AI Won't Fix Bad Ads: Stop Blindly Trusting the Algorithm

We’re back on For The Daring Growth Seekers podcast, and this week I’ve spent a bit of time auditing a brand new client’s Google Ads account. And as I was going through the account, there’s something that really did stop and make me pause. What they’d done is they’d gone into Google Ads and they’d blindly switched on a tool called AI Max. They did it a few weeks ago, and the algorithm ever since is now running wild. It’s pulling in search terms for regions that they didn’t even service. Money was literally going out the door for clicks that had zero chance of converting.

When I sat down and asked them why did you turn this on, they literally said, well, everyone’s saying AI can do it better, and Google’s literally shoving it down our throats as well about how good their tools are, so we just thought we’d give it a bit of a go. And that right there, that is the problem. We’re trusting these AI tools blindly.

And I actually recently came across an article as well on an industry publication called Search Engine Land. And I think it also really hit the nail on the head. Now, it came out as a very spammy title to this article. It was “Three PPC myths that you can’t afford to carry into 2026”. But the article itself, it actually broke down exactly why I think that blindly trusting AI in your paid ads is a lazy strategy. And I’m not saying not using AI in your paid ads strategy. That’s the complete opposite to what I’m saying.

So instead, I’m going to unpack it a bit for you. I’m going to combine it with what I am actually seeing in real clients’ accounts at the moment, and can actually give you a pretty clear plan on how you can actually use AI properly in your paid media. If we haven’t met before, my name is Jordan. Welcome to For The Daring Growth Seekers podcast.

It has been a hot minute between podcasts, and I appreciate that. It’s been a bit of a break. And the reason being is that we have had a bit of a change here at Done By Nine. We have moved from our shipping container. So we’ve been recording this podcast, and we’re running our entire business, out of a renovated shipping container, if you’re unfamiliar with our original story. But after five years, we have upgraded. We’re now on a new premise, literally next door to where the shipping container is, but we’ve moved to the front of the street. So if you’re familiar with New Plymouth, we’re on Devon Street West.

So right on the street front now, we’ve got a bigger space. We’ve got more space for the team. We’ve got a brand new workshop space that we’re working on. And coming very soon, we’re also developing our own podcast studio, because the podcast has been a big part of what we do day to day, and so we want to have a designated space for that. So that is coming very soon. But I appreciate it has been a little bit of a break in between episodes so far.

Let’s get into today’s episode. And as I said in the intro there, essentially what we want to do is we want to stop blindly trusting the algorithm, stop blindly trusting AI to be able to do some of the work that we think it’s going to do a great job for. But it just doesn’t have a lot of context. So just going back to what I was saying in the intro, here’s a little bit about what we’re seeing across the board right now.

The main issue is that businesses are switching on AI features that we’re seeing in Google Ads and Meta as well, without understanding what those features are actually doing. So in Google Ads it could be AI Max, it could be Performance Max; in Meta you could be using Advantage+. They’ve obviously got a whole heap of different suite of tools in there. So essentially we’re seeing that people are reading, I don’t know, blog posts that are saying AI targeting is outperforming manual targeting. They’re consolidating everything, they’re letting the algorithm loose. And then they wonder why their results are, well, pretty average at best, if I’m being honest.

And that article that I was referring to in the intro, Search Engine Land, it actually broke down the three big myths that spread like absolute wildfire back last year. But I’m still seeing a lot of those myths that are starting to play out in our client accounts right now. So I’m going to get into just a couple of those myths today that I think are relevant in today’s day and age.

But before I do that, I do just want you to understand one thing, and that is that you know that I actually think that AI tools are the future, especially for marketing, especially for not only generative marketing, but being able to help us to analyse and to get better results for our clients’ campaigns, and also the campaigns that I run as well. So AIs and algorithms, they are incredibly powerful. They’re taking a lot of that guesswork out of the marketing for us, okay? So that’s the first thing I want to get across.

But, and this is the critical piece I want you to understand as well, that these algorithms and these AI tools are only going to work when you feed them the right data. Bad signals, weak foundations, they just get scaled faster, okay? So that’s it, that’s the game that we are currently playing right now.

