Name: The AI Survival Guide for Performance Marketers
Uploaded: 2025-11-20T17:16:16.424Z
Duration: 1 h 2 min 32 s
Description: The AI Survival Guide for Performance Marketers

Transcript for "The AI Survival Guide for Performance Marketers": Hello, everyone, and welcome to the webinar today. Thank you to everyone who's here. I'm buying on time for the start of the session. I'm sure we'll have a few more people joining us in the next couple of minutes, so unless they continue to join, I will quickly introduce the brilliant guests that we've got lined up for today, cohost of the Prompted podcast, which covers everything AI related, in performance marketing. Nick Hanley and Chris Nightingale. How are you guys doing today? Doing all good. Yeah. All good. Yeah. Yeah. Chris, I don't know about you, but I I saw you're well, we're all in the Northwest Of England at the minute. I saw literal icicles outside my window today, which was, impressive, and the heating's been on nonstop. But, Nick, you're you're traveling tomorrow. Right? Nice. Yeah. I, find South Africa tomorrow. Currently, Nottingham, where it is minus two, I believe. So, yeah, it's quite cold. But it'll be 24 degrees where I'm going. I'm jealous. nice. I'm jealous. While we're waiting for people to, join the session here, I just wanna bring on screen, this slide, and and just you you guys have got together and and, you know, you're cohost of the prompted podcast. And I'd love to know, a bit from you guys about, you know, the kind of thought process that went on prior to you guys launching the podcast and, how you found it so far? Because I know from experience that running a podcast takes a lot of work. So, yeah, it'd be good to know what it's been like so far for you guys. It's a bit of a bit like a random one for us, wasn't it, really, Nick? I think me and Nick obviously have a passion about everything, performance marketing and AI, and we just had all these conversations naturally that spoke and you went, you know what? Let's just, like, let's just film a podcast and just put some stuff out there to people. And I think, yeah, reception's been really nice so far. I think a lot of our kind of peers in the space have really enjoyed it, and we just enjoyed sometimes what feels like waffling. But, actually, I think with the good old world of AI, we can actually condense them down to some pretty nice clips as well. So, yeah, it's been it's been really good. Nice. And, just a quick note to say, if you aren't already following Nick or Chris on LinkedIn, you can find links to their profiles in the docs tab here in Goldcast as well as a link to the YouTube channel for their podcast. Or you can just scan the QR code on on the slide here. And I also just wanted to mention that, my interview with Miles McNair, who I'm sure many of you will know, is dropping next week on the Lunio YouTube channel. So be sure to subscribe there as well, to catch that. I've hosted several webinars with Miles in the past, but, this was his first time on the podcast, and it is a good one. But you guys, Nick and Chris, are you gonna stick with the, not the Anco studio, the other studio? Is that your preferred location for recording now? Yeah. I think, we'll stick with Outset Studio, which is in 4 way house in Manchester. If anyone wants to go about, we'd highly recommend it. It is a great first to film in. But, yeah, I think that's gonna be our home for now, unless Chris thinks otherwise. No. No. I agree. It's really good spot. They're really welcoming. It's really good setup. And it's great for guests as well. Like, the the the whole customizable thing with the sofa or the chairs is really cool. So, yeah, if anyone out there knows any super, super cool performance marketing slash AI people, then we're always looking for to guests to squeeze on as well. Probably looking to do, you know, an episode a month or something at least in the next next next year or so. So, yeah, all ears. I I was just gonna ask, so you guys have three episodes out so far, and have you got any, like, rough idea of when the when the episode four is gonna drop? Good good point. Sorry to put you on the spot. TVC yeah. TVC, I mean, Nick, I mean, we have already filmed it. It's a bit different to our last episodes, I'll say. Right. So hopefully. coming out the end of the month depending on how crazy Black Friday is, I'd say. Do you agree, Nick? Yeah. I think that's a a fair call. End of month, early next month is probably gonna be better for us than it. But, yeah, it's gonna be an interesting episode. A bit of q and a episode over what we've done before. So it's quite a different format. Nice. Well, I look forward to seeing that one when it drops. So we've had a few more people join now. So, I will get going. Again, welcome, everyone, whether you're watching live or, back on demand. For those that don't know me, I'm James. I'm the content lead at Lineo. I'll be hosting the webinar today. We're aiming for around fifty minutes to get through all the content. We have got a lot of stuff to get through today, so we might have to rush a little bit towards the end. But we'll try our best to leave at least ten minutes at the end for q and a, which I'll say more about, in just a second. A few other quick bits of housekeeping before we get into the agenda for the session today. We are recording the session, and it will be available to watch back on demand afterwards, on the Lineo YouTube channel and on our blog as well. If you have any questions about anything that Nick or Chris are covering today, please pop them into the q and a tab that you can see there beside the docs tab here in Goldcast. And that just means your question won't get buried in the chat, and we'll try and get run to as many of those as we can at the end. You can also upvote questions that you're interested in, which will help me prioritize which ones to ask Nick and Chris first at the end of the session. So please do make use of the the Q and A tab there. We've also got some resources in the docs tab here in Goldcast, and I'll be directing your attention there at relevant parts of the session today so you can get more context and helpful tools that relate to some of the content we're going to be covering. And we will also be running a few audience polls during the session, which you can vote on in the polls tab. So please do give us your responses there. So you should see a notification pop up whenever a new poll question is open for responses, and you can respond by clicking on that tab and giving us your answer. So on that note, I'm just gonna bring our first poll question, up on screen here. And the first poll question is quite relevant to the main topic that we're covering today. So the question is, how do you feel about Google's upcoming AI powered ads advisor tool? And we've got a range of responses there, where you can let us know your thoughts on that. But while we're waiting for the audience responses to come in, Nick and Chris, it'd be great to hear a brief take from you guys on, what you think about this, even if you haven't had the chance to use it, yourselves yet. I think I think from my side, I'm slightly positive about it. The reason is I think it can hopefully give us some additional insight that we might have missed, I think, or, you know, whatever else. But, equally, I'm always skeptical with tools when they come up from Google. I think we all know what Google's fundamental goal is. It's to make loads of money. And, yeah, as we all know with, you know, recommendations and opticals and all the rest of it, slightly skeptical, but I'm, you know, always need to try it first, and see what I can uncover before I kind of downbeat it too much. Yeah. Nick, similar thoughts to Chris? Or Yeah. I think some of our thoughts. I've seen a bit of feedback just from LinkedIn. I think it might be just asking Gales who brought it up saying that it actually gave outdated information from an old article instead of the newest material. that's available. So I think that there's gonna be some teething teething problems and and whatnot. But, yeah, I think I'm with Chris somewhat positive, but I'd like to test first and get a bit more feedback from the team. Yeah. For sure. So I think from the audience responses that we've had, I think the winner here is neutral. I'm undecided and and waiting to see how it how it performs. So I think most people kind of sharing broadly similar sentiment to to both of you guys. So I'm gonna close that poll down now. There will be other polls that will be opened up throughout