How to destroy your business with AI in 10 minutes

Most small to mid-size businesses are screwing up their implementation of AI. They think AI saves them money, but what it's really doing is it's costing them in revenue. I've scaled two agencies to a combined 160K a month in revenue using AI, and I've been at this since the early models, GPD2 and GPD3. And I can tell you that most businesses are doing it completely backwards. Today, I want to show you three AI mistakes that I've seen that cost companies $500,000 or more over a year or two, as well as obviously what you can do to avoid those mistakes, whether you're implementing AI for a client or for yourself. So for those of you that don't know, I actually built my first like successful company uh on one second copy around some very early language models. These were GBD2 and GBD3. And one thing that became very clear, especially with the technology back then, is that the biggest strength of AI is its flexibility. And the biggest weakness flexibility, which is kind of a catch 22, right? If you don't accommodate for that flexibility, you often end up with just enough rope to hang yourself. And we as a company have done so many times. So what we quickly realized is that you need lane guards or bumpers. Basically like in bowling. Uh you need to corral all that flexibility into specific areas and pressure points where AI actually makes sense to apply. But what most people do is they get carried away cuz AI is sexy and shiny. Then they slather AI and automations into literally every nook and cranny of their business. Then wonder why they're not growing. So let's

get into it. The first major mistake is when you save money to sacrifice revenue. And this one is really dangerous because it does feel logical on the surface and it does feel like you are making some progress as a business. To make a long story short, you do not want to use AI for front-facing sales applications, specifically ones that are high risk and then high ticket, at least not in its current form. Because if you do this, you will probably lose more money in revenue than the amount of money that you have saved in profit. Let me explain what I mean. With the current state of the technology, companies often will look at AI and immediately think, "Wow, all this AI stuff is freaking fantastic. we should implement it in every level of our business and because you know a lot of AI tends to be natural language based the natural step is hey uh why don't we start by automating stuff that is natural language heavy like our sales so then they prematurely fire their entire sales team or maybe their customer success team or something like that and they start replacing them with an AI voice agent or chatbot and you know this makes sense on the surface um customer success managers CSM might cost like 5,000 bucks a month AI voice agent costs 100 bucks a month so on average the FaceTime savings are like 4,900 bucks. But the unfortunate reality is if you're not careful, AI then significantly reduces your conversion rate because it's not as contextually adept as a human being, at least not yet. You know, it'll miss nuance. Uh it'll miss some subtle cues. Could be small like uh I don't know, maybe it doesn't know the date of an inquiry cuz you forgot to include some variable in the agent. Or maybe it's bigger like the person that you are talking to actually reached out 4 months ago and talked to the salesperson or something like that. Or maybe they left a review somewhere. or maybe they use a specific phrase that indicates that they want a manner of being addressed that maybe the AI model doesn't understand. So sales is a perfect example of a mislication of AI because there's just there's too many things that could go wrong, too many reasons why a prospect would be turned off by it. And then the risk if you do end up making any of those mistakes is really high and it impacts the most important part of your business, which is your revenue. So, a real example that I've seen with numbers, a business owner saw an AI chatbot on the internet, one of the similar channels to mine, I suppose, that talked about how great AI chatbots were, and they thought, "Wow, I could fire my setter and I could save myself $5,000 a month. " Which is obviously understandable in theory. $5,000 a month is quite a bit, but because he did not understand what I just talked about, he failed to see the second order consequences of replacing his employee with AI. Yeah, you save money up front, but what happens next? So, obviously, what happened? Well, because the model was shitty and contextual and it was some dumb chatbot like most of them are, he saw a dip in conversions. It wasn't a massive dip. Okay, his CVR uh conversion rate ended up dropping maybe 5% from 25ish to 20%. And this may not seem that big um up front, but what that means is his topline revenue, aka the new revenue that he was driving in every single month dropped by an equivalent relative ratio from 25 to 20%. So in his case, his topline revenue is 62,500 bucks a month. And if you do the math on that, 25 to 20% drops about $12,500 a month of new revenue down to 50K, which meant, yeah, he saved $5,000 a month, but in doing so, he lost $12,500 a month. You just implemented this thing and the thing, yeah, saved you a little bit of money in profit, but it has now lost you two and a half times approximately that much in revenue. Is that really a trade-off you want to make? So, I see this all the time at Leftclick, which is my consulting agency. Uh, clients will come to us after trying to automate everything. They will get really excited about the technology obviously and they'll try implementing it at every level of their business and they'll say hey we did this but then our revenue has dropped significantly and then I think to myself well yeah I mean of course it did you applied AI in very high value low volume processes where every single interaction you have with a customer has a massive impact on revenue does that make sense you can't just replace you know human judgment in processes that directly impact revenue and then expect the exact same results at least not with current levels of technology so my takeaway is avoid AI in high value high-risk and low volume bottlenecked process processes like sales and like Legion, unless you use it really intelligently in a pre-existing process to help like generate ice breakers for cold emails or something. And if you want more on that, just check out this video up top. Uh we'll make sure to link

it. The second major mistake is when companies think that AI creates new opportunities. But in reality, AI does not create new opportunities. It just makes pre-existing workflows and processes a little bit better. By definition, AI does not introduce new capabilities into a business. At least not yet. At least not at the current levels of technology that I know of. Uh there's no new functionality here. It is just trained on the outputs of people and pretty much everything AI can do today, a human being can also do as well, just a lot slower usually since we are not all made of silicon. Here's a perfect example, a personalized cold email. Before AI, if you guys wanted to write truly personalized cold outreach, you would have to spend 10, 15, 20 minutes scrolling through somebody's Facebook feed, finding a picture of their Yorkshshire Terrier from 2021, and put together this elaborate, beautiful connection request, and send it. Uh, and I know this because I did this all the time. I had to do this in excruciating detail. And as a soloreneur, as somebody that ran my entire business on my own, this ended up being a very fundamental bottleneck to my processes. So obviously, it was technically possible, but when you do it all day, it's pretty unprofitable at scale, unless you're selling something mega high ticket, which I was not at the time. In my case, I could personalize maybe 30 emails a day. Needless to say, not very scalable, not something super high throughput. The thing is with AI nowadays, you can research and personalize 30,000 contacts in the same amount of time that I used to do 30 in. And because it is text, and because in my case, I use it. Well, there isn't actually enough bandwidth to tell that it is not a person. So, the end result is it is for the most part the same process. It's just a lot faster and cheaper. And that's one of the reasons why cold email is so powerful now. But the key that a lot of people miss there is that cold email actually worked before, right? It was a pre-existing established revenue generating process in the business. This just made the margins of that process better. So many people think, I'm going to use AI to generate opportunities out of thin air. I'm going to use AI to like revolutionize a business model. But in reality, at least for services, which tends to be what I talk about, there is not some magic paliative bullet that generates cash or results from nothing. There is no free lunch here. You need an existing, you know, sales process that works. You need proven messaging. You need a validated offer. And basically, I just don't recommend trying AI to do anything that hasn't been validated by your business before. because only if it's before do I think that it would be safe enough that AI could make something that was previously possible but inefficient possible efficient and very scalable.

The third mistake that I see all the time is when people spam AI like it's a grenade launcher in Call of Duty. They think that AI applies equally to literally all problems and hopefully as you've seen the case that I'm making throughout this is that you need surgical precision to really be able to get the 8020 from these models. So, like we talked about in the chatbot example from earlier, I would not use AI in a high-risk customerf facing application. This was already a pre-established revenue generating process. And the reason why is because small inaccuracies with AI have a way of compounding down your funnel. I think of it kind of like the butterfly effect, right? Where uh if you have a really small mistake early on, that mistake will compound and multiply over the course of the entire customer life cycle. And a lot of the time, this leads to you just not getting the client, which is obviously the biggest mistake and error of all. So the process that we use with leftclick clients is we look for pre-existing workflows that are already validated revenue producers or profit savers. Has to be one of those two things. That's stuff like the cold email example, but also basically anything that people do that currently takes a lot of time and energy for comparatively little result. This could also be administrative stuff. So stuff like invoice processing, uh data entry for sure, report generation, scraping, any sort of like manual back and forth appointment setting. And then we automate those processes. And the reason why is because these processes are in a class of tasks called permissive tasks, which are things that you just need to do in a business. They're kind of like a checkbox where, you know, all you need to do is get that checkbox from 0 to one. All you need to do is check it. But there's no like quality scale. There's no, you know, if I put more effort or time or energy into this thing, it will yield better results than if I put less. It is for most part just a 0ero to one. Which means if the quality is poorer, does not impact the quality than result. You could take that hit and then you take the time, money, and energy that you saved and then you put human beings with all this extra resource onto a task that does scale with quality. And the things that scale with quality tend to be high bandwidth stuff like sales where you interact with customers directly. So just a few months ago, I had a manufacturing client who wanted to automate the vast majority of a sales process. Uh they were doing something like $5 million a year and their contracts were also pretty high ticket when they built a relationship with a retailer who they would distribute to. It tended to pay quite a bit. But instead of just jumping at it and automating the entire sales process, we thought about it for a little bit. And we ended up saying, "No, what are the processes that are currently taking an inordinate amount of time in the sale, but they don't scale with quality? Why don't we automate those first? Then we'll leave everything that's human relationship building based alone. All the facetime, all the calls, even large portions of the email flow. " So we ended up doing is we automated specifically all the generation of their sales assets and then also we ended up automating some of their inventory management which opened up a bunch of time in both the sales and fulfillment teams. Then we just applied all that additional time to sales which at the time was their bottleneck which ended up making them a lot more money. And so this isn't really like automation. AI. What this is this is like process. When you get really good at process you realize that automation and AI aren't really doing anything new that human ingenuity hasn't been able to do over the course of the last century or so. So that's the framework. You automate the permissive stuff so humans can focus on highv value high context work, not the other way around. All right, hopefully that helped. Obligatory. If you guys would like help

implementing any of this stuff, identifying where AI makes sense in your business or calculating the real ROI of an implementation or just building systems that work, that's exactly what we do at Leftclick. We've worked with companies across every industry from small businesses to multi-billion dollar portfolio companies. Book a call with my team. We'll walk you through your specific situation, show you where you guys can use AI to help as opposed to just destroy your revenue. Aside from that, thank you for your time. Looking forward to the next video. And special shout out to anybody that subscribed to my channel so far. Really, really appreciate every moment that you guys watch my stuff.