WEBVTT 00:00:00.080 --> 00:00:30.040 So this is how I'm using Claude Cowork in my creative strategy workflow today. So here's the thing. Most creative strategists and digital marketers are using AI completely wrong. And it's not necessarily because they're bad at prompting or even that they're using the wrong tools. It's because they're asking AI to do the wrong job. So I've spent over a $100,000,000 on meta ads, and I've worked in this industry as a creative strategist and media buyer for over ten years. And I will say that AI only really started to click for me when I stopped using it as my replacement. 00:00:30.200 --> 00:00:33.880 Instead, I treat AI like it's my junior creative strategist 00:00:33.880 --> 00:00:49.545 or my marketing assistant. So today, I'm gonna walk you through my five best use cases for Claude Cowork and creative strategy. You'll see my actual prompts and how I'm thinking through this platform right now. Because the goal isn't to replace your strategic thinking, it's to amplify it so that you can spot opportunities 00:00:49.545 --> 00:01:13.935 faster that you would have never seen without it. But first, I wanna give you a sixty second rundown on co work. Please feel free to skip it if you're already familiar. So Claude has three big functions, chat, co work, and code. Chat is your basic chat interface. You probably already have a lot of experience with that. Now co work is where Claude can actually do the work for you. It can open your browser, read your files, and complete tasks on your behalf. Some people might call this an agent. 00:01:14.255 --> 00:01:35.610 Encode is for developers. It's a command line tool that helps you build apps and websites. So if this isn't you, just feel free to skip that one for now. Now this video is very specifically about Cowork because I want us to concentrate on getting real work done, not just information being generated. Now in order for you to access Cowork, you are gonna need the desktop app, not just in browser. 00:01:35.690 --> 00:01:40.815 And as far as pricing, I will say at a minimum, you are going to need the pro plan. I personally 00:01:41.295 --> 00:01:42.975 use the max plan, 00:01:43.135 --> 00:01:53.375 but that's just me. And the final thing that I'm gonna note here with co work is that in order for Claude to access different parts of your computer in different apps, you're gonna have to give it access 00:01:53.375 --> 00:01:54.655 through connectors. 00:01:55.130 --> 00:02:03.130 And for everything that we're gonna go through today, make sure that you at least have Chrome connected as well as Slack. Now let's dive into this first one, analyzing 00:02:03.130 --> 00:02:04.170 ad libraries. 00:02:04.170 --> 00:02:18.015 Now essentially what this does is it's going to analyze a Facebook ad library for you, aka the ads that are currently being run by any brand right now, and it's going to give you a report that contains incredible 00:02:18.015 --> 00:02:29.050 and actionable data for you to use. You can do format breakdowns and a deep dive on the creators they're working with via partnership ads and just an overall scope of their creative strategy. 00:02:29.130 --> 00:02:44.805 Now the reason why this is so impressive is because I used to do this exact report for my clients every single month. Now here's the prompt that we're gonna be using to generate this. Let's go ahead and put this into Claude CoWork and see what happens. So the basic prompt structure looks 00:02:44.805 --> 00:02:45.445 like 00:02:45.685 --> 00:02:46.325 this. 00:02:46.645 --> 00:02:56.660 Number one, you're gonna say, can you do an analysis on RAND? Let's go ahead and just do Ridge Wallet. Right? And then we're also gonna get a link 00:02:56.900 --> 00:03:08.020 that goes directly to their ads library. I found that when I'm doing this, if I just say go to Ridge's ad library, it often won't get the right thing. So I just wanna make sure, hey. Let's have it 00:03:08.625 --> 00:03:09.345 exact. 00:03:09.425 --> 00:03:20.465 So we have our prompt right here, and the first thing that you're gonna do is just select a project that you're gonna work in. I always just have everything be put into a clod folder. 00:03:20.705 --> 00:03:26.760 And then another thing that you can potentially select is, oh, do you want it to be Opus, 00:03:26.840 --> 00:03:34.360 which is like the highest computing power? Sometimes I use Sonnet, but since I have the max plan, I'm gonna go ahead and just use Opus for the best results. 00:03:34.680 --> 00:03:38.280 I already have the Facebook ads link 00:03:37.515 --> 00:03:39.835 directly to the ads library open, 00:03:40.315 --> 00:03:42.075 and we're just gonna go. 00:03:42.555 --> 00:03:44.635 And you can see here immediately, 00:03:44.795 --> 00:03:52.700 it starts the progress. So it's gonna show you everything that it's gonna do step by step. And you can also see that the connector 00:03:52.860 --> 00:03:56.300 is also searching. And you can if you want to 00:03:56.700 --> 00:04:10.275 really go through and see how it is starting to conduct these tasks. Now what you'll see here too is, oh, Facebook's domain is direct from direct fetching. But if if you can see right here, Claude, not me, 00:04:10.435 --> 00:04:12.595 just pulled up this 00:04:12.995 --> 00:04:14.195 entire window, 00:04:14.355 --> 00:04:30.110 and it's starting to go through Ridge Wallet's ad library, which is pretty cool. So I'm gonna go ahead and collapse that and let it do its thing. Sometimes it can take a few minutes to actually generate the entire thing, and you can actually see it moving in the background. That's not me. It's crazy. 00:04:31.885 --> 00:04:32.685 Ah, 00:04:32.685 --> 00:04:34.285 and there it is. 00:04:34.605 --> 00:04:37.085 K. It's giving me a little summary. 00:04:37.165 --> 00:04:39.725 Let's open it up, shall we? Wow. 00:04:39.725 --> 00:04:43.165 Okay. So we can see for Rigs, they have 291 00:04:43.330 --> 00:04:47.490 active ads for product lines, for creator partners, and 94 00:04:47.490 --> 00:04:50.370 ads were launched on April 4 alone, 00:04:50.690 --> 00:05:13.505 and the longest running ad is a 158. Alright. Let's dive in. This is such cool data. K. We have a video versus image breakdown. Pretty even, fifty fifty split. We can also see the video duration distribution. I actually see this quite a bit with videos where they're forty five to sixty seconds long. Now I can see the percentage of partnership ads is pretty low, only 4.5%. 00:05:14.090 --> 00:05:27.930 Now something that I'd probably do a follow-up here is ask, hey. Where are these ranking in terms of impressions? Sometimes this report will automatically show it. Sometimes it doesn't. Core messaging strategies, durability, toughness, lifetime guarantee sales. 00:05:27.930 --> 00:05:28.890 Wow. Wow. Wow. 00:05:29.965 --> 00:05:55.860 And we can see, like, the big pillars here, specifically for wallets, it looks like. Now this is what I really love is seeing the inferred target personas. Now, something that I really like to do as part of my process, especially when it comes to persona mapping, is mapping the personas that you seem to be targeting from your creatives versus who your actual customers are, aka the analysis from the actual reviews. And then the top 10 ads by 00:05:56.020 --> 00:05:56.820 impressions. 00:05:57.315 --> 00:06:00.595 Okay? None of the partnership ads, it seems, 00:06:01.075 --> 00:06:06.515 but some of them are, like, the legends in Bloom. Of course, I think they're actually probably pushing 00:06:06.595 --> 00:06:12.470 spend to that, maybe on, like, a reach or an awareness campaign just because it is limited edition. 00:06:12.790 --> 00:06:41.800 Next up, let's talk about some other social media reports that I'm generating. Now one is gonna be for creators because in addition to owning an ad agency, I am also a content creator. And then the other is for brands that you're interested in spying on, aka your competitors. Now let's first dive into the creator one because I actually made that one first because I actually used to have my assistant generate this report for me every single week and month, and now it is completely automated by Claude. Okay. So here's the prompt. Let's run this thing. Okay. 00:06:41.880 --> 00:06:50.120 So again, you can have it compile in whatever type of format you want via Pitch Deck. I like the HTML graphics, 00:06:50.520 --> 00:06:55.525 but you could also do it in a sheet format if you wanted. So I have all of my links here, 00:06:55.845 --> 00:06:58.485 and I don't have any Chrome tabs open. 00:06:58.725 --> 00:07:07.045 Sometimes too, what I'll say is it'll hit a wall and say, hey, we can't access these. So I will tell it to make sure you open 00:07:08.010 --> 00:07:09.290 these links 00:07:09.850 --> 00:07:11.290 with the Chrome 00:07:11.930 --> 00:07:12.890 connector. 00:07:13.850 --> 00:07:20.805 Let's go. Oh, and this is cool. It's gonna ask me a user question, which I love. When you say last week, which date range do you mean? Great question. 00:07:21.285 --> 00:07:22.965 Um, let's do 00:07:23.525 --> 00:07:33.605 let's do this one. What engagement metrics matter the most to you for best performing? All of the above, baby. How detailed should the do less of recommendations be? High level detailed 00:07:33.605 --> 00:07:40.640 with examples. Call me out. Actually I did this maybe a week or two ago for something, and it read me to filth. 00:07:40.960 --> 00:07:46.080 Up. And it just pulled up my tab too. We love to see it, and it's on my LinkedIn. 00:07:46.640 --> 00:07:47.520 Crazy. 00:07:47.520 --> 00:07:49.680 Okay. I'll get back to you when the report's ready. 