Hi everyone. My name is Jana and I’m a Google logistician. I’m in the business of delivering targeted businesses with the right metrics and helping to scale online sales using mostly AI-enabled tools in Google Ads.
What is AI Max and what is its novelty
Until 2025, the main AI tool in Google Ads was Performance Max. Today we’re going to talk about another new feature in the advertising cabinet, which Google has already announced and which will be available for almost all advertisers in the third quarter of the year 25. This is AI Max for search advertising campaigns. Let’s run through the difference, what are the innovations of this campaign, where you can enable it in your ad cabinet, what are its advantages or disadvantages.
How many calls and sales will I get by ordering contextual advertising from you?
What is AI Max? From the campaign name, it’s already clear that these are search campaigns with AI-ka functionality. But what makes them different from the other set of ad campaigns we have? Because a lot of people are now writing about their disadvantages or pros. Some people admire them, saying that they are an innovation. Some people don’t like it at all because it reduces the amount of control over your campaigns. I’d like to run through these points a little bit, where Google got this AI Max thing from, why they’re betting on it, and why it’s going to be trending for the next few years. Who watched my old video about Google advertising trends, I was saying that they’re going to revamp search, DPOs to something very super smart and they’re going to make a tool like this. And lo and behold, it’s been two or three years, they made this ad tool. You can think of me as almost Vanga, but it’s all based on the general trends of dynamics that we see in the advertising office.
What is AI Max? It’s search with AI-key functionality. Now I’m going to tell you what it is, the logic of using these campaigns, what their advantage is over what we have now in classic search campaigns. To bring you up to speed, the logic of how this AI stuff works. I would compare this type of ad campaign so far to this type of campaign….. Let’s put it this way.
What does this AI-ka do? AI-ka selects keywords correctly enough. It selects a landing page for these keywords and can generate a text ad. When I heard about these three AI-ki features, it reminded me a lot of a type of already existing advertising campaign in the advertising cabinet, which is called DSA (Dynamic Search Ads).
What is this beast? I’m going to talk a little bit about it now so that it’s clear what AI Max is. Look, the DPO-shka, what did it do? When we created it, it could do what? It could pick up keywords from your site, if there is a way for Google to index your site. Whether it’s open for Google to index, for Google bots to index, or whether it has this whole SEO story for Google, it could pick up those semantics from your site and not use match types. Because usually when you read something about AI Max, you’re going to hear a lot of words about them not using match types. The DPO-shell also used a broad match type. Is that a bad thing or a good thing? Actually, this kind of campaigns, DPOs, are some of the most powerful for e-commerce, or, for example, if you have services there, you’re selling and the site is available for indexing. In fact, yeah. The broad type of compliance in the last two quarters of ’24 and ’25 is a major trend.
People who are still trying to maintain some manual control, to hard-switch these toggle switches, are still losing more than they’re gaining. The advantage of this broad type of matching, even in the old DPO-shka format, was that you don’t have to worry about whether or not you’ve got the right keywords that will lead to conversions, to your desired results. Plus it took into account a lot more low-frequency queries or queries that are entered once and not repeated. We’re all used to “Siri”, “Okay Google”, searching “Okay Google” for some products, there’s a lot of these queries now. In fact, your search advertising campaigns can now generate more than half of all queries from this block of low-frequency queries. This is already a separate topic of video, whether you can somehow optimize them or not. Today is not about that.
So, this DPO-shka picked up keywords from your site. They could be minus, that is, something negative selection, but the selection of these broad words was on the fact, on the essence of the keyword, and not on what would search for a user who is interested in buying a certain product. Maybe a little bit of an overreaction to something, but there’s a difference there. Further on, perhaps you will understand.
I mean, if we were selling, for example, elephant figurines, we could have queries like “Pink/green elephants”, “Elephant photo”, something like that. Of course, if you don’t filter those queries, they kept showing up and unless you have an automated strategy there for conversion or for value, you could reduce that amount of untargeted traffic a little bit.
