it. And Celeste is going to share with us exactly what's her process. So you can do it as well and have a systemic scientific way to create content that converts on LinkedIn. This is obviously extremely important for anybody who's in business today.

2:45And so I'm really excited and honored to welcome Celeste to the show. Celeste, welcome to Leveraging AI. In 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.

3:33Thank you. That was a really nice introduction, by the way. This is what I do. Yeah, but you're absolutely right. It's not blog. And I share that post sharing my process. That's the way I use AI is to create the strategy and to optimize the content creation. So in that case for the post, I was focusing a lot on the hook and that the idea was valuable. You need to validate your ideas with data, right? And also sharing content that is just strategic according into the goals of your audience and the pain points. But of course, any type of strategy, it's going to be useful if you don't deliver in the content.

4:13So it's like you promise something, you optimize that with data, but we need a lot of practice to be able to master the content because even someone that knows nothing about content but just creates something really valuable can create something that resonates with others. But I'm here today to share how to optimize that. Also for the ones that we are just starting, So we can optimize as much as possible. So in the level, whatever level we are, like if we are beginners, very medium level or advanced, we can increase the results that we already have. So yeah, that's the goal. So I'm going to start sharing this screen.

4:48Yeah, fantastic. So yeah, let's just dive into your process. I think the key here, and that's maybe your biggest and most amazing differentiator is the data side of things. A lot of people know today how to create content with AI and the people who know it better know how to create it to sound like them and so on. But your really unique thing is that you've learned how to harness and harvest the data from LinkedIn itself in order to actually make this a lot more scientific. So let's jump in. And by the way, for those of you who are listening and not watching the screen because you're not on LinkedIn right now, first of all, join us every time we do these lives so you can see this live on LinkedIn.

5:25But if you're not, if you're just listening, we're going to describe everything that's on the screen. So if you are driving or walking your pooch or doing whatever it is that you're doing when you're listening to the podcast, you can still follow with us. Yeah. So the first thing that I'm going to use is chat GPT. And for this, we have this workflow for LinkedIn. So of course we need LinkedIn data. And the way that you can gather is that already gives you all of all the things that you need. So if you go to your profile, right now I'm on my LinkedIn tab, and in the top right corner, I see an icon that says export.

6:03So you need to make sure that you have selected the data of the last year. Of the last year, you'll see a button to make that happen. And also, you start exporting the data, you click the button, and you will have your analytics ready to go. so the next step is that and by the way for those of you who are again are listening it doesn't give you the data right away some of the data it has to compile and then sends you an email with the rest of it so there's some that you can get really quickly and some of it comes a few days later yeah the great news is that for example you can analyze the last 30 like the last year so you can have as much data as possible but if you want to see how are you improving then you can select that you want the last 90 days, the last month or the last two weeks and something like that.

6:54So it depends on the type of things you want to analyze. But the next step after that is to go to your Google Drive and then you upload that data to Google. And this is pretty important because with the download that we just did, we are going to have all this data compiled together and ChatGPT needs these small pieces of content to analyze. So once I open this in Google Drive, I have my file and I will go to, again, file. I will go to download and then I will go to tab separated values or comma separated values. This is just a way to download the data so you can have, ChatGPT can interpret this better.

7:44And you will do this for every single one of these tabs. So I will do this for engagement. Then I will click post, like the next tab that I have, because it downloads all your data. It downloads your engagement, the top post that you have. It downloads the followers. And what I like about the followers, it's really cool. Because here you can see my dark past in the screen. Because when I started, I just got one new follower for post. I was getting like zero followers every time I was posting. And it was increasing until I was getting like 200 followers per post I was doing once I have the strategy.

8:22So you can, with all of this data, you can really analyze how it has been growing and everything. But yeah, it's amazing. Again, for those of you not watching the screen, we're looking at an Excel table or Google Sheets in this particular case that starts a year ago because that's the data we downloaded. and it shows the date of every post and how many new followers you got. So there's a date, number of followers, date, number of followers. And the numbers in the beginning are 1-0-0-0-3-11-8-0-0-2. Very small numbers. I'm not trying. No, I think this is awesome because this is a year ago. And now every post is in the dozens of new followers and sometimes in the hundreds of new followers, meaning this process works.

9:06Now, as far as the specifics of what Celeste is saying, is that it comes, when you download the data from LinkedIn, it comes in several different tabs and ChatGPT doesn't know what to do with the other tabs. So you basically save each tab as a separate file. So then you can load each separate file to ChatGPT. Okay, so now we have those files. What do we do then? Yeah, so now once we have these files, we go to ChatGPT. And here you can see that I already did this because I wanted to take the time to really think about the questions that I wanted to ask and really think about the strategy I wanted to develop.