the session today, so so do be on the lookout for those. So now I'm just going to, move on to the agenda for today. So I'm gonna bring that slide up on screen here. And we've got eight main points on the agenda today, before we move on to the q and a. So first, we're gonna explore what can be automated and what shouldn't be automated. After that, Nick is gonna run us through some of the material from his recent Bryton SEO talk outlining how you can set up your data team and processes for success next year, including some of his prompt engineering best practices as well. Then we'll look at how the web is becoming increasingly dominated due to our increasingly dominated by bot activity due to LLMs. And we'll briefly look at some data we published at Lunio in relation to how AI Max for sales campaigns increases your exposure to invalid traffic. After that, we'll look at real world examples of AI applications built by Nick and Chris. And there are some super useful resources to point you to towards, in that section if you're interested in building your own, tools, applications, and workflows. And, of course, we'll get guidance and advice from Nick and Chris when it comes to getting started with building your own tools if it's something that you're new to or or haven't tried out before. Then we'll close with some predictions for what lies in store next year, and the kinds of things that you can do now to prepare. Because even though AI is not gonna fully replace marketers, it is definitely gonna change the focus of the role quite a lot. So that's everything that we've got on the agenda for today. So we'll get into it first if I just find the right slide here. So we all know that AI has has changed marketing significantly, and I bet that everyone that's here today in attendance is probably using it on an almost daily basis for certain tasks in their roles. But that said, there are still things which AI and automation can't do or or is quite bad at doing. So, Nick, to pass over to you first. From your perspective, which parts of performance marketing today are are genuinely ready for automation and which still definitely, you know, you wanna keep a human in the loop? Yeah. I think, it's a pretty fast, because I do definitely think we're in the automation era of AI. I don't think we're there to see things that, you know, people do talk about, you know, agents and, you know, having things for. I generally don't think we're at that point yet, especially with what we do. I think a lot of the work that we need to look at is not replacing people from the end of value that's brought to clients, whether that's creative or strategy. I generally do think it is the stuff behind the scenes, whether it's, you know, pulling reports, whether it's, you know, looking through search term reports. So I'm gonna get them into a position that makes them a bit easier to look at or even pulling data from multiple sources to give a bit of an understanding of whether there's something larger than a Google Ads account, you know, a matter account that could be causing the change itself. So, yeah, I think it's behind the scenes that should be automated in a lot of words, but I still think keep that human element there. The reason I say that, we saw literally Cloudplay went down yesterday and took down so much in terms of, like, essentially half. Internet, but things like GPT went down the rest of it. So, yeah, I think it's keeping those skills very much human, but automating as much as you possibly can with AI, using it to proofread to check, but don't get it to do the bulk of the work unless it's preparation for data and analysis. I still think that should very much stay a 100% human, which is the reason that graduates are still incredibly useful and the rest of it, because, like, people still need to learn this craft, not just on it, but back end out of everything. Yeah. I think we're gonna come on to some of those warnings and some of the potential risks that you can, expose yourself to if you're over relying on AI for certain tasks. But if we move on to this next slide, so both of these slides that you can see here, they're taken from your, deck, Nick, that you presented at Brighton SEO a few weeks back. And I just wanna say upfront that you can grab a copy of the full deck that Nick created for his presentation of Brighton SEO from the Docs tab. I highly recommend doing that because you will find a lot of additional context and useful information there, which is kind of complementary to what we'll be covering during the session today. But, Nick, in your Brighton SEO talk, you highlighted, the shift between or sorry, the shift towards enabling humans to do great work better. So in your mind, like, what does that look like in practice today? Yeah. I think it's a lot of what I've just mentioned beforehand is actually doing the heavy lifting for people that, you know, is laborious. It's causing challenges or brain drain more than anything. What we need from, in my opinion, marketers in the future is to become a hell of a lot more strategic, to become stickier with clients, to, you know, be a bit more innovative. So freeing up headspace by automating some of the boring stuff that sits behind it. Or shall I say the things that can be automated that, you know, is a daily task is how we get people to do, you know, great work better. I think it's very much like we have the mantra of brains and bots, where it's like we want to enable our brains for people to do better with bots, which is the automation side, the AI side, and enabling people to do that is a 100% proactive. Pretty much most of my job at the moment. And, yeah, I could walk on for this for a while, but main points are there, James. Yeah. I know. So, just we'll get into some interesting stuff later on, again, from your Brighton SEO talk about what you can do specifically to your data and how to improve the hygiene there and also looking at some promptings in your own practices. But while we're we're still in the the kinda introductory phase here, Chris, I'm gonna pass over to you on this. So if we're talking about expectations and and stakeholder management, there's a lot of headlines out there. And I think it's very easy for some people, sometimes outside the marketing department, to start to think that everything can be automated. And, you know, why can't we just replace everyone and then have AI, work some magic? But as you can see on the slide here, it's not magic. So how would you go about or how would you approach setting realistic expectations with both clients or internal stakeholders if you work in house, about what AI can achieve and what it can't achieve right now? Yeah. I think some business leaders at the moment either think AI is gonna solve all their problems or they're really, like, nervous about it and don't wanna do anything at all. I've not seen many people that it's actually in, like, the middle ground, but, the way that I'm trying to approach it now is that, you know, trying to tell these these business leaders, the the clients I'm going to, don't expect, like, magic and fireworks to happen when you go in there. And equally, for a lot of companies that I've worked with historically, there's so much, like, red tape. For whatever reason that is, they just don't want to put sensitive data into any of these machines and AI motions and the rest of it. But, really, I would just the way I kind of, go through it with them is is find what those pain points are before kind of just slapping, you know, Gemini into anything or, you know, Nano and I and whatever that is. Any tool that's AI focused, don't just slap it in and think it's gonna solve a problem before you know what that problem is. You know, there are some edge cases where, you know, these chat models and AI are genuinely brilliant. I saw them the other day where you can now start to generate three d worlds and go into them and have a walk around, but they're super edge cases. If you look at kind of my world in PPC and and generating copy and and images for clients, it sticks out like an absolute sore thumb when someone's, created, like, a chat GBT image for an ad as I I posted yesterday. So, yeah, I'd always start with pain points. Don't don't jump into thing things too, quickly. And just to echo what Nick said, really, it needs to be human led strategies, I think, best served from from humans, and AI is just there to kind of emphasize the work we're doing, I think. Nice. And, Chrissy, you know, you you're kinda hands on with this all the time with