Nicolas Korvessis shows how to use Claude to rewrite Facebook ad hooks every morning in five minutes. He tracks hook rate and hold rate in Ads Manager, feeds underperforming hooks into a custom Claude prompt, gets five ranked variations back, and tests the best one. One ad's hook rate jumped from 32% to 42% in a week.
These are my hook rates from last week, and this ad went from a 24% and a 29% hook rate all the way up to a 42% hook rate, and I didn't even touch it. Claude rewrote the hook, I swapped it in, and it took only 4 minutes. I do this process every single morning, and here's the exact setup. Most people write 3 to 4 hooks, pick the one that feels the best, launch it, and they check back in 3 days hoping that something worked. And I used to do the same thing, but the problem is that you're guessing. And by the time you know that a hook's failed, you've already burned money on it and it's too late to go back. So I decided to build a full system where Claude and AI analyzes my hook performance every 24 hours and writes me replacement hooks based on what's actually working. And let me show you exactly how I do it. So here's what happens every morning. I jump into the Ads Manager and I find the hook rate data for my ads currently running, and I'll find anything that's on the lower side of things. So closer to 20 or 25%. In this case, for this week, I found this one. That had a 32% hook rate and a 9% hold rate. Then what I do is I take that data plus the hook from that specific ad. And in this case, since it's a video ad, I'm actually taking the hook on this first line here of the actual creative, and I'm going to feed that data to Claude with this powerful prompt. Claude will then give me back optimized hooks ranked by predicted performance, and the whole thing takes me less than 5 minutes. Let me break down each step for you. So inside the Ads Manager, first you want to find your hook and your hold rate. Now these are actually custom metrics that I've built, and these are really, really great for video ads. So I highly recommend that you do the same thing. To actually get your hook and hold rates in here, the hook rate is going to be 3-second video plays divided into impressions, and then hold rate is going to be video plays at 50% divided into impressions. To make these, you'll go to the columns here, and then you'll go to Customize Columns. And then you'll go to custom and then create a custom metric. Then you'll drop in those metrics that I just explained. You'll call the hook rate hook rate and then hold rate hold rate. From there, you'll be able to add these custom metrics to your columns and then actually see how well is your ad hooking people and how well is your ad holding people. Now, from what I've seen, hook rate is good if it's over 25%. And if your hook rate's below 25%, that hook, the first few seconds of the video or even the first line of the ad copy, is gonna be the problem. If it's above 25%, but your hold rate is dropping, that means that the body is the issue. Now today we're just gonna talk about fixing hooks, but just as a little bit of context on what you should be expecting and what you should be looking for. Now I'm gonna show you the exact prompt that you need to use near the end of this video, but I want you to understand why it actually works in practice, because if not, you're just gonna get these generic hooks that don't really drive your results up and instead maybe improves vanity metrics that don't actually mean anything. So I want to explain why. So for this past week, what I did is gave Claude this hook here that was broke coach versus full-time coach. Like I said, I'm using the video hook in this one because with video ads, the video is going to make the biggest kind of impact versus the copy itself. And then from there, Claude gave me back 5 different hooks that I could test based on the prompt. Picked hook 2, and what ended up happening was really, really cool. This is a hook that I was using previously, this tactical one, and then it was kind of a before and after. Which was working pretty well. It had a hook rate of 32% and a hold rate of 9%. I gave Claude all those metrics with the prompt, and then from there it spat out 5 new hooks that I could test. From there, what I went ahead and did is I picked hook 2 because I felt like it would work the best. I tested it this past week, which was adding broke coach versus full-time coach at the beginning of the video, and that actually bumped it up by 10% on the hook rate, 42% hook rate and a 17% hold rate, which makes a really big difference all across the board. And this crushed it for us. And it was a very simple process of just giving Claude the metrics, having a really supercharged prompt to get hooks. And then we didn't just blindly pick. We really decided to pick on which hook is actually agitating people or which hook is actually going to show people, hey, this is the dream outcome that you could get. I really want to be on like one side of the spectrum, either really, really kind of digging the knife into people and making it really painful. Or really showing people a very cool outcome that they're really wanting to achieve. Okay, and now here's the actual prompt that I'm using. I'm telling Claude, you're my Facebook ads hook optimization specialist. Your job is to analyze ad hook performance data and generate replacement hooks for underperformers. So you're gonna go ahead and drop in your niche for coaches. My target audience is coaches and experts. My brand voice is, I'm just gonna say direct. Rules for generating hooks. Every hook must pass these steps. Use these proven hook structures. Keep hooks under 12 words. Never use generic filters. So then what I'm going to do is I'm going to go ahead and actually drop in both of the hooks that I've got here. We'll just copy and paste those in and then fire this off. Now, this prompt, like I said, will be below in the video description. You can also grab it on screen right now. Should be up on screen. And let's see what Claude generates for us. Okay, so here are some variations. We've got this tier 1 closest to the proven structure, tier 2 with more pattern interrupt stuff, and then specific pain callouts and curiosity gap. All these are great, and it's very easy to actually start testing them. So once I'm happy with whatever hook that I'm testing, I highly recommend testing things in a 1 ad per ad set structure. So I'll just duplicate what was working. I'll pretty much keep everything the same, but just test the hook. Now, you could test this in a way where you're doing maybe multiple ads per ad set. But I just suggest doing one ad per ad set for some of these more granular tests, because it can really make a big difference. Like I said, this takes me 5 minutes every morning. I do it with everyone that I work with, and I do it for all the ads that we run internally as well. And it's honestly been a huge game changer. So if you want access to the prompt, just see that in the description below or earlier in this video and try this out for yourself. Look at your hook and hold rates and check the link in the description if you want more deep dives and access to stuff that is going to take you way further in your paid ads and in your marketing. Now, if you want the full creative system where we actually show you how to do images, videos, etc., all from Claude on autopilot, I'm breaking that down in my next video. So make sure you subscribe so you don't miss that one. Thanks so much for watching and I'll see you then.