9:47So you go to chat GPT and you have an option to upload your files. So you start loading the chart that we just downloaded and we downloaded one by one. So this means that we have four charts on the screen right now. And I say, I start with something important. I say, okay, analyze the data internally. And once you understand all the things that are going on, answer okay. And I will start asking you questions because I want to give chat GPT this time to start to start analyzing, because if you just bombarded chat GPT with questions, it will be like, okay, here's the first thing I found. And probably it's not pretty accurate.

10:26So you want to make sure that you do this thing first. And then here's one thing to start to get interesting because I say, okay, from this chart that is a chart of top posts, give me the URL of the ones of the 10 posts that have more like significant follower increase. So after I asked this question, it started, it gave me a chart with the date of the post URL, the impressions, engagements and follower increase that I was having. and this has been pretty useful because now I can click in the post URL and I see what are the exact posts that are giving my more followers. So I say, okay, if I want to increase the audience that I have, if I want to grow, what are the types of things that I need to keep posting in order to be able to achieve that, right?

11:25And the same happens with, for example, the posts where I have these last followers and I have to have like most updated data. And this is pretty interesting once I was analyzing at this because in the chart that I'm seeing right now, that I'm looking at right now, I see that, for example, I have a post with 80, 80K of impressions. 18. yeah 18 000 impressions just over almost 19 000 impressions that's huge yeah but not it's interesting let me tell you why because with the follower increase i have 30 new followers with this post so lots of impressions that's nice 30 followers huge but now in the chart where i have the ones that give me more and not the ones that give me less i have another post that has 12 12 ,000, which is less.

12:21Those are less impressions. But what happened here? I have more than the double of that of new followers. I have 97 new followers. Whoa. Okay. Are you telling me that impressions are not the same as followers? So this type of vanity metrics, I love impressions, but are they actually being useful? So what I did in this, I just need to analyze what are the things that I'm doing in these types of posts that are not giving me, that are not helping me to grow, but are giving me a temporary, a temporary win. That is just the amount of impressions I have today, but it's not going to help me to keep growing my audience.

13:04So I want to pause it just for just a second to do a quick summary and also talk about the pros and cons. First of all, this is incredible, right? We took data from LinkedIn that is free. You have access to your own data as long as you've been posting. If you haven't been posting on LinkedIn, maybe you have a trick for that as well. But if you have been posting on LinkedIn, then you can see all that data and you can download it and you can use it. What Chachupiti does, which is really magical, it goes to the four different files and combines them in its own head to one big database of information.

13:38And then it can combine all the parameters together. So basically the date of the post, a link to the post, because that's what Celeste requested, the number of impressions for each post, the number of engagements of each post, meaning people who liked it or wrote a comment and so on, and the number of followers that were added because of this post. Or it's not really because of this post, but it's in the same day. So that it doesn't really know whether it comes from that post, but it's people who started following you at that day. And what she found out is that there's no direct correlation between the number of impressions and the number of new followers.

14:15Like in some cases, there are higher number of engagements with less followers. And for Celeste specifically, her goal was to grow the audience. And in this particular case, that means that you need to optimize for the posts that achieve that versus the post that gives you higher engagement, which again, depending on your goals, like you can say, I want to get a lot of engagement. That's fine. It just depends on what you're trying to drive so that's where we are we have this data we have the links to all these posts the posts that drove a lot of engagement and the posts that drove a lot of followers what do you do next yes so the next thing is the analysis right because i have if you notice the way i am prompting this and you need to be careful that you learn how to prompt because just because you know how to write doesn't mean that you're going to be able to express your thoughts in a way that a machine can understand.

15:09Because we forget that because chat GPT sounds like our little friend, we forget that it's a machine, right? But this is pretty important. For example, I'm giving it, I'm saying, okay, from this specific chart, I want you to give me this. And another thing is that I'm specifying the format where I want to get the information. In this case, I specified that I wanted a chart format. We are going to see another type of format in the next steps we also i also specified the date impressions of engagement like all the columns that i wanted to get in my chart so i could analyze everything so it's not blog that i'm having this output that allows me to yeah so so going back to good prompting you need to tell it what hat it's wearing in this case probably like a social media analyst you need to tell it what you want it to do.

16:04So describe the type of data and what you're looking for. And you need to specify the format of the outcome because the format of the outcome could be a tweet, could be a blog post, or in this particular case, could be a table. And if you're doing a table, tell it exactly what you want in the columns. So you're going to get the exact outcome you want. And even within the columns, you can tell it what format you want it, like date and time, just date. Do you want it in thousands? Do you want it in hundreds? Like whatever you want as far as the outcome. So you don't have to reformat it afterwards for whatever need that you have.