your clients. So could you maybe give us a real world example of a time where you initially thought, oh, this is gonna be great. I can automate this, and I'm gonna speed things up with AI here. But you find that it didn't work out, and, ultimately, it really required that human touch. Yeah. For me, it's anything that's kind of client facing at the moment. Even when that comes down to data, in the early days, I used ChatGPT to really just give it, like, my entire Excel document and pick out things in the data for me. And even simple things that I was starting to put in presentations, I thought, hang on. That's not actually quite right. Even when I challenged it and it told me how great I was for spotting the mistake, it still made the same error. again and again. And it was just it was so frustrating. Actually, like, four x the amount of time that I wanted to spend on this single slide of data. And there's actually something quite, like, nice and freeing about going through the data myself as, like, a PPC nerd and just understanding all the numbers and pulling out the trends that I just don't think AI is capable of replicating, and I'd lose a lot of value by not seeing that bigger picture that my clients need me to see. So I'd definitely be wary of just analyzing data at scale. And anything client facing, I tend to not really use it at all if I'm honest. Yeah, man. I I've had some nightmares with putting CSV files into chat g p t and trying to get it to work with the data that's in there, and it's just it is a headache at times. So, yeah, I feel the frustration there. But as I said, there's a lot to cover, so we're gonna move on. So so, Nick, the next few slides have either been lifted directly from your Bryton SEO deck, or I've kind of heavily drawn on them for inspiration from the content that's on the slide. So thank you for making my life easier and putting such a solid deck together for Bryton SEO. But as part of that presentation you gave, you focused in on how marketing leaders can prepare their data, teams and processes for AI adoption. So if we start off with data, first of all, obviously, you know, looking at the slide here, you emphasized that data hygiene is the foundation of AI success. So could you talk people through the different rules you've identified on the slide here, and why they're so vital for for companies to get right? A 100%. And I think this doesn't just go for, I think AI, but it goes for anything that involves data behind it. Like data in, data out, you're gonna get better results if you do not actually give good ingredients. I've always called them ingredients, especially when talking internally as well. But making sure you got things like consistent naming conventions. You know? AI finds it really hard to read. Much like things like if you got data in BigQuery or any other tables, you know, data repository is that, like, we really need to make sure that we're having clean fields, making sure you've got structured content, and giving examples as well. You generally find that putting data into, you know, an LMM or, you know, providing or layering an LMM on top of, you know, data sources. If that data is fully complete, it's gonna have a really good time, when when analyzing it and a bit better than it has done. Try to put it in JSON format as well, and it it seems to like to read a hell of a lot more in terms of just from a standard standard file. But always include an example as well. I think this is one thing that people do miss. So when you give data to an LM or anything like that, give an example of what the data actually is before, you know, you allow it to analyze. It can have a lot more context when you actually give it an example or give it an understanding of what certain fields are or how they should be used as well. And it has helped us a hell of a lot when actually analyzing client data or understanding how to, you know, work a a bigger analysis and use AI as part of it. So, yeah, I think the biggest thing for me is and this isn't just an AI thing. Clean your data, clean your feeds, and clean everything to make sure that it's a 100% the right web. You can actually use AI to do it as well where it can highlight those sorts of things. Just to on that examples point, so, what what do those examples look like when you're putting them into the the LLMs? It depends on how you how you want to put them in, but quite often, I'm using I'm using, like, a CSV or I'm actually overlaying an LN onto that BigQuery datasets. So there's multiple ways of doing it, but I think the fundamental is is quite often you're just giving it a sheet or you're giving it an adjacent format to make sure that it actually is able to, you know, read it correctly. Yeah. 100%. So if we come on to the next part, so prompting is obviously crucial. And as you can see on the slide here, it is a skill that takes learning and practice. I think anyone who has a background in copywriting like myself has a slight advantage here because, essentially, to prompt well, you do need to be able to write well. Not in the sense that you need to use fancy words or grammar, but more in the sense you need to be able to write with extreme clarity and specificity that's unlikely to be misinterpreted by the LLM, which does take practice and and that is a skill. But there is a lot more to it than that. And I think, Nick, you're going to walk us through that in just a second. Before he does, I just wanted to quickly highlight a resource from Google that Nick has flagged, which will come in handy here. The overview of prompt and strategies, you can see on the slide, and there's a link to that in the docs tab as well if you'd like to check it out for yourself. But for now, Nick, I'll pass it over to you to kinda outline how you would advise people thinking about prompting and where they should start if they if they wanna get better and and build their Sound. Oh, sorry. skills I. thought we were gonna switch slides. But, yeah, I think, in terms of prompting, how how it actually works as well, I there's it's really simple in a lot of ways is to create great prompts like Google, like Google AI prompting guide that James has linked in the chat is actually a really good place to start. But ChatGPT now has one as well. So if you go into the help of ChatGPT, there's a few documents on there and how it's great, really good prompts. As James said, being, you know, having a copywriter background is a 100% gonna help you in this. From a paid media background, we're quite good with specifics and quite good with specificity itself on the outcome to analysis. So the five things that I've told my team in particular to use on the day to day when we're talking about it is define, like, task clearly, understand, like, what we're actually wanting out of them, tell them what the, you know, what the outcome is that you're actually wanting to achieve, Provide as much context as possible as well. Do you know what the background? Is there anything they need to know? Is there anything and I said, very, like, it's a human. I'm personifying AI at this point. But, realistically, it's one of those things we need to buy background off so people, you know, and AI can actually understand a bit more about the specifics of what you're trying to do. Give it an outline as well of what you want it to format it and how you want it to output. It feels simple to say, but I've seen a lot of, like, prompts where, you know, people aren't, you know, giving a structural, an actual outline, or even a reference. Like, if you can provide examples or, you know, putting constraints of the, like, what you actually want it to do, you're gonna get much better outputs than, you know, you would if you didn't give it those guidelines or, you know, those guardrails. And the last one is, like, quite often I find is I'm writing prompts, and I'm essentially using another AI to iterate and re and, like, refine them. Quite often, I'm using, if anyone's used gammadot, gamma.ai, I think it's called, which is a a deck builder. I'm actually putting the prompts through chat gbt and then tune it to Gemini and then throwing it into Gamma to get my final piece of work and then passing it back to ChatGPT to do the heavy lifting of making it into brand styles and fonts as well. And this is all by using just, like, having a clear outline and vision of what you're trying to achieve and creating almost a prototype in your head before you actually start building. But, yeah, these are the five things that I'd a 100% say as a start. Just start doing