00:07:58.315 --> 00:07:59.115 Okay. 00:07:59.115 --> 00:08:00.715 Now it's here. 00:08:01.515 --> 00:08:10.100 Very cool. Let's open it up. Now, of course, you can give it other specifications if you want it in a different color or whatever, 00:08:10.580 --> 00:08:22.115 but let's see. Okay. Last week, I did 10 posts on LinkedIn. Not bad. Six on Instagram, one on YouTube, which was a short, two on Twitter, and two on TikTok. And here was my top performer. 00:08:22.115 --> 00:08:24.435 It says it went viral. It did not. 00:08:24.835 --> 00:08:27.315 They put ad spend behind it, but 00:08:27.795 --> 00:08:28.915 good to know 00:08:29.475 --> 00:08:35.410 your brand breakdown and process content drove the highest saves. Not surprised. So the Oats Overnight post, 00:08:35.730 --> 00:08:39.090 that many saves, and the three processes real. 00:08:39.250 --> 00:08:52.555 Very good. Yeah. I did get a lot of impressions on that Caraway Home one and the AI video content, of course. See, it's just so cool to have this type of data at a glance for creators, 00:08:52.875 --> 00:09:02.315 especially too when it does it by platform in the different posts that I have. So I get a LinkedIn breakdown. That's awesome. Very cool to for me to refer back to. 00:09:02.980 --> 00:09:03.860 And then 00:09:04.180 --> 00:09:05.780 Instagram breakdown. 00:09:06.180 --> 00:09:08.900 Do less Amazon partnership post. Whoops. 00:09:10.820 --> 00:09:11.780 Fair enough. 00:09:11.940 --> 00:09:12.820 Observation 00:09:12.820 --> 00:09:14.980 had an incredible 4.8 00:09:15.275 --> 00:09:16.315 save rate. 00:09:16.635 --> 00:09:20.715 And this is where I'm like, ugh, you're reading me to filth. YouTube and x are significantly 00:09:20.715 --> 00:09:21.755 underserved. 00:09:22.075 --> 00:09:27.595 Do less generic sponsored content. Happens. You're cross posting heavily on LinkedIn and IG and TikTok. 00:09:27.995 --> 00:09:28.315 Yeah. 00:09:28.840 --> 00:09:29.560 Fair. 00:09:30.280 --> 00:09:59.650 Personal pros without a professional hug. Okay. But I love getting this report. You can also tell co work, hey. Be sure you send this to me in Slack. And now that they have scheduled tasks, you can make sure that this runs every single week, which is what I do. Now the next one is for a brand or competitor analysis that I like to run for a few different use cases. Number one, you can do this for your competitor. So I'll actually have co work do this analysis for me monthly so I can keep tabs on what the competition is doing organically. 00:09:59.810 --> 00:10:07.970 Again, I used to do something like this, which while it did focus more on the paid aspect, I would also pull in elements from time to time organically. 00:10:08.050 --> 00:10:29.620 And another way that I'm using it is with my clients' competitors to really map out what their strengths and gaps are and if there's anything that we should capitalize on. And sometimes just because I'm hella nosy, I'll run it on some of the big brands or brands that I really admire. And what I really like about this too is when you're finished generating this report, you can also ask co work follow-up questions to that data. 00:10:30.020 --> 00:10:37.220 And here. Now, again, I'm having the final format be in a summary, a spreadsheet, and an HTML file. 00:10:37.700 --> 00:10:43.235 And you can really do this for as many brands or as few brands as you want. 00:10:43.875 --> 00:10:48.835 I've worked with a lot of beauty brands before, so I'm like, oh, I wanna keep tabs on these specifically. 00:10:48.995 --> 00:10:56.530 Let's go. And now we wait. Now the end result is going to be this. So I did have the summary right here, 00:10:57.170 --> 00:10:58.530 and then I have 00:10:58.770 --> 00:10:59.970 the report. 00:10:59.970 --> 00:11:07.495 So what's really cool is at a high level, because the follower counts, wow, Laura Gallo and Jones Road Beauty are neck and neck. So cool. 00:11:07.815 --> 00:11:11.415 Celebrity collab are 10 x multipliers. Not surprised, 00:11:11.575 --> 00:11:12.375 for beauty. 00:11:12.695 --> 00:11:33.410 And founder led content always punches above its weight. And then it gives top reel by likes because I really specifically wanted to look at the video content, and then it breaks down the top five reels each, and I can go ahead and it has a direct link. So cool. I love that effect. And it also, like, points out what they're doubling down on utility 00:11:33.410 --> 00:11:34.130 driven 00:11:34.290 --> 00:11:41.475 content. Same thing for Laura Galloway with their their but they're doubling down on celebrity partnerships and media moments. Interesting. 00:11:41.955 --> 00:11:42.915 Multiphase 00:11:42.915 --> 00:11:49.635 product launch campaigns, which is also very interesting because they're the only one that doesn't have a really front and center founder. 