Why AI Is Quietly Sabotaging Your Marketing...
In this YouTube presentation Elliott Botterill warns marketers that misusing AI can dramatically hurt campaign performance. He illustrates the danger with a client case study where a sudden drop in landing‑page conversion was mistakenly blamed on copy, leading the client to replace the page with AI‑generated content. Botterill shows that the real culprit was a newly‑added, cumbersome form—a change that AI missed because it lacked the business’s contextual knowledge. He then outlines a repeatable diagnostic process: maintain a weekly, color‑coded scorecard (rolling 13 weeks) that tracks spend, leads, cost‑per‑lead, sales, etc.; layer business context onto the numbers; identify a single constraint; test one change per week on a fast cadence; and log every change. He reinforces this with three “golden rules” – test one thing at a time, act fast (weekly), and keep a detailed change log. Finally, he introduces the 10‑70‑20 principle for AI usage: 10 % human idea, 70 % AI execution, 20 % human editing, positioning AI as a tool, not a decision‑maker. The video ends with a call to adopt this framework to avoid AI‑driven pitfalls and achieve steady, data‑driven growth.
Chapters
Ladies and gentlemen, welcome back to the YouTube channel. In this video, I am going to be talking about why AI is quietly sabotaging your marketing. So if you're a business and using artificial intelligence to help you create your ad copy, create your landing pages and your emails, you're going to want to watch this video because a lot of people are seeing their results get absolutely tanked. And I'm here to show you why. And how to fix it. So in this video, we are going to be covering why using AI can indeed kill your results despite it feeling like it's doing such amazing work for us. Then I'm going to share the diagnostic process that we use before we use AI that enables us to get amazing results so that you can replicate it and do the same. And then I'm going to share the 10-70-20 rule as well, which is our guiding principle when it comes to using artificial intelligence inside our business. So for those who are new to the channel, hello, my name is Elliott. For those who are coming back, welcome back. I appreciate you. Brief little bit of history. So I have been in the marketing world for over 10 years now, despite my young and youthful appearance. I started off at M&C Saatchi Performance working as a search and social executive. Before I quit, started my own business in my mum and dad's front room, grew that into a 7-figure agency with a bunch of awesome team members, and then in 2025 exited that and joined Entrepreneurs Circle, where I currently am the Chief Marketing Officer. During that period, I've had the pleasure of working alongside and with many of your favourite gurus in the space and even speaking at events and blasting some fire. That was One of my favourite moments ever. Within our current Business Entrepreneur Circle, we have a software company, we have a high-ticket coaching programme, and we have got a low-ticket community, our core business units. This is a snapshot of our last 30 days' performance, and we spent about £72,000 in advertising, drove 6,000 leads, 600 sales calls, 6,202 sales, and £1.1 million in revenue. Total revenue. So I share that so you understand the context where this information is coming from, because you should never trust random strangers on the internet. So this channel is all about showing you how we drive traffic, how we convert traffic, and how we close sales— both the guiding principles and the tactics that we deploy within our own 8-figure business. And we share them to you because that's what life's all about, and we want you to have a good time as well. So Artificial intelligence, if not used properly, can be a wrecking ball to your results. We've seen this ourselves when we have actually created campaigns almost entirely with AI and barely made any changes, and actually it performs pretty bad. But just recently I saw this with one of my clients as well. And so to illustrate this point, I must tell you about one of my clients I work with in my secret vault. And this client, they're a coach. They do around, well, they spend around £10,000 per month on advertising. They have a nice webinar funnel. Every single week they do a webinar, spend a couple of grand on ads, fill the webinar, deliver some value, pitch, sell an online program. And they are just having a great time, and they're crushing it and getting lots of results. Now, we've been working together for a very long time. I want to say, uh, well over a year. And long story short, In their last webinar, which took place about a week ago, they noticed their landing page conversion rate decrease. It went from like 15% where it was at to about 7%. Now firstly, this 15% for a free webinar, normally you'd say that's very low. I would benchmark 35% here as a good conversion rate. Key point of context, they have multiple qualifying questions. So we are adding friction to get higher quality leads. So we're actually okay with the 15%, but the drop to 7%, damn, that is steep. And we don't like that. We want to get that back up. So obviously we noticed that drop to 7% and they sprung into action, which is good, which is what a good business owner, good marketer should do. Let's make some changes. Let's get that back up. And so what they did