when you work at Brompton. Yeah. Definitely. So if anyone wants to take a quick screen grab of that before, we move on, I'll leave that slide on the screen for just a second. But, Nick, once a business has got a firm handle on data hygiene and prompt engineering, you identified three main ways you can empower your teams to enhance their work, which we can see on the next slide here. So give them the give them their copilot, arm them with tools, and bring in automation. And just before I pass over to you, Nick, I just wanna caveat this here and say that due to time constraints, this is going to be a bit of a whistle stop tour through these aspects. Nick goes into much greater detail on each of them in his Bryton SEO deck, so definitely check it out if you're interested to learn more. The main intention here is just to give you an idea of what the main pillars are, to focus on. So with that said, Nick, could you kind of briefly talk us through each of these and what they look like in practice in a marketing team? Yeah. I think the actual the actual deck went through is have thirty minutes to do a lot of this. In in general, like, we we use Gemini internally. I shouldn't also call myself a Gemini girl at Brighton SEO, which makes me feel quite sad inside. But, I'm I I actually love chat GPT, but I think giving everyone a copilot, yet again, very simple thing to do. But making sure that you're arming your teams to be able to, you know, just ask. Like, before, you'd ask a team as we're working a bit more fractured and, you know, hybrid as well where you may not always be in office or get quick replies. Give them a copilot. Get them to, you know, ask questions for you, but also keep a human on and do QA it with the team as well. I think one of the things I didn't mention is anything that goes through a chat GPC or an LOM, always sends check for work. It can sometimes get it wrong. The second is Armway tools. I use NAN on a daily. I absolutely love it as a platform. For those that don't know NAN, it's, or as Chris likes to call it, NAN. It's a tool that allows you to create automated workflows where you can include AI agents or monosol to essentially create some critical workforce. There's AI there's AI mode in Google, and the other bits that allow you to do that elsewhere. But ARM people with tools to allow them to speed up parts of their roles that are causing the biggest headaches. Yeah, we as I said, we use AI Studio internally now to try to, you know, look at search term reports and the bits like that to speed people's time up. But I'm with the tools and allow them to go there. That can be even platform tools as well. I do Merchant Center now has, you know, the product studio where you can change backgrounds and other things like that. It's just making sure people give them time to actually do it and that you have advocates within the team or yourself. And Boston is bringing automation. My god. I have a thing that saved my time in the past couple years is bringing automation into Google Ads accounts. Well, that's using an element that says, you know, help me write scripts. Like, I've got background JavaScript. I won't send the best code in the world. So I use it to help me do that. But also, like, it says, you know, I suppose, like, create automations or, you know, create automations for your clients that are actually going to be useful instead of it just for you. So, yeah, those were three whistle stop things that I would definitely do. The full deck has a bit more information in each of these sections if anyone does wanna ask me any questions after this on that. Yeah. I've just I've just dropped the full link into the chat there, for for yourself, Lisa, and anyone else who wants to to check that out. But before we move on from this, slide, you know, Chris, based on this kind of framework that Nick has outlined, have you got any kind of go to checklist for determining whether one of your clients is truly ready to implement AI tools and and automation? Yeah. I think that the biggest thing for me when I'm talking to clients is, like, do they have a problem that that a human couldn't fix? As in, are they just chasing something that isn't actually gonna help them out and just waste a load of time? The way I always like to think about it is, are they gonna have to do something once that will free up their time forever, or is this problem just something they're gonna have to continually battle every day? Everyone's probably, like, super time poor. Right? And I think AI is here to help us with time, but it's if it doesn't do that and if it just means we've got. more things to check that aren't broken, then that's always been my number one, is that it has to have the goal of, like, freeing up our time and not messing up anything else. Some of the other quick things are just like, what's the the quality of the data like? I think not Nick touched on it before. The quality of the data has to be good in their in their back end in terms of what they're getting in. Do they have data layers when they're pulling through, like, the ecommerce, like, ad to basket, ratios and all that kind of stuff. And is it, like, client facing as well? So a lot I know a lot of people will quite happily put headlines out there and descriptions out there in their ads. I'd always kind of say, get it to help you with the ideas, but actually then tweak them yourselves with that human led approach. You know, take out those and dashes, take out the Americanisms. It can still be really easy to spot AI for sure. And just have a look at if there's any tools that do these things for you. You know, building things in a in a n. It's really time consuming. It does save a lot of money and time in the long run. But, actually, are there tools already out there that exist that you can also then potentially jump on that's maybe, like, eight or nine pounds a month that that do these for you as well? Nice. So just in the interest of time, very quickly, I just wanna go over this, relatively recent study published in Axios. The link to read the full thing with all the details is in the docs tab. They go into great detail on how the team there distinguished between human created and bot created articles because it is tricky. They did develop a pretty robust methodology to produce a best gas estimate, but even the researchers themselves acknowledge that a definitive and exact kind of AI made content online isn't possible. With today's tools and definitions, you can just get a best guess approximation. Anyway, with that caveat aside, this is just to illustrate that the amount of new articles generated by AI is increasing massively. If I was a betting man, I would put money on the fact that we're going to see that pink line continue to rise next year and significantly overtake that green one, representing the articles that were human created. A 2022 report from Europol estimated that 90% of online content would be generated by AI by the '26, which is pretty crazy. A link to the Europol study is in the docs tab if you wanna check that out in full as well. But the reason I'm highlighting this is to show you that we really have entered a new era online now. And for better or worse, bot activity of all kinds, good and bad, now dominate the digital world. Nowadays, you just have to look at the comments section of any, popular YouTuber or celebrity's Instagram post, and you're gonna see a lot of comments there which look suspiciously like bot activity. And we're still only in the infancy of this new era that we're moving into. So in this new landscape, it is important to be mindful of the potential threat that this increased level of bot activity poses to your paid search and social campaigns, because we're seeing at LUNIEO firsthand how much money can be lost when high levels of IVT or invalid traffic are allowed to persist for a period of weeks or months. So I just want to very quickly present the results of some recent data analysis we did for one of our clients who took part in the initial beta test for Google's AI Max, which rolled out in May. So if I just briefly give you the context here. So we looked at data from one of our larger clients in the luxury retail space, where the entry level starting price for their products is around the £500 mark and all the way up to the $10 mark on the high end. So this business, I've been using Lunio for more than twelve months prior to January 2025, and that had cut their overall IBT rate in half across the Google search campaigns, bringing it down from just over 7% to a stable baseline average of around 3.7%. And from January to May 2025, we saw a steady maintenance of that baseline average. But then in May, of this year, they took part in the open beta that Google ran for AI Max. And as part of that, AI Max was enabled on around half of their search campaigns. So this gave us an opportunity to take a look at what the impact that had, on their IVT rates. So in total, eight ninety five search campaigns were included in the analysis. 