16:35So you can specify the format in very higher detail, and then you'll get the kind of results that Celeste is getting. Okay. Awesome. What do you do next? What I do next is to start analyzing. Of course, I already said one more of the parts of the analysis, but the second thing I noticed is, as I told you, this post, like the post that has 18 ,000. was about AI. It was about artificial intelligence, but it was about artificial intelligence itself. It wasn't about AI plus content, but almost seven of the posts that I have in the top are about AI plus content. So I already see what are the topics that are allowing me to grow.

17:23So I know that I don't need to post only about AI, Even if that's trendy, that gives me views, but that's not helping me to build the audience. But when you niche down in a specific topic, like in my case, it's where I see more growth. So that's also one of the things I notice. And now what I do next, other types of questions that I like to ask is, for example, what's the average amount of engagement that I get through the week? and again as we talked before i asked to have the information in in the graph so i can see that visually i'm a visual person i like it and now i see that the day where i have more engagement is usually on mondays and wednesdays like the best day is a monday and the reason of these it's because I see after analyzing this that the audience that I have takes LinkedIn as a job.

18:25So it's usually content creators or people that have a business on LinkedIn. So they take this LinkedIn game as a job. So they usually are pretty active on Mondays and on Fridays, they're like, okay, maybe a little bit less of commitment on Fridays. So if I want to increase my engagement, I already see that probably I can move the days I post instead of being on Friday, maybe on Thursdays. Because I can get an increase on engagement there. And if you're looking at this right now, on Thursdays, I don't have a lot of engagement, but it's pretty decent. It's still pretty decent. And the reason that I don't have a lot of engagement is because I don't post this day.

19:08Yeah. So even if I don't post, I still think that there's opportunity to grow my engagement on those days because it's already pretty decent, even if I don't post. So it's just something about analyzing the data and understanding how it works. I want to touch on what you just said, just like on the underlying concept of this. You need to remember that the fact you don't post on LinkedIn doesn't mean there's no activity around you on LinkedIn. because your post, yes, it gets the most amount of attention, probably in the first 30 minutes and then maybe a few hours, but it's still there. Like I get the really weird ones.

19:42A few weeks later, somebody will comment on a post that I did a few weeks ago. I'm like, how did you even find it? That's buried like the height of the Eiffel Tower, scrolling down on LinkedIn to find it, but it still happens. But definitely within the following day, following two days, people are still commenting on all their posts because that's when they got in. And that's what the algorithm showed them. And you can still benefit from additional engagement, especially if you engage back. If people comment on your post or like your post and you go back and you connect with them or comment back or ask them a follow-up question, you can drive additional engagement without posting.

20:16And again, the graph that's on the screen right now is clearly showing that. There's three days of post, which shows a lot more engagement, but there's still 40 to 50 engagements on days that do not have posts just because people are still engaging with posts from previous days. This is phenomenal. Okay. So I see that according to my data on Saturdays or Sundays, I don't get a lot of engagements here. And that's because usually, like in general, it doesn't need to be like this for you, but most product profiles are not active on these days. So I see that in my strategy, I usually post my post of promo, where I promote my services and everything else helping people on Wednesdays.

20:59Because this allows me that, as you said, the post keeps showing to other people. If I post on Friday, a promotional post, it will almost die the same day. Because LinkedIn usually promotes the first hour, then the next six hours, and then in the next 24 hours. On Sundays, I don't get a lot of engagement already. But yeah, that's the goal. And the next and last question that I like to ask to ChatGPT is what is the amount of impressions that I get through the week? And again, I want to see this type of analysis in a graphic too, so I can analyze that with my own thoughts easier. and I see that again on Mondays I usually get much more impressions and I want to analyze this I can also ask for the links that like for the posts that are giving me this the biggest amount of impressions but the way I like to use this data it's to learn how to grow engage and monetize an audience.

22:09So to grow, like we already said, the first workflow where I analyze what are the things that helped me to have more followers and grow a community. To engage, I have the second question where I ask what type of engagement I have through the week. And I can analyze if the days I'm posting are actually the best ones for my type of audience that I'm getting. And then to monetize my audience, I usually like to have this amount of impressions. Because, for example, when I work with or collaborate with brands I like and they say hey can you help us to promote this post or this tool I say okay yeah and but I will do it on Monday like it's always like that I will do it on Monday or I will do it on Wednesday because I know that I have more reach on those days and I can make sure that the workflow or the value that I'm adding to that company or to the audience it's trying to like reach the most amount of people or also if I want to grow the audience and I have a piece of content that I know it's really valuable we usually we have this feeling right when we create content we are like oh my gosh this is the content is one of the best ones I have done or it's adding a lot of value because I have this new piece of information that I didn't have before and I think it can be useful for more people I usually recommend these to post on the I usually recommend to post this on the days that you have more reach because something's going on there.