is go straight to Manus, our AI tool of choice, and they fed it the landing page and they got it to rewrite the whole thing. And they presented it back to me. They said, look, Elliott, Manus has had a go at our landing page. It's improved it. Here's all the new structure and copy. What do you think? My response was Chuck the whole thing away. Chuck the whole damn thing in the bin. You see, not because the copy the maintenance had produced was bad— in fact, it was very good copy— but because in this scenario, they were about to solve the wrong problem, and they were about to replace this entire thing with what AI had produced. And so Today is first all about showing how to diagnose what the right problem is and then how to use AI in the right way to solve the problem. This is how we get good results. So you see, the main problem with AI is that when we give it a task like build me a new landing page, it is working off its own knowledge base and its own context. It's up to us to share as much information and detail with it as possible if we want to get high-quality output. But again, the challenge that AI has is it is self-contained in this box, and so it only knows what we have given it and what it can find on the, on the interwebs, okay? And this is where the first disadvantage exists because AI sees your page as text. And we see it as a key part of our system. And one of our biggest superpowers we have as a business owner is our level of context. Our level of context is supreme. And until we reach AGI and AI is plugged into every single thing that we are doing and is monitoring our thoughts, we are always going to have better context than it because we know our business, we know how we operate, all of our interactions with our customers, what's going well, what's in the pipeline, and so on and so forth. AI does not unless we explicitly tell it. So this is why we have to make sure we deploy and utilize our context in our decision-making before just getting AI to reproduce something straight out of the gate. You see, we have weeks of data. We know when we have made changes. And AI frankly has got none of this. Okay. Unless we specifically tell it. So today I want to show you how we diagnose what is actually wrong first so that we can then use artificial intelligence properly to fix it. First, we need to understand the funnel. We are going to have some ads, our source of traffic, that go to our landing page. On this landing page, there's going to be some kind of form, and then we're going to have our thank you page as well. Now, when something breaks, when we look at our stats and all of a sudden we're not getting the results we want, our cost per lead is a little bit too high, where is the first place people normally look? They normally look at the ads because that's typically where they notice the cost per lead has increased. And then they normally look at the landing page. Now, I have observed that since the rollout of artificial intelligence, people default to these two things more than ever because actually new ad copy and new landing page copy AI can produce in like 30 seconds. It's an easy fix. So we actually use our own internal bias and practically project that that is the constraint because we know we can fix it really, really quickly. and in some cases it is just not, and this is where our context comes in because in this particular instance with my client, they were thinking the landing page was the issue and I had a quick little look at the scorecard and at the page and lo and behold, the culprit was not the landing page, it was the form, and I was able to get there because I had applied my extra layer of context. That I had. So here was the decision-making. These guys are running this webinar. They've been running it for a very long time. They had at least, you know, 20 weeks of profitable webinars under their belt. That is important context because we know this thing works. We know the landing page conversion rate was at around 15%, and then all of a sudden it dropped to around 7%. This is the dip that we have observed. Now, what we also know is that they have made some changes because every week they're tweaking and trying to improve things. And we know that the week before, they actually changed the form. So they changed the form to this new form, and the form operated very differently. It had lots of fields that you had to scroll through, okay? And this was a key change that the AI hadn't picked up on. So whilst they made this change and because of the landing page conversion rate dipping, instantly assumed it was the landing page, the reality was very different. It was that they had applied this new form and all of a sudden conversion mechanism became less efficient. The mechanism that takes traffic and turns it into a lead was now not as good as it was before because everything else was the same. The ads were the same, the landing page was the same, and they've worked previously. So why would they all of a sudden break? It was because we changed the form. So therefore, rewriting and redesigning the entire landing page is not going to fix this problem. It's not going to fix the form. So we have to recenter around the action and look at the form. This is the impacting thing. So this is what I mean when I talk about context. You have to be aware of the things that you have been doing, and you must understand how the actions you take