404 of those had AI Max enabled, and four ninety one of those did not have AI Max enabled. And our data science team performed some statistical analysis on both those sets of campaigns to assess the impact of enabling AI Max on IVT across search campaigns. And the findings are presented on the next slide here. So in short, for search campaigns with AI Max enabled, average IVT rates rose from three point seven percent to five percent with highs of up to six percent recorded. So when we look at that change in the baseline IVT rate from three point five to 5%, that represents a 35% increase in the amount of IVT those campaigns with AI Max enabled were exposed to. And for our clients spending large amounts, like this particular client, that represents a significant increase in wasted ad spend. We've actually written a full article on our blog detailing the analysis we did, and exploring the increase in IVT exposure caused by AI Max. So if you're interested in reading that, you can check it out in the docs tab. But before we move on to your next topic, I would very quickly just say if you are planning to enable AI Max or you've already done so and you're spending quite heavily on Google Search, it's well worth. auditing your traffic to see what level of IVT exposure you're dealing with. Because ultimately, until you look under the hood, you don't really know how big of a problem you've got. And once you've got the data in hand, you can then decide whether it makes commercial sense to try and address that. Depending on the industry and market you're in, an audit may reveal that you're only dealing with 2% or less invalid traffic on search, which in most cases is gonna be less of a concern. But if your average IVT rate is consistently hitting 5% or higher across search, that's usually, there's a strong business case to be made there to do something about it when you run the numbers using your average CPC value. So if you are interested in ordering new traffic, you can find a link to that in the docs tab, and our team will be there to help everything or to help you get everything set up. So now we're gonna come on to a really interesting, part of the presentation. Nick and Chris are gonna walk us through some real world examples of tools they built with the help of AI. So I'm gonna do much less talking here and pass over to them. So, Chris, we're gonna start, with you here in the Script Sensei custom GP built by the legendary Nils. I never know how to pronounce his surname, and I butcher it, so I'm not even gonna try to attempt it here. But, Chris, can you let us know what's the or how you've used the script sensei and why you think it's a useful tool for people? Yeah. So I I was lucky on on a to to meet this guy when he did his talk, and I'd never heard of the the script since then before. I absolutely love the name, by the way. And he talked it through and actually made sense as to not just for myself, but particularly if there's, like, juniors in your team or people that are just getting started with the scripts. I think this is a really good kind of bolt on the chat GBT to to give a go. I think it's completely free. And what it does is it basically just talks you through, right, what are you trying to achieve with the script, and then it goes out and helps you build a script that's not just a script that anyone can use for any platform, but it's almost, like, dedicated and ready for Google Ads itself. So it can just help you guide with some ideas. Maybe you're looking to, you know, do a script that changes campaigns at a certain time of day. And for certain campaigns, they need to switch on or switch off ad copy. Like, the list list is endless, like, the scripts. And I know that there's a lot out there that are kind of premade scripts that you copy and paste and then change a few parameters. This one, if you tell it all the kind of parameters and insights that you need to upfront, it'll just bake it in, ask you a few questions about it, and then go and build it. And it's been, yeah, really useful. I think a really good tool for anyone who's just starting to use Google Ad scripts more broadly, but needs something that's more customized than just a copy and paste job. Nice. And, Chris then, I don't know whether, I think this might have been built completely independently of Script Sensei, but I know you put a lot of work into building this tool for your clients who can't do marketing, which allows people to track brand mentions across, LLMs like ChachiPT and Gemini. And for the audience, there is a link to Chris' LinkedIn post that you can see on the slide here, detail in the tool in the docs tab. And, also, I'll just say that, the link to the Script Sensei as well is in the docs tab. So be sure to click on that and bookmark it. But, Chris, I'll pass over to you to talk us through how you built this, why you built it, and what people might learn from from the process you went through. Yeah. I I guess, this feels like a little bit of a shameless plug, but, like, when I was going through and talking to clients, they're always gonna how where am I showing in AI? How am I showing? And at the at the time, there's loads more tools out there now that do very, very similar things. But at the time, there's only really one other provider that provided a service to really pull AI results, and it was hundreds of pounds a month. I just thought there must be a better way of doing this. So, I used NAN, which Nick kind of intro'd me to. And it was really useful because what you can do is take data and insights from different chatbots and stitch them together. So in this example, what I decided to do was ask, Gemini and ask Chat GPD questions like so I pulled this example for a 0.com, which is where I used to work. I used to pull things like, where's the best place to get a washing machine? Would you say AO is a trusted electrical retailer? What are the top, electrical retailers in The UK? All those kind of questions, brand and generic, and then basically stitch those into a look at yeah. Into a a Google Sheets and then into a look at Studio report, which is what you can see here. Basically, it just stitched all that data together to understand how many times a brand showed, was the brand sentiment positive, where were the gaps. So it's a little bit like SEO in terms of, you know, what were the keyword gaps, what what queries did we, prompted a negative response or no response at all, and what queries actually favored competitors or also what competitors showed. So it's a really good starting point, to build this. And it's not that expensive. You know, when you start to get into the like, I didn't know a lot about APIs and and webhooks, but LAN was a very drag and drop software. You can code if you want to, but, actually, if you flip it around, you can just use a very drag and drop kind of diagram builder if you like to say, right. Put this query through here, then take the data and put it over there. So that's been really useful for me, and I'm actually rolling this out across, a lot of the clients I work with now predominantly on on the paid front just to see what that landscape looks like because I think it's it's informing clients now what to do next with blogs or where to focus on even from, like, a customer service point of view. Oh, I didn't realize that ChatGVT thought this. We need a blog post about that or to fix that problem there. So, yeah, as a starting point, it's been really useful. This isn't available kind of publicly, but I'm more than happy to have a chat with anyone about it who wants to kind of see the workflows that go on in the background. Yeah. For sure. Yeah. Be sure to reach out to Chris if if that's something that you're interested in. And, yeah, Nick, maybe maybe he will make it public in the not too distant future. We'll keep our fingers crossed for that. But we're gonna come on to the three, kind of examples from you, Nick, of things that you've, been involved in creating. So first of all, we've got the Alexis chatbot that you and the team at Impression put together. So can