00:11:51.430 --> 00:12:16.215 Very, very cool. And the final one is how to build persona research decks. And we're actually gonna be breaking this down into three concrete steps because at any given time, you might just want to do one part of this process. Number one, I'm gonna show you how to scrape for reviews into actually break that data out into personas and then putting that data in a finalized deck. Now one thing I'll flag here is that for a brand like Ridge that has 47,000 00:12:16.215 --> 00:12:21.290 reviews, you're not gonna wanna do all those or it just might take a while, but for purposes of this video, 00:12:21.690 --> 00:12:25.850 I'm gonna do an even 3,000 just so that we can go a little bit more quickly. 00:12:26.250 --> 00:12:26.970 Let's go. 00:12:32.785 --> 00:12:37.985 Now I'm gonna go a little bit ahead. You can see here that it was able to generate for me the CSV 00:12:37.985 --> 00:12:40.145 of all these reviews. 00:12:40.145 --> 00:12:40.785 Amazing. 00:12:41.025 --> 00:12:44.465 We love to see it. It even broke it down into the different 00:12:44.785 --> 00:12:46.065 product variants, 00:12:46.065 --> 00:12:51.960 which we love. And now what I'm gonna do after this is I'm going to be pasting 00:12:51.960 --> 00:12:52.920 another 00:12:52.920 --> 00:13:13.125 prompt right into the same chat of that. I'm gonna be saying, hey, from a CSV of these reviews, do the following. Now what I like to do with this first is to have it put everything in a document for me so that I can review it and really understand, okay, what type of personas are they locking in on from these reviews. 00:13:13.365 --> 00:13:15.445 This can actually make a really excellent 00:13:15.780 --> 00:13:21.860 context document that you can upload to a cloud project. If you're interested in hearing more about that, definitely let me know. 00:13:22.660 --> 00:13:36.485 But I also like it because it's easier to edit. So I can export it and edit it in Google Docs. Awesome. But once it's in a place where I think it's pretty good, then what I'm gonna tell it is, hey. I want you to turn this deck into a presentation 00:13:36.645 --> 00:13:47.740 with visuals that has a graph, yada yada, and it's gonna work. And then you're gonna have something like this that you can review with your team. Now being able to have this kind of data and documentation 00:13:47.980 --> 00:13:53.180 without you actually having to do it, I can't emphasize truly what an unlock it is. 00:13:54.140 --> 00:14:06.315 And if you're seeing this and you're worried about the branding and the blah blah blah, all of that can truly be edited. And I will say too, Claude has a connector directly to Canva. So there are some specific like Canva style documentations 00:14:06.315 --> 00:14:10.555 that we prefer. We will actually have Claude go directly to 00:14:11.115 --> 00:14:11.595 Canva 00:14:12.210 --> 00:14:31.065 aspect. And if I'm just doing my own personal work, Gamma is something that I also personally use. And full transparency, I have partnered with Gamma on my Instagram and on my LinkedIn. They are not sponsoring the YouTube channel at this point, but it is genuinely a tool that I use and really, really like. So if you are creating a lot of reports, 00:14:31.225 --> 00:14:33.145 definitely try out the 00:14:33.545 --> 00:14:35.385 native PowerPoint, 00:14:35.545 --> 00:14:37.785 but Gamma and Canva, 00:14:37.945 --> 00:14:43.545 like, there's so, so much you can do in automate. And honestly, it has made just a massive, 00:14:44.100 --> 00:15:18.650 massive impact on my day to day. Because it's, you know, kinda funny. Right? Like, as creative strategists, I think we often think about the immediate output, which in many cases is the actual ads or the briefs. Right? But the majority of my time as a creative strategist is actually spent researching, and it's so funny actually. Ogilvy, when he finally had his namesake agency, he didn't bill himself as the creative director or even the owner. He billed himself as the research director, which really just goes to show how important research is in this process. And, yes, while I do have a lot of research 00:15:18.730 --> 00:15:57.714 here that is all automated, I'm still doing a lot of it manually, but it is this type of stuff and the visuals particularly that has just really done so much for me, and I love it. And I hope you guys enjoyed it too. Now I know that a lot of this video was super research heavy, but to be honest, as a creative strategist, I spend a lot of my time conducting research. So these have been a massive, massive unlock for me. However, if you're curious about how we're implementing AI into the rest of our workflow, specifically in briefing and QA, we've really made some great strides there. Let me know because I'm happy to expand on this into a series. And I'll see you guys next week. Thank Thank you so much. Bye.