23:34Maybe you have more movement in your audience. Maybe, yeah, you have something going on. And I thought this wasn't really important. Like you hear people on LinkedIn talking a lot about what's the most important thing that you need to comment a lot. Some people like to complain that they don't want to comment one hour per day to be able to grow their accounts. but it's just the way the algorithm works it's built to to create networking it's built to connect professionals so the way the algorithm works is that it's just that when you when the algorithm knows that you have connection with someone else like in the comments in the dms it will promote your content yeah absolutely the i want to add one thing to what you said as far as the importance of this.

24:26A lot of people, when they define when they want to post and so on, just go to blogs that say, hey, the best time to post on LinkedIn is on Wednesday at 10 a.m. And I'm like, that's ridiculous because that's specific to your audience, your content, your type of stuff, your industry. It's different for every single individual. And so posting when everybody else is posting, first of all, it's not always the best idea because then everybody else is posting and there's more noise to compete with. But also you have the opportunity right now to literally look at your data and see what's working and what's working and what's not working or what's working better or what days.

25:06And you can even run an experiment, just test it out for a month and a half and just say, I'm going to just post on Mondays and Thursdays moving forward. And I'm going to post this on Monday and this on Thursday. I'm going to see if there's a difference. And you can test different things out or different things in different weeks and see what actually works best for your audience with your topic, with your niche, and make decisions based on actual data versus on general data that everybody else is using. Yes. And it's pretty useful to be able to make these types of things. And if you're looking at this right now, you will notice that I was opening my documents to be able to see my carousels.

25:46And it was because I want to show something. I was creating one of these pieces of content that But I thought it was going to be pretty useful because it's a piece of content that was saying about how it was explaining how scammers were making money with AI. And I wanted to share to people how to provide these types of things. So I posted this the first time. And then I posted again the same piece of content after I was growing a little bit more my audience. And you will see, if you're looking at this right now, that the first post has 300 views, like literally 300. And I was like, oh my gosh, how can this be true?

26:28I thought it was going to be pretty valuable. I spent four hours creating this carousel. It can have 300, please. So I was like, okay, there's something going on right here. And I started analyzing these data. Of course, I had less data on those days. It was just a thing. And I look at these types of statistics that I was showing you at the beginning, the days where I have more engagement, the days where I was able to grow my audience a little bit more, and I was starting to apply this data. And the shocking thing for me here was that the second time I posted this, it got 56 ,000 impressions. Like 300 to 56.

27:13this. And I understood a really important lesson when I did this at the beginning, that it was the fact that your content, like, it's going to hurt, okay, it's going to hurt, but your content doesn't matter. Like, the content, it can be as valuable as possible sometimes, but it's not going to matter what the content says, but who is saying that. so because i want to add to what you said because i think it's critical first of all i got a question before i'm going to say what i'm going to say was the post itself the same or just the content within the post was the same did you change like the caption no nothing it's the same same thing exactly yeah so i want to say something here that is critical that there are so many parameters that impact the success of a post and you need to optimize at least for some of them but the interesting thing is if you know that and that's where i disagree as far as the content doesn't matter this is a good piece of content you spend a lot of time and research and effort and if you were watching this you can see it also looks amazing like you thought this content is going to be successful.

28:33If you post something that you think is valuable to your audience and it doesn't work, try to figure out why. If you already know your audience and you know what works and you're like, holy crap, I thought this is going to be like my best post ever. And I got crickets. Something is broken. And assuming something about your audience, it cannot be random. And this could be the day you posted it, the time you posted it, what you actually wrote. So maybe the piece of content is awesome, but you didn't use the right hook. Maybe you should have mentioned somebody in the post itself that is relevant to this to capture the audience.

29:08And if you do this, it's perfectly fine to repost old posts if they did not work or even if they did work. So now you have a bigger audience, more people are following you. Take something that did okay six months ago, change the post a little bit now that more about your audience, research when to post it and post it. Again, in this particular case, it went from 300 impressions to what, 56 ,000 impressions. So you can reuse good content again and again and get bigger and bigger ripples of impact if you learn in between and make the right adjustments. Yeah, that's absolutely right. And there are more ways to use AI for content.

29:52And this type of workflow was just for analysis. But I have a framework that I like to preach. I always tell people to use this framework, but it's the C framework. And it stands for structure, example, and analysis. So you can always use AI for those three things. And those are my favorite things to use AI for. Because I think some people fall into the hype of you can create content and ask Tachibiti to have this amount of posts and go viral and be amazing or rocket emoji and everything else. But the fact is that because, for example, just ChatGPT is a probability-based model, it creates content based on the highest probabilities that something has to happen, but something is happening.