you talk us through the problem that this was designed to solve and how other marketers or agencies might be able to replicate something like this for themselves? Yeah. For context, it's, Aaron Aaron Dix, our tech director, but is the main owner of this. But, yeah, it's it's really impressive. But the reason we decided to start creating, element of the load in our dataset, which is all housed in BigQuery, which pulls from all of our clients. Obviously, it's, you know, GDPR compliant. Got a white background number. They also post, you know, external data from from the web as well. Webcasting, you know, hashtags and, you know, usages on TikTok. However, it's now certain trends that we're seeing across Google or, you know, other platforms. It also has been this. One of the biggest things that we found is that was causing our team challenges is that it takes an absolute lifetime to pull data and to query in a way that, you know, allows you to see it quite quickly, especially around client calls and clients are asking about their product procurement, are they understanding why things have dipped or why things have spiked. Or even in this case, as you can see on the right of this, where our talk at, you know, like, where an analyst sat within within a dataset and seeing whether, you know, it was anything that we needed to query. But, yeah, as as I mentioned in it, so I went back to the the base of my Brighton deck as this, where it's we actually worked quite heavily to incorporate a lot of different datasets, within our one of our BigQuery stack. We have used multiple different parties to to, you know, pull that data in. But we now have an in house dev team that do it for us, which is, I suppose, incredibly useful being an agency of our size. But then we've uploaded a simple chatbot on top, which I believe is called Bright Chat, if anyone wants to walk into it. But we've essentially started laying in and adding tools in via MCP, which gives access to adding a BigQuery. It gives it access to things like, you go for another data sources like that that are built via API, and it's allowed to still integrate. It's really cool, setup where the team can use it, and it enables them to find results quite quickly instead of having to troll through a lot of different data sources and unionize them together. It does it for them. It's still a work in progress. We still have challenges with it, but it is getting there, and it's becoming incredibly useful for us internally. Yeah. I can see that that has the potential to be a massive time saver. You know, obviously, as you said, there there's always problems that need to be ironed out, but, yeah, massive potential with that. And this is, one of my favorite names for Atul ever, Nick's non shite music recommendations engine. So, again, Nick, if you could give us the rundown on this one, you know, it's not strictly marketing related, but I think that the principles that you're applying to build this can definitely be, cross transfer to to marketing related projects you might be working on. So, how did you go about putting this together? Yeah. For context, I absolutely cannot stand Spotify. It literally recommended Dolly Part the other day when I was listening to Tyler Childers. If anyone knows country music, they're completely they're they're they're miles apart. But, yeah, one of the things I wanted to do was create my own, my own sort of, like, recommendation engine, which actually is housed in Google Sheets and is connected in via, Apps Script, into things like Spotify, into Last FM, into Apple Music, and a few other places. But, essentially, I've got a status. I listen to one new album a day, which sounds smart when I say it out loud, but I absolutely love music. And it essentially allows me, which you can see on the right, for instance, the 321 albums, and I took the screenshot for James. But it reads the date that's in that spreadsheet, and it looks at the vibes, and it looks at, you know, the the other genres that exist within that. Then it recommends me albums that I've not heard before, and it gives me a reason why it's recommended it. Then it crosschecks it against my Spotify to see whether it's an album I've listened to recently, and that's all connected via the API. I built this by essentially using app scripts and then n a n, to sort of speed up the process. But it's a really simple build, and I actually just asked ChatGPT how to do this initially and then just learn the time a bit on YouTube of how to build and connect into APIs and getting our access tokens and other bits like that. So, yeah, it's if anyone wants more information on this, I'm more than happy to share it with them. And if you want to see the back catalog of the past three years of albums I've listened to, you're more than welcome. Nice. Yeah. Shout out to emo being the top genre there. I don't know if you've ever heard of the the guy nothing nowhere and his his, Will It Emo album, but he's been, covering popular songs in an emo style, and I've been a big fan of that. So, if you haven't checked that one out, Nick, I recommend it. But we'll come on to the final one, and I hope you're not, offended by the image that I've gone for here. But I didn't have a screen grab, so I just thought, you know, we're gonna slap on the Yorkshire tee. But, for those that don't know, could you explain what the Northern agent is. and and how you created that? I actually forgot to give you a screenshot for this, James. Apologies. But, yeah, long story short, I I have to QA a lot of my own emails. I'm an incredibly blunt man when it comes to writing emails, and I absolutely am not the biggest fan of writing them. So I've actually done this. I've got a agent that goes through an AM where it essentially reads my emails before I send them and sit in draft, and it puts my tone on top of it, but also tells me where I'm being a bit of an asshole or being a bit too harsh with people. And, yeah, it's called my Northern Asia because the way that I write emails and whatever I write in general is very much how I speak. It's an incredibly, you know, northern slang. I'm from Leeds. It's something that I've actually really enjoyed, something that I don't feel, I believe. But when I've QA'd emails through ChatGPT before or, you know, Gemini, it completely strips out any form of personality from them. So I built an agent that has a brand guideline in the back end of it, with an AM, within the memory of it, and, essentially, QAs it as if it's me QA ing it, which is really, really useful when I've got, you know, a frustrating email that I'm writing and trying to be a bit kinder when writing it as well. 100%. So, that's a a solid use case for sure. We're just gonna move on, and I'm I'm conscious of time here. So we'll see how fast we can get through these final few slides here. What I want to do here is, for Nick and Chris or, you know, one of you guys, either or, I'll let you decide who answers this one. But if you could give any generalized advice or guidance, to people, you know, for anyone who wants to to follow in in your footsteps and build some of their own tools, any generalized principles that would help them. And then after that, we'll come on to this warning that people can see, on the slide about AI assisted reporting. So, I don't know, Chris, if you wanna jump in on that in terms of guidance, so what people should do if they wanna build their own tools. Yeah. I guess the first thing is to to understand the problem, first and foremost, and be curious. Like, be commercially curious. Like, I think about my old boss or clients and their bosses. They they don't really care about PPC if it's generating, like, CTRs of, you know, 10% or CPCs that are down by 50 p. They're just really wanting results. So anything that I'm doing, I always start with the end in mind in terms of, like, what is this gonna enable me to do? Is it gonna mean I can free up two hours that I can then spend on strategy? Is it gonna mean that, reporting is better so I can actually run it through this report through different chatbots that kind of validate that. And it always start with the end in mind, like, what do you actually wanna achieve, and be patient as well. Like, ChatGPT will get it wrong countless times, and you will need to, you know, help it learn countless times before it gets it right. But once you got the finished product or script or tool in place, it can be really, really good. But, yeah, it's definitely not a not something to just hand over and leave it. Yeah. And then, Chris, maybe