on the different parts of your funnel, how they influence the different metrics. Because you need to know for every single metric, how can I impact that? What actions can I take to move that metric to the right place? And if I've made a change, if I've updated the form, what is likely to change? How can I tell if that has had a positive or negative effect? And that way you're able to constantly deduce are you taking the right actions or are you taking the wrong actions? This is just the cycle of being a data-driven marketer. Now, you can only be a data-driven marketer if you have got data. And it's why for all of my clients, one of the first things we tell them to do is get a proper scorecard in place. Now, the scorecard is a beautiful thing. It is a Google Sheet that we manually fill out every single week. We never automate it because we want the ritual of populating these numbers to make us aware and give us a deep sense of accountability. I've always found that when things get automated, people don't have much attachment over it, and therefore stuff doesn't get done. But so the scorecard We are going to track weekly. So every week we fill this out and the scorecard is going to track the key metrics of your funnel. So this would be things like spend, it would be things like lead, it would be things like cost per lead, it would be things like sale, it would be things like cost per sale. And every single week you input what that number is. Now we make this sucker incredibly colorful because we are going to color code it. So if a number is on target or getting better, then it will be green. If the number is off target or getting worse, then it will be red. And that way we can just look at our scorecard every single week and immediately know Which part of our funnel is working and pulling its weight and which part is not? Therefore, which part should we focus on? If I know which part to focus on, that immediately narrows down the set of actions that I need to take in order to fix that thing. Firstly, if my spend is in the red because I've underspent, then I need to figure out why my campaign's underspent. Did I just not allocate enough budget? Have I turned off all my ads? Is Meta throttling my campaigns? What is the reason? Did my credit card get declined? Whatever. If my cost per lead all of a sudden has gone up, then I need to investigate that. Is it because my ads and my cost per click or my click-through rate has increased? So I've got some creative fatigue or Meta's allocating my budget to some inefficient ads that I now need to pause. Or is it the landing page? Did I get the same cost per click and click-through rate, but all of a sudden my landing page conversion rate has dipped? Is that because something's changed on the landing page? Is something broken? Did I make any changes to the landing page? Update the mechanism, the lead form. Is the lead form working? Has it broken? You know, do I need to reset it or something? And then same with the sales process. And we use this to direct our actions. Now what is important about this is you look at it every single week, and over time you have got weeks and weeks of context. We like to have a rolling 13 weeks worth of data on this bad boy. Because it enables us to see the trends. It enables us to see over time how the actions we are taking are moving the metrics in our funnel. And this is key because as a business owner, as a data-driven marketer, we want to develop our intuition. We want to develop our deep-down knowledge of what we're doing and how it actually makes effects and impacts on our marketing. Because we consistently want to be optimizing our week and our time based around The highest revenue-generating actions, the things we can do that are going to move the needle the most. And if you operate in a world where you are not completely on top of how your actions are impacting the numbers that our entire life revolves around, then that is a bad place to be. So we want to solve that, and we look at a rolling 13 weeks, and we develop this intuition and feel, and it keeps us focused on the right key actions. So once we have got the scorecard, then we can actually look at the diagnostic process. Now this is honestly pretty easy, but number one, look at your scorecard. So this is the first thing you want to do on a Monday. My team are responsible for filling in the scorecard immediately on on a Monday at 8:00 AM, and then we have our team meeting, and the focal point is our scorecard. And we look at it, and then we discuss what we did last week. Did that move the numbers in the right way? What's our constraint for this week? And what are we therefore going to do next? Now, once you're looking at the scorecard and discussing it with your team, you need to layer on your context. You have got to think, you have got to apply your knowledge and your experience from what you have been doing and observing within your business to the numbers, because the numbers are one part of the story. You have got to project your opinion and insight on top of it using all of the additional context that you have. Then you need to identify your single constraint. Emphasis on single, because in any business you have got one constraint at a time. And so we need to decide what that is. The one thing that is throttling our performance more than anything. The one thing that if we solve would have the highest impact on our