So let’s start with the first myth, okay? And that first myth is that manual targeting is outdated and AI is doing things much better. And look, I actually think that there’s a lot of truth in that statement, that manual targeting is getting pretty outdated, but it’s under very specific conditions. So AI performance does depend entirely on your input. And this isn’t just with running ad campaigns, but also with your generative AI and everything like that. If you don’t have enough volume, you don’t have enough data, the AI can’t learn properly. So if you don’t have things like clean conversion signals, the AI is optimising for the wrong thing.

I’ve just got out of a new client’s Google Ads campaign, and they’re facing this exact same problem. Literally, the conversion data is pretty much nil. They’re still using an old conversion setup back with Google Analytics, Universal Analytics, which I think was stopped years ago when we all got changed to GA4. And if you’re a marketer, you know what a shambles that was. But they are still connected to an old conversion signal that is just not doing any work. So AI Max, if they turn that on in that account, it would be optimising for the complete wrong thing.

So if you’re a big e-commerce brand, and maybe you’re feeding purchase data back into your Google Ads, you’re consistently getting, let’s say, at least 50 conversions, let’s say, every single month. So in that particular case, yeah, using an AI targeting tool would make sense. You’re feeding a lot of data back into Google Ads. They’ve got volume. They’ve got a lot of signal quality in there as well. But if you’re a service business or a smaller business, maybe a B2B company, you’re tracking leads instead of purchases, and you’re not getting hundreds of leads every single month, logic is going to break down really fast. So you do need to be really careful.

So I’ve got a client that I want to take you through here. They’re spending literally between $750 to $1,000 a day across multiple campaigns. And essentially what we found is that they weren’t hitting the budget every day. Essentially, we’d started to hit a bit of a ceiling with keywords and keyword targeting. And we couldn’t figure out how to reach more people with the keywords that we were targeting. We didn’t want to go too wild on them as well. So in that case, we tested out AI Max.

Obviously, we firstly set it up properly, made sure conversion tracking was really clean, made sure the website was solid. We set up our negative keyword strategy. And you know what? For that particular client, with that type of spend, $750 to $1,000 a day, it worked. It’s doing really well. We’re finding new customers every day that we didn’t even know that were searching for us, based on some of the keywords that AI Max is actually finding, and the search themes that AI Max is actually finding for us as well.

But then this week, as I said in the intro, I audited a different client who turned on AI Max. They didn’t have any of that groundwork. They didn’t even have the proper location exclusion, some of the real fundamentals of a Google Ads campaign, that messy conversion tracking. And despite them thinking that they had the location settings configured, AI Max was actually showing ads to people in regions that the client didn’t even serve. And it was a physical business that people actually had to go to as well.

And I think it was doing this because the algorithm was looking for anyone who might be interested in that service, regardless of whether the business could actually help them or not. It was essentially going out blind, trying to find as many clients and customers as it could and trying to help them. But this is exactly what can happen if you blindly trust that algorithm without understanding what signals you’re actually sending it.

So before you hand over any targeting decisions to AI, there’s three fundamental questions I want you to ask if you’re going through your Google Ads account. The first thing is, are your campaigns actually optimised against a business-level KPI? And I’m not just meaning, oh, I just want to get more leads. But I’m saying, is the KPI that I want to get leads that cost less than X, or get sales above so-and-so percent margin, for example?

Okay, the second question I’d be asking myself is, are you sending enough conversions or data back to that ad platform? Because you do need volume for the AI to learn from. And the third thing is, are those conversions reported quickly with minimal delay? Okay, so are they actually going back into the ad platform quick, or is there a delay? Does it take a while for the conversions to actually come through?

So if you answered no to any of those questions, I honestly think that you do need to go back and start fixing up some of your marketing fundamentals. If I’m not making it clear, and not only Google Ads, but across all your marketing, across everywhere right now, data is the key. We need to be getting data back into these models so that AI can actually help and do these things better for you.

And don’t be afraid to go a bit old school when you are faced, and when the situation calls for it. So that article I mentioned before actually gave a really good example of this. The author literally doubled the client’s margin by implementing a match type mirroring structure and pausing broad match keywords. Okay, so if you don’t know what that is, that’s okay. Essentially, we’re saying we’re just going for as tight a targeting as we can, and not letting AI Max do its thing. And that runs completely counter to what Google’s telling us to do right now. But it’s working because they’re able to have a look and understand their data, and see where their client was at at that given time.

So it all depends on, for example, where your ad account is currently at. How much you’re spending per month. If you’re only spending $300 or $400 a month, most likely some of these AI tools may not actually be the best bet for you just yet, until that technology catches up. But this is a great opportunity for you to be focusing on those fundamentals right now.

So look, I guess the example is that if you’ve got broad match keywords, and potentially, I don’t know, they’re delivering leads at $33 each, and you’re like, yes, this is awesome, I’m getting such good cheap leads at the moment, except those leads had a customer acquisition cost of, let’s say, over $2,000. And so that’s essentially saying that while you’re getting a lot of leads into the business, they’re low-quality leads that are not actually converting into paying customers, which is obviously a problem.

Meanwhile, if you’ve got some exact match keywords that may be costing you $40 a lead, or $45 or $50 a lead, but then you can actually acquire that person for $450 over an entire spend, that’s obviously a win compared to $2,000. So even though your cost per lead is slightly higher, that actual cost to actually acquire a customer is actually four times lower. So that’s just a really broad, general example.

But it’s essentially, Google’s algorithm is doing exactly what it is told to do. If you’re telling it to deliver the cheapest leads possible, it will deliver cheap leads. But those cheap leads potentially aren’t going to convert into customers, and that’s obviously a massive problem, because that’s your marketing budget just being thrown fully away. Then, obviously, if you take back control, you focus that budget on the exact match, like that example did, the keywords are actually able to convert, and the margins were able to double, which is great.

And that’s the point. The algorithm is literal. It is literally going to do what you tell it to do. If you tell it the wrong thing, you give it the wrong data, the wrong conversion data, you’re going to get the wrong results as well. Okay, so that’s the first big myth I just wanted to clarify.

The second thing is around creative volume. Okay, so in 2025, Meta rolls out Andromeda. And suddenly, everyone starts talking about how much more creative is going to mean more learning, which means better results. And look, again, I get there’s a bit of logic there. The more creative variations we can give a platform, the more options they have to actually match the messages to different people in different contexts. But in practice, I honestly think that this myth is more reliably going to actually increase just your creative production costs, and how much ad creative and copy and work you’ve got to put into there, as opposed to actually improving your results.

And this is why I say that: because creative volume is only helping when the ad platform is receiving enough high-quality conversion signals. Okay, so without any of those signals, more ads is just meaning more assets for it to rotate through, and the AI has got nothing meaningful to actually learn from. And especially if you’ve got a low budget into that account. If you’ve got thousands to spend in your ad budget, obviously, the more creative, the more is better, because obviously you’re able to allow Meta to signal through a lot of different pieces of creative. But with smaller budgets, you’ve got to make sure that you’re keeping your creative targeting tight.

And I see this all the time. Businesses pumping out 20 different ad variations, most likely all of them generated by AI as well. And then they’re getting confused as to why their performance is staying flat. Because suddenly, they’re feeding all this awesome creative, and it should be doing amazing. But it’s because they haven’t fixed the conversion tracking. They haven’t thought about things like their customer journey. They’re just adding more creative, more noise to a system, using all these different ads. And they’re adding it to a system that doesn’t actually have the data that it needs to actually make good decisions.

So, when resources are limited, and let’s be honest, when you’re a small, medium-sized business, resources are always limited, I hear you, then I honestly think that volume and creative, so your creative volume and the amount of creative you have, is not your best move. Instead, I think you need to be focusing on conversion rate optimisation instead. So, the first thing here is, again, reviewing your tracking. Track conversions, improve your performance. So, if you’re only tracking form fills, but the real value is really when someone actually books a consultation, then you should be tracking the consultation instead.

So, I’m going to go back to that client I was talking about before, who runs a quiz. So, this quiz on their website actually self-qualifies leads before they book a consultation. So, initially, they were tracking quiz completions as that conversion. So, what we did was, we switched to tracking actual consultation bookings. The algorithm started to change its behaviour. It started to find people who were actually more likely to complete the quiz and book a call with one of their team, not just who would click through from the ad. Because that’s not the person we want. And that is the power of providing these systems clean conversion signals.

And then, after that, you need to start thinking about improving your customer journey. Make it really easy for people to convert. Then, map that higher-margin product or service, so the algorithm actually knows what actually is making you money. And only then, when you’ve got clean data flowing through, that’s when you should start testing new creative variations, and trying to do it at scale as well.

Look, if I’ve lost you, I’m going to pull it right back and tell you right now. If you’re considering using AI features in your Google Ads or your Meta campaigns, here is the checklist you need. Okay, this is all you need to take away today.

Firstly, conversion tracking. Is your conversion tracking accurate? And I mean actually accurate. Not just, oh, yeah, we think it might be working. I want you to be entirely sure that it’s accurate. If you’re tracking people filling out a form on your website, but the real value is in the qualified lead or the actual customer, you’re sending the wrong signal back to Google or to Meta. If you’re tracking clicks to a website, but not what happens after that, honestly, you are doing things blind. And I saw another Google Ads account this morning that did the exact same thing. Because the AI is making decisions based on what you’re telling it is, that’s a conversion.

Okay, so make sure you’re tracking the actions that actually matter to your business as well. I think that’s the first big thing I’ll be getting you to think about. The second one is your website quality. Always be reviewing. Is your website fast enough? Is it mobile friendly? Is the customer journey clear? Because, again, if you’re turning on AI Max or Advantage+, it’s actually going to start trying to send traffic to different pages on your website. It’s going to do it automatically. So you better make sure that those pages are actually working. Because if they’re slow, if they’re confusing, if they’re broken, you’re essentially just paying for a bad experience, and you’re not going to convert those people.

The other one on the checklist is data volume. Okay, you’re getting enough conversions for AI to actually learn from, or enough data back into the platforms for it to learn from. If you’re running a campaign that gets five conversions per month, I don’t really think that AI is going to be able to do much with that. You need more volume in your account. So I think the general benchmark is at least 50 conversions per month, depending on your bid strategy. That’s just a real blanket benchmark. I can take a look at your account if you’re interested, though. But if you don’t have that amount of conversions, I honestly think just stick with your manual campaigns. Focus on fixing your conversion rate. Trying to get those conversions up. And until you do, then you can start playing around with AI Max as well.

The other big one is your signal quality. So are you tracking negative signals? So for example, if someone’s filling out a form and you mark it as a conversion, but then 80% of those conversions, those leads, turn out to be complete junk, you’re teaching the algorithm to find you more and more junk. What’s the point? So make sure you’re tracking conversions that are actually going to lead to more revenue. And if you can, make sure you’re sending negative signals back to the platform when a lead turns out to be low quality, so you can actually tell them this is not the type of lead that we’re actually looking for.

And the final one is your campaign structure. Okay, so I want you to check: is your campaign structure built for AI, or is it a mess from your 2018 campaigns that you’ve been patching together ever since then? Honestly, AI right now is working best with clean, simple campaign structures. If you open up your ad account, your Google Ads or your Meta, and it’s got 47 ad groups in one campaign, all targeting different things, you need to stop. You need to consolidate before you even look at playing around with some more AI tools in there.

And again, we can give you a bit of a hand with that as well. You can reach out to myself or the rest of the team. We are more than willing to go through and audit your Google Ads account, and give you a bit of a heads-up as to giving you some clarity as to how you can actually shift your campaign structure to work better for an AI-driven world.

So I’m hoping with today’s podcast that you see my main point, and that is that the problem isn’t the AI tools. The problem is its misuse. So platforms are doing exactly what we’re asking them to do. They’re optimising against the signals you’re providing there, within the constraints of things like your budget and your structure of your campaign. And when your fundamentals, when your marketing fundamentals are broken, AI is not going to fix it, and it can’t fix it. It’s just going to scale your problem faster.

So this year, I honestly think you continue your focus on getting those foundations right. Feed the algorithm clean signals, and then let it do what it’s actually good at, which is helping you find more customers, more clients. But the main thing I want you to take away from this podcast today, it’s pretty simple: that AI is obviously pretty powerful, but it is not magic. Okay, so if you’re not feeding it the right signals, you’re not feeding it the right data, you are just going to burn your marketing budget faster.