could you just give us a few words of warning to to kinda emphasize this point that's on the slide here? I know we spoke about reporting being a really common use case or application of of AI and performance marketing, but there is a a line to be drawn here. So what would you say to people on that? Yeah. It's a really interesting one. I would when I'm approaching reports now these days, I've actually got a a tool again. It's in n a n. But what it does is it is it'll run that same query or same report through three different models, and then it will come back to me and each three models will give me an answer. And if those answers are any, are any different from each other, then they start to quiz each other in terms of where did that data come from, why is this not matching up, etcetera, etcetera. And, hopefully, they'll come to the right answer and agree on all something. It's almost like a committee or like a voting procedure. All three have to come back to me with the same answer, and if not, then questions are asked, basically. They're all in trouble. So the way I do it now is they're almost marking each of those homeworks. And, you know, I still, to be honest, on occasion, have to just sense check and double check the data for my own sanity. But more AI bots are better than one is what I would say. And if you've got different models or different providers giving you the same answer, that gives me more confidence that it's generating the right results. Nice. So just in the interest of time, I'm gonna skip over one slide that we did have, and I'm just gonna move to this because I feel it's probably directly relevant to a lot of people in the audience. So, Chris, I know, you know, when we've been preparing this, I know you've you've had quite a few pointers listed down about how you expect things to change in terms of, you know, Google Ads interface and and meta ads and what people are actually doing in the ad platform. So maybe you could give us a quick rundown of the kinds of things you're expecting to see within platforms like Google Ads and Meta, next year. Yeah. I don't think it's gonna be anything as drastic as keywords are gonna disappear, but I definitely think this is something I said, at SCM stories, which is great event, last month was, I think. queries are gonna be even longer. They're gonna be, ten plus words. And I think now at the moment, if you put a 10 plus word into Google, it's gonna go, no. There's not enough search volume for that. I'm not even gonna bother showing it. And, you know, we all know in some cases, it still does. But I think as people become more, like, comfortable sharing more information and also with AI mode where they can follow-up more information, search query is gonna be longer, which means that there's a big opportunity then to actually be more specific with our targeting, whether that's the headlines, whether that's the type of pages we're bringing people to, and all that kind of good stuff. With the creative side of things, think a lot of these platforms, MetaGoogle, we've already seen it, are gonna allow you to do AI stuff built into the platform, which is free. There's been some great use cases, simple things like adding backgrounds to product images, adding lifestyle images onto things. They can work great. Obviously, I wouldn't necessarily use them by and large, but on certain occasions, they have worked well. And I think that will only start to just accelerate, as they try and basically all these platforms are going for people that just want to set something up really nice and quickly. So the big kind of warning there is I would still be all over the detail, all over the settings. We all know that the horrible location setting by default is is, like, target and interest in. And I think. Google is just hoping that we just jump on and set things up in their way to spend a load of money, and, actually, we still need to be very aware of that. But without a doubt, AI will make a big impact to how the marketing of the next year looks for sure. Yeah. Do you do you reckon that, we are gonna see the say goodbye to DSA campaigns in the wake of AI Max? Oh, I think I think next year maybe. I think it's like, it depends on the rollout of their side with AI Max. If everyone's using it and no one's using DSA, they'll just wipe it. But I I know there's a lot of DSA DSA funds out there still. Yeah. Yeah. We'll we'll we'll wait and see what happens on that. But I know, Nava Hopkins, previously have optimized her in our Microsoft ad. She's been waiting on the deprecation of DSA campaigns for the past twelve months, so they're still hanging in there. Still using what they still can in 2026, but we'll we'll see where we get to there. So before we move on to the q and a, I know, obviously, Nick and Chris, you guys live and breathe this stuff every day, and there's a lot of stuff that you guys will be aware of that people in the audience and and myself, you know, aren't gonna be, aware of or or paying attention to. So before we move on to the q and a, I just wanted to open it up to both of you to kinda give a shout out to any particular content creators out there or any tools that you find really useful that we haven't mentioned so far. And just for the audience's sake, myself, Nick, and Chris did speak about this before we obviously went live here today. So I've tried to round up as many of the things that they've mentioned, and include a link to them in the docs tab. So hopefully, most of the things that are going to be mentioned, though, will be referenced there. But I'll pass it over to you, Nick, first to talk through anything you'd like to shout out. Yeah. I think there's two creators that are more YouTube based that I've that I follow and I've listened to probably on the on the daily, if not, yeah, if I listened to them this morning. So there's Tina Huang, who is an amazing person to learn AI agent fundamentals and to understand how to use, things like NAM and, you know, other tools like that. And she's she's just she's fantastic at explaining things about, you know, making it too too complicated. And the other one, if you've not heard of as well, is Jeff Jeff Sue. Yeah. Again, very similar, but it's more on the automation side and is very much an advocate for, you know, no code but still having the the human input. So those are the two that I highly recommend people to check out. I haven't given you the ones correct. Sorry, Chris. James. I will, grab go grab them and drop them in the chat just so you can have a look at them, of their channels as well. But, yeah, I think I'll pass over to Chris because Chris has an absolute plethora of his. Yeah. Chris, jump in there. The main the main one is this one on screen here. So so Kenny Skelton, who's an absolute legend in the digital world. I used to work with him, at Mediacom. He's now a place called Choreograph. He's actually I mean, me and Nick have got our own podcast, but people can listen to two podcasts. Right? He's got a brilliant three? podcast. What three? The All three or more. podcast as well. Yeah. There's loads of podcasts. Yeah. Yeah. All three. Yes. Of course. Yes. It depends how time poor we all are. Right? But, yeah, these, these they've recently set one up called Punks of AI. It it's just really funny but also educational. So I really enjoyed it. I think it's worth a shout out for anyone who just wants to give it a go. I think on his last one, Kenny built a tool that, pulled out five ingredients. You you got, like, food heaven or food hell, and then he bought them out on set for people to have to, like, make a recipe with or taste or something. It's it was funny. So, yeah, go and give that a go and give that watch, I'd say. Nice. Yeah, there there are a few other tools that were mentioned there. I know, Nick, very quickly, is it you that uses Descript? It's actually Chris that uses Descript, to currently edit our podcast. We may be moving tools, but, Chris, I'll I'll give a floor to you to to mention Descript. Yes. To be honest with you, it saved me so much time. Like, I am not a video editor, and hats off if you are in the editing game, because you know a lot more than me. But, it has enabled me to do so much. So it's got something called Underlord. And, basically, what it means is if you've, if you've messed up, which full transparency, me and Nick do, like, take retakes of things all the time, It kind of knows when you're doing the retakes and strips out all of, like, the first take. It can remove, like, filler words, and it can it can do loads of