Advantages and limitations of DPO
One more thing. There are nuances to this DPO. It’s a classy tool. It’s still gonna be relevant. It’s not going anywhere. But there are nuances to it. You have to worry very well about the purity of the traffic. Another advantage of this DPO was that the very header we set was also dynamically generated. We didn’t have to set headers. We only set two descriptions. There are also additional attributes there: additional references, footnotes, structured descriptions, and so on. That’s what’s included in the “Object” block, which used to be called “Extension”. Where did it get that header from? It was also indexing your page title and pulling it up there. Plus the third advantage of this DPO was that it actually dynamically matched the landing page. It was very likely that if a person was looking for a pink elephant, the landing page was either a page with pink elephants, depending on what you had indexed, or a particular pink elephant that led to a product card. In most cases, all of those DPOs that we ran, the vast majority of them were targeting product cards. That gave an incredible boost to the conversions of those campaigns, as opposed to classic campaigns where you pick keywords, you prescribe this ad, you assign a landing page. Yeah, because usually if we want to sell those elephants or iPhones, we would lead to a catalog with iPhones or with those elephants, and here it’s a direct product card. It was reducing the amount of time a person spent on our site. And of course, there were more conversions that way.
AI Max: improved version
You see, these same things are also present in this AI Max for Search. They are also present, but in a more pumped up version. What does the pumped up version mean? First of all, it can already generate your text ads on its own, but not in terms of what it will snag just the site. Although here also, I think they will practice this in the beginning, just pulling ready-made texts from your site, so carefully. I’ve had a situation where Performance with the same AI-ca functionality pulled texts that don’t quite fit our service. For example, a company writes AI agents for certain business tasks and they have a section for a person to get a job with them. She just pulled up that text, like “you can come work for us,” even though it’s not relevant. That is, there is such a possibility that she can also pull texts from your site. Therefore, you need to be very careful for the time being to minus the pages that you do not want them to be indexed, Google took them into account. Next I will show you where it can be done.
But the texts will really be generated, taken from the site, and they will be of a more commercial nature. Their quality will be more in line with the query that the user is really looking for. What is the advantage of these campaigns? Yes, because perhaps it is a disadvantage – they take away our ability to manually influence the price per click. Because before, in order to reduce the price per click in search to certain lower bars, we had to work on the quality of the ad, the quality indicator, the relevance of the ad, the CTR, and so on. This functionality is being taken away from us, the AI will generate text ads by itself, selecting them in such a way that they maximize their relevance to the final customer of Gula, or rather, the client. This is, respectively, the one who clicks on the ads. It will not be done for us. We will be indirectly profiting from it.
But there is a very big advantage in this, because you do not have to think about how to improve this ad, because you can not SEO-texts to change constantly on your site. Especially since Google won’t have time to index it and the OSRC won’t be able to pull the right texts. But DPO-shka is cheaper than classic search in terms of ad relevance and price per click. AI-ka will most likely be an improved version of this whole story.
Example of generating a relevant ad
What do we have in AI-ka? The same functionality that was in the DPO-shka, but an improved format for generating ads, i.e. there will be better ads. For example, in this screenshot, you can see that a person is searching for, for example, “best serum for dry sensitive skin”.
This is a person searching for the best serum for dry sensitive skin. And here you can see what kind of ad the automation can generate and what page it can lead to.
Improving semantics through AI
I’m not going to say it’s going to be straight up super perfect. They’re still going to target what you have according to what you’ve set SEO in those headers, descriptors, meta tags of your page. It’s still going to need information. But it’s a signal to you that you can improve on your own site. Plus, the more you give useful information on your site, the more likely it is that the textual content that the AI-scheme will compose will be better. Accordingly, the keywords will become much better.
We can see here that the low-frequency entry is pulling in some pretty good text, right? We’ve got “Face Serum for Dry, Sensitive Skin.” We’ve got pretty much all the keywords we need in a text ad. It’s a very cool story, actually. The DPO-sheet was pulling up your SEO text, but maybe it needed to be improved somewhere. That is, it simply took, extracