Is AI for Businesses All Talk and No Walk? High Failure Rates!

Foreign.

Hey, I want to jump on here today and talk about failure rate with AI. And the reason I want to talk about this is because I've been heavily in the field as a digital marketing agency and now an AI automation and consulting agency for the last almost three years. Pretty much at first a test subject, testing it on my own product and my business before ever launching it with any other clients. And the reason I'm jumping on here today, because there is a lot of information going around and I think, I think media tends to latch onto the negative and then find evidence to prove it real. And one of the things that I see popping on a lot right now, and it's, it's this one, is businesses are, are experiencing a failure rate between 80 and 95% when they use AI. And by this I mean, it's like, okay, they bring it in, they think it's going to work and then it fails. And they say, yep, this AI is nothing but smoke and mirrors. It doesn't work and it's a failure. Well, I have to tell you that, yeah, it's new technology and it can be one of those things where it's a bright shiny object and it's, it could be, and I'll get into more detail, it could be one of those things where it says, hey, it is going to solve all your problems and then it doesn't. And it can be frustrating and it can be expensive. But I want to talk about first of all, the items that, when I did some research, why we have such a high failure rate with, with AI. And the first one is misalignment with business needs. And let me give an example. Let's say for instance, a company comes to me and says, hey, we would like to do X, Y and Z. And I say, okay, great, let's use AI to do that. And what they're saying is they want sales and I'm talking marketing. Those are two different things. One is for the future, one is for the now. Sales are for the now. But let's say, for instance, I use a program that we built, we call it the Content Accelerator and it creates all the content weekly for every customer, you know, including a blog, video, all the SEO and aeo, or if you might have heard, it's called geo. That's so the artificial engines can find it. It does all the social media, it writes in an email. And all of this is done using artificial intelligence and automation. But they want, they want ads run. And so there's a misalignment there. And so we have to make sure that this fits. Now, let me give you, let me get, give you a flip side of that. A pro of that. Recently I went into a business and we looked at their business, said, hey, what is something that they do? And I want you to, if you're watching this, whether it's live or the first time, think about this. What AI right now is great for about, I'm going to say 90% of getting you there. 90%. If you go into it thinking, hey, I'm going to be able to just put in this, this request, like, you know, I'm going to put in a vibe coding for a website and it's going to come out and I'm going to be able to launch that dude today. You're going to be disappointed. Now, it's going to get you 80 to 90% of the way there. But you, me, we're gonna have to tweak that. It's gonna do most of the heavy lifting. It's almost like if I wanted to dig a pool in my backyard, a backhoe would do most of the heavy lifting, but we'd have to get in there with shovels and do the fine tuning before we started building the pool. Well, think of that like AI. Now, I'm just coming across as someone who is in the trenches doing this every day for local businesses, including my own. And we've had some really great results. And have we had some weird results? Absolutely. But we caught them before they went live for a client. So the first thing is misalignment with business needs. And it says projects are often launched to solve a vaguely defined problem or use the latest technology rather than addressing a concrete business need for users. Another example of this, well, somebody says, well, let's build an AI agent for you to handle all your potential leads. The flip side of that would be, well, we'll build an automated workflow that has a drip email system or whatever it may be against the AI. And they're saying, well, I can replace it with an AI and it's not really ready to go yet. Right. And so it feels like, oh, that was a failure, that didn't work. Let's hire, let's hire Sam and Sandy back in taking those phone calls. So one of it is misalignment. The next one is data issues. This is poor data quality. Maybe, like it's insufficient and irrelevant data or, excuse me, insufficient relevant data and lack of proper data management are often significant barriers to training effective AI models. So as an example, and I share this with my clients, when we're using AI and they say, well, I remember early on, they say, well, yeah, it sounds good, but it just doesn't sound like me. So what we would do is we would start getting content or creating content from that ideal customer. This video, it's written, whatever it is, audio. And we would train our models to say, we want this to sound just like Jeff and using the same phraseology, the same terms, the same tone. And once we put that in there and we trained it to do that, and then we ask it to perform a task like writing a blog. And then I would read that blog and I'm like, oh, my gosh, that is so good. Well, you can do the same thing and have those same oh, my gosh moments. This is so good. But not if you don't train it. It's almost like having a new employee and you give them keys to the castle and you walk away, and they don't really know what you want to do. They're not acting like you. Just recently today, somebody said, hey, if they would only act more like me. Well, this is the way to do it. You have to train them with that information. The cool part about this, in my opinion, is the training can happen so quickly, literally, if you have, like I did1,1 AI program that it's an AI avatar, it acts and sounds just like me. It will take take calls, it will talk to you, it will counsel you. And it has almost a million data points that we uploaded. It's over 15 years of marketing, all the. All the videos, all the articles, everything. I gave it, all of that. So it has this tremendous resource of information, and it's also synced as we put new content out. So it really is my brain in another in the Internet that can help you with. So when you properly put that information in there, it really does it. So lack of data, relevant data can be a. Can be a problem. The other one is, and I think this is a big one, unclear objectives and expectations. I think it's really good because what can happen is somebody says, again, I'll go back to an example where I got excited because I know what AI can do. And I went into a company and they said, we'd really like to think about having an AI voice agent. And I looked at their entire company and the way they had it set up, there were some redundancies that we could now use AI to help them with and again free up more of their time, increase efficiency. And I showed that to them, and they weren't quite ready yet. Now they went ahead with the AI voice agent and that's going to be fine. But knowing down the road that this is kind of almost like having an ace up their sleeve, they know if they get to the point saying, hey, for us to scale, we have to do this next step. And we've already seen behind the curtain, right, and we know that it works, that's meeting expectations. So sometimes having an unclear objective or expectations, and it's a gap between, hey, what they want and what they really need, I think that's important for us. That's where we started shifting and saying, hey, we're more of a consulting, like we'll come in and then we may spend a day, we may, it depends on the size of your company. May spend a day, may spend a half a day, may spend a week reviewing everything you're doing and saying how, how and can artificial intelligence, the way we have control of it now, can we build a system to make, to make it better for you? These are things that if you don't do that, it can certainly make a difference. And I'll be the first one to raise my hand and say, hey, I've purchased AI stuff that I got excited about and once I start unboxing, I'm going to, why did I even get this? This, this doesn't, this doesn't really help me. As I say, it doesn't help my, my boat go faster. It doesn't get me any closer to my goal. But I was so excited about it because it made so much sense. And you know, I can give you dozens of those over the last three years because the technology and it was always evolving. Now the next one is insufficient resources. And to me, this is, projects may lack the necessary infrastructure, the people to do the work and the financial resource to manage the data and deploy AI models efficiently. Now for Swingpoint Media, what we do, our goal is to step in and say, let us build this, let us put it in your business, let us train your staff, which is designed probably 99% of time, it's designed to work alongside your staff, not to replace your staff. Just like when we started switching over from full time writers to using AI to write our content. What we saw was we retrained many of our people to say your job now is X. Whether it's now you're, you're now posting, you're now editing, you're now over, you know, you're looking at the, at the final piece of content, you're doing the images. In other words, they just got a Different role. And in a situation, let's say you have an AI voice agent for your business. It can be set up to say, you know what, I want to answer 100% of my phone calls. And it leaves me free now to take care of the walk in traffic that may be your business. There's a lot of people walking in. Last thing you want to do is keep interrupting somebody or putting on hold while you can take care of your customers or patients right in front of you. The other might be, hey, you know what? If the phone rings more than two times on the third ring, the AI voice agent will answer that call and it will now act as if it were you and perform the duties that we've designed for. And it could be, hey, it's making appointments, it's answering phone calls, it's actually transferring phone calls. It's now calling people the day before to confirm that they know their appointments are ready to come. And if not, if they need to change it, then they'll make those adjustments and ping the owner. So, you know, there's a lot of little things that can be tweaked, but if, if, if those objectives are, the resources are insufficient, you probably can't do that. It may be one of those things where, you know, there's a saying is you've got a, you've got a champagne taste, but a beer budget. And what that means is, hey, we want all of this. But it does take time and money because most of this right now, it's not like software. It's not coming out of a box. It's. It's this artificial intelligence. Imagine it's a bunch of puzzle pieces that if you know what you're doing, you can put them all together and it's customized just for you. For us, as an example, we were using a product to help us distribute and post content. And as we started looking at this, we said, you know, A, it's quite expensive, and B, because we're using it for many of our clients, we said, what if we built our own? And it's taken a couple of months to do this, but in the long run, now it's ours. And we did that. We put together all these pieces of the puzzle, and now it's our product is something we can do and we're in full control of it. So that's something you want to take a look at. But if you're not ready for that because it can be expensive in time and money, then you say, you know, maybe I want to do a little bit Differently I'm going to take, say I'm going to roll out a four phase plan and it's phase one, phase two, phase three, phase four and it's over, you know, 812 months, whatever the number is. So insufficient resources can be an issue. The next one is.

What we call focus on sales and marketing pilots. And a significant portion of AI investment is funneled into sales and marketing pilots which are easy to pitch but often fail to deliver measurable enterprise wide results. I'll give you an example. Just recently one of our clients, I was on the phone with him and we were doing our weekly call and, and they said well we want to do something with our database, we want to gather more information on the current people that are database. And so they were saying well we'll can we, can we hire these people out? And I said well gosh, I said we can probably build an AI agent that will give them the names, we'll give them the fields that they want in their end use in their CRM and then we'll have it go out and search public records to fill in the gaps for them. And they said wow, you can do that? I said well I think we can, let's try it. Had I done that up to this point? Absolutely not. But we went back and in a very short period of time we trained up an AI agent, gave it access and then gave it a small working test portion and said here's a hundred names and this is what we want. Here are the fields we want. We want you to go out and research and find out as much as that as you can and then put it right back into the spreadsheet and then we'll put it back into the CRM and it will update that well in a few hours. It did the job of what would have taken someone individually because we've had people, we have somebody on our staff right now that scrapes the Internet, they do that for clients and it's a long, tedious and quite frankly it's an expensive process. This helped overcome that. Now the other piece of that would be cool is saying, okay, now we've turned this over and now maybe someone, a human will now overlook all that information and make sure it works. But this is one of those things where, you know, some of these things come out as a pilot and nobody really tells anyone else how to use it. And I think that's one of the biggest mistakes is we get excited about oh, it'll do all our emails, oh it'll do all our videos, we can make videos with it, you know, and crazy stuff on the Internet right now will make you think, oh, I can do that. I can make a full blown, you know, whatever. And all I have to do is give it a few words, a few prompts, we call it. And then it turns out to be a mess. And, you know, that's just. I think that's from giving. It's almost like giving a child a set of keys to a car and wondering why they crash it before they got out of the garage. It has nothing to do with the car, it has nothing to the child. It's the training piece in there. And that's why with SwingPro Media, we like to make sure if we turn any of this over to our clients, most of the, most of we manage in house. But if we turn it over to our clients fully, they are fully trained and they know what they're doing, just like our staff would be fully trained on how to do this. So they know what they're doing. Now the. One of the last ones is lack of maintenance and evolution. I can't. This one, this one used to pull my hair out because quite honestly, for me, it was moving so fast. AI, it's still moving so fast. Just about the time. I remember when we first built our AI automation, we call it the content accelerator. And it would do all the work that we would take a week and 17 writers to do and video editors to do. It would take, you know, just, let's say, let's call it a day instead of a week, it'd take a day. And at first it was like, wow, this is, this is really good. And then what would happen is a week later, so some new tool would come out that would enhance it, or new version. Instead of a ChatGPT 5, it was ChatGPT 3 or 2.5. And it kept evolving, so we had to keep updating that. Well, if you're not willing to do that, you're going to see some deficiencies and if you're not willing to maintain that.

We're building a voice agent right now for my company. And it's really designed for two reasons. One is it'll take my calls and it'll answer all the questions, do appointments and things like that. But the other piece of that is I wanted to be a demonstration. I want people to come to the website@swimmingpumpmedia.com when we have that on there and say, well, let me try it out, let me hear what it's like. Does it sound like a robot anymore? No, they're very interactive. Is it perfect? No. But this is the worst it's going to be today. It doesn't mean tomorrow or a week from now. If that improves, we will update it. A lot of people think it's like software and it's not. We're just going to set it and forget it and move on. And in wrapping up, that's what I want to really stress. I said it early on. I said, hey, if we're doing this using AI, it's going to get you about 90% of the way there. So that means you or someone on your staff will have to do the fine tuning, that extra 10%. And doesn't mean you'll ignore that other 90. You'll review it and say, boy, that's spot on, or I need to make a little tweak here and there, but you'll do the other 10%. And to me, I think this is probably the largest reason, in my opinion, of the failure rate, because we have a mindset of it's going to do the entire job. It's artificial intelligence. It's, you know, it's got 160 IQ. It's. It's got 148 IQ. Whatever it is, it should have all the answers and should get it done the first time. That's not it. That's not it. You know, we do prompt engineering. We don't. We do some conversation, but most of us prompt engineering and we test it and make sure that when we put it in, it comes out consistently. Think of it like a recipe. If I follow a recipe to make these brownies, if I do it the same way every time, the outcome should be the same every time. Unless there's some kind of freak thing, you know, that I got a bad batch of flour or whatever, something was rancid. Everything should be.

Good in, good out, not garbage in, garbage out. So the big thing is it's not going to take you all the way. It's going to do most of the heavy lifting. Again, think of it like a backhoe. It dug most of your pool, but it did not do the fine tuning that's with you or the crew out there with a shovel or know, even a trial and fixing it up and making it right. So before you get sucked into this thing that says, oh, you know, AI is. AI is overblown or it's. It's not what it should do. It's really not doing the job. Why should I get involved now? Because such a high failure rate, because it's getting better every Day. And some of these points that I've made that if you look out going forward, you're saying, wait a minute, before we do it, am I really needing all of this? Or do I just. I just need somebody to write my emails. Great, we'll train it. Or I want somebody, I want a tool that, that can connect to my emails and go through it and figure out which ones I need to respond to. And it will write a response in my voice and then, and then let me look at them in draft and hit send. Maybe that's something you could do. And maybe, you know, if you're. Someone gets, you know, like myself, we get between 1 and 300 emails a day. Having that help will make a huge difference. Because some of this stuff is like, yeah, this is, this is not a hot potato. I don't need to deal with every email that comes in, but it needs to be handled. So maybe there's a, you know, level one, two, and three type of emails. And this artificial intelligence has been trained to understand how to handle that, and it can save me a ton of time. So I just want to pop on today because it was a little frustrating. I keep hearing this. I just, just got a reading before I got started on this that I think it was Harvard Business Review was giving another diatribe on how negative the results were for people that adopting AI. In closing, I also want you to remember first, we are barely, barely, barely out of the early adoption phase. And the people who do early adoption, they're geeks like me, that we're trying things. We want to be on the cutting edge. We're testing this stuff out. And when somebody sends something over to me about, you know, a tool that I might have used, and it didn't work out really well because it was in, you know, it was in, in version one or version two, and I gave a negative response. Now maybe the, the media or, or in this case, Harvard Business Review comes back and says, oh, that was a failure. That was a failure. Value didn't work. Now we're geeks. We're testing stuff out. You know, imagine how many times, you know, go back and look it up. How many times did Elon Musk blow up a rocket before he got it to go up? And now look at his numbers. It's unbelievable. You know, sometimes it feels like he's sending something up once a day, whether it's from the east coast or the west coast, he sent. Or even from Texas. He's sending rockets up and returning them safely. But he blew some rockets up on the way. And I want you to realize this could be the situation. So, anyway, I just want to point that out. I want to share this today, and if you have questions about it, please put it in the comments. And I'd love to hear from you. And if you want to talk more, my name is Jeff Harrison, and, you know, I'm with Swing Point Media. I'm the. I'm the owner and CEO of the company. And this was just something that was kind of like, I don't want people to get the wrong idea about this. It really is a lot of good. All right, have a great afternoon.