things. It can do, like, multicam shots so it knows who's speaking, and it knows when to get a reaction shot from the other person and flip it back and forth. So if you are watching any of the the prompted podcasts, all the kind of the shots one, shots two are all kind of completely dictated and orchestrated by the AI on there. And it it's quite good. It can also, like, interject and put in words if we want to change, like, Manchester for Liverpool or we want to change Gemini for track GBT. Because it's obviously listened to hours of our voices, it can in like, change the words for us and and put that into the into the, the podcast as well, which is kinda scary, but there's loads of stuff on there. It does. So it's been it's been really useful for our podcast so far. Nice. Yeah. Something that I think, we might need to check out, at Lunio and see it might, speed up things for us. So I am going to, just stop sharing the slides there, Noy, and that's it for the content. And with the short amount of time we've got left, we might run over by five minutes just to around five past five. So we'll go through the questions that are in the q and a's tab. If you haven't submitted a question yet, you can still do so now, and you can also upvote the other questions that you're interested in hearing an answer to. So, Nick and Chris, I'll leave it up to you guys, in terms of if you want to respond to one of these or if both of you wanna respond, that's fine as well. So the, first question that we have, I'll share on screen here. As Chris shared AI, especially chat GBT and Copilot, isn't always right with the data. Even taking an average of the last seven days spend was coming out wrong on both places. What is the right way to use it so the outcome data is right, or do we do these manually and not try to rely fully on AI for these kinds of, data related tasks? I'll I'll pass it over to you guys. I think I might I might have answered this one already with the workflow that I'm using. So at the moment, I will just basically really condensed version is is kind of get it to challenge each other with the workflows. Different chatbots will kind of have to come out with that same answer. I know now that the workflow methodology that NAN kind of is built around is also coming to other chatbots, though. Is that right, Nick? Yeah. But it's, I believe it's coming to ChatGPT and Gemini. I believe in the future, Gemini already has AI Studio, which is starting to do that but with less autonomy. But, yeah, I think it is coming to others quite soon. I think it might be next month, Right. actually, but we'll see it sitting in charge of EPT. Nice. Nice. So we've got, this is an interesting one, and I feel like, some other people might be in a somewhat similar, situation. So it's a question from Lisa. I have a CEO that wants me to cancel my contract with our agency who does everything from graphic design to video production, and he wants me to use AI for everything. Am I, one purse I am a global or sorry. I am a one person global marketing team, and I need advice on how to head this off. So if anyone out there is in that kind of situation where where they find themselves where the senior leadership team or or the CEO of the company is pushing them to get rid of resource and and just automate, how would you go about navigating that tricky interpersonal situation, I guess? can I can jump in to start with on this because I think it's a it's a it's a question we have as an agency when we are chatting to people and people could automate all of this? If it's creative in particular, just show how bad the creative can be when you're doing it en masse. Show the flaws of AI or elements or whatever you're using. It's not the show if I was to keep a team, in my opinion. I think it's also showing where it's actually really visible and useful. And, like, I think maybe speak to your agency and see where they are using it and see whether it could even be a bit you've got to send them with a new agency that is, you know, a bit more AI or LN in tune or using automation, which then allows you to get more value out of what we are doing. They are, I think, the show of a force that AI has. We're all it's not perfect. And I think if people think it is, it's surely it's misplaced. Chris, I'm not sure if you have anything to to add on there, but I. think show show where 100% about. agree with that. If you just look at, like, the latest Coca Cola advert, which they claim was by AI, no one's just gone into a script and some AI made me a Coca Cola advert that's good enough for TV. Right? There's been probably hours and hours of refinements and tiny detail that's gone into that. There's been human led. And I'm at the very the same conversation, like, great if you move to an agency that's very AI orientated, but there's never gonna be a solution at least so I can see for the future that's just like pure only AI. And for me, maybe it comes down to, like, a commercial decision. Is it that the the agency is too expensive? Is it they're not doing enough? What are the what are the KPIs? What are the reasons? What is he wanting he or she wants him to do by by moving to AI? Is it just a costing? And, actually, yeah, I don't think there's a solution even if you paid, like, an AI company, like, tons and tons of money, they're not gonna be able to produce anything near the same level. So, yeah, that would be my 2p on it. Nice. Yeah. Solid advice there. So we got time for one more question, very quick one. It it's, from Simon, and it's I know we've already been through this. Can you guys please share a list of AI tools that you guys use? But I just noticed that there were, a couple of tools, I think, used by you, Nick. So one being make.com, which I don't think we've touched on, previously. And, also, Nick, you did mention, in one of your answers that you're using, Google AI Studio. So to kind of close this off with the q and a, if you could kind of talk us through make.com and the kind of things that you're doing with Google AI Studio. So the quick one, make.com is essentially compared. It's anan. So you can use either. There's so many different automation tools out there. I like nan, but I have used make.com, when it was on free trial, and I've still used it for some stuff at work. But it's an automation platform that allows you to incorporate AI and AI agents into workflows. So it'd be useful. And then Google AI Studio is where prompting becomes incredibly useful. It's essentially a no code by coding platform, run by Google that allows you to create simple apps and web apps, that allows you to input data. I create a little bit of data that allows you to change people into Victorian portraits just from a very simple prompt. But it's all prompts there, so you're not having to go into the minutiae of intricacies of nnanrm.com. It's all proper. So, yeah, if you wanna check something out that could potentially do a bit of workflow, like, I e, work through, you know, search term reports and bits like that, via doing a CSV upload, get Google AI Studio. It could be a good show. Nice. Yeah. Definitely. Well, we're checking it out. So that is it. And I think we'll need to wrap up there so so we don't run over time for for too much. But I just wanna say a massive thank you to to both you, Chris and Nick, for joining us today and and sharing the insights and advice that you guys have have, you know, developed from from your experience there. And before we wrap up, I'll pass it over to you guys if there's anything you'd like to say or mention before we before we finish. No. I think thanks everyone for joining. Go and, go and listen to our pod, Shameless Plug, prompted podcast on YouTube and Spotify. Right? I think that's everywhere, is Nick? Yeah. So go and give it a go. Thank you very. much. Good checking out. Yeah. If you have any questions for us, you can contact me directly via james.dini@lunio.ai, and I'll be happy to help you or point you in the right direction, if I can. As As I mentioned before, if you're interested in auditing your search and social campaigns and figuring out your average IVT rates across search, across PMax, across demand gen, you can click on the link in the docs tab and our team will be there to help you out. Lastly, big thank you to everyone in the audience, tuning in, whether you're here live or if you're watching back on demand. We hope to see you again for the next one. But for now, take care, and best of luck with all your AI experimentation over the coming weeks. Cheers, guys. Cheers, Gotcha. guys.