overall results. And it is always one thing. But then it's time to take action. And controversially, we typically want to test one thing, one single thing, because what most people do is they detect a constraint and all of a sudden they mix that with their ideas and boom, they're creating new ads, new landing pages, new emails, new offers, new campaigns, new ideas, new this, new that, new everything, and they throw dozens of brand new variables into the mix. In order to have stable long-term growth where we are just consistently getting better and better and better every single week, we want to test one single thing each time. Because if we test more than one thing, we cannot truly say which thing caused the result. Did this change actually drive performance up or down. If we've changed multiple things at the same time, we can no longer assign the causation. And it's critically important that we do have a level of causation we can apply, because that's the only way you can just systematically move things up. And people rush and they try and do everything, and as a result, they actually completely invalidate a lot of their testing, and they end up spinning the wheel because they don't get the clarity on what is going to move the result And therefore they just do no, no, no, no, no, no, no. And they're in this perpetual cycle of like being stuck and it sucks. So we want to test one thing. So this brings me to the three golden rules. This brings me to the three golden rules of testing and optimizing and making changes. Number one, do one thing at a time. Don't destroy your experiment. We want to establish the causation. This is why you have to get very good identifying the constraints. You're always changing the thing that's going to have the biggest impact. Number 2 is we act fast. We do this on a weekly cadence. Scorecards update every Monday. We have the meeting, identify the constraint, identify the action, execute the action. Next week we do the same thing again. So we're doing one key thing every single week. This is important. Most people are too slow and they think, actually, if I've made this change, I need to allow many, many weeks, or yes, let's assess against next month. That is a death wish. The world we live in is very fast moving, and it's normally less to do with the amount of time something is running and more to do with the total amount of people that see it. So especially if you're spending you know, over £1,000 a week on your advertising, then you should be fine to do this weekly. People fret about stuff being statistically significant, and most cases that is just a silly way. We're better to move fast, make a decision, and let's move on to the next thing, because over time that speed really begins to compound. Also, the algorithms these days learn, they learn so fast. It's a myth that it needs hours and hours It's a myth that it needs weeks and weeks to properly learn before it gets good. It's way more powerful than it used to be, and you see results very quickly. If it's working, you will know. The third one is where a lot of people go wrong, and this is to just log everything. So when you make these changes, keep a log of it. What date did you make the change? We have a big mirror board where we have screenshots of what we've done and like a little timeline. So we can stay on top of them because we make lots of changes. Sure, yes, it's critically important we remember, or our team can pick up over time, which ones have been made and which ones actually drove the results. So we don't want to learn something and then just forget about it and stop applying it. That's again how we guarantee ourselves a steady trajectory versus a shaky trajectory. Final thing I wanted to talk about briefly is the 10-70-20 principle. Which we use now when it comes to using AI. So when we are going to use artificial intelligence to help us create one of these changes, whether it is new landing page copy or new ads, we apply this. And what this means is the first 10% is our original thoughts, our own ideas, knowledge, opinions. The 70% is the AI then doing the heavy lifting, creating the ad copy, creating the landing pages, doing the analysis, whatever. The final 20% for me is the most critical. This is our human editing. So we spend actually quite a big chunk of time now editing and humanizing what the AI produces, and that is where all the magic is. Because if you stick to just AI-generated stuff, its effectiveness is not very good generally. But if you are able to actually apply your own human knowledge to it as well, then that is where the magic is. So more on that in a separate video. The key point for today is that AI equals a tool. It is not the decision maker. And so yes, use AI as a tool, use the 10-70-20 principle to get very good human outputs, but you are the decision maker and you need to use it as a tool to aid you in your decision making, but not to do the decision making for you. And having your scorecard understand the diagnostic process. Having your scorecard understanding the diagnostic process in order to every single week identify the key areas you need to focus on and the specific action you need to take to move that is going to get you moving in the right direction and avoiding the common mistakes that many people have. So I hope that was useful. Happy diagnosing. Don't forget to like and subscribe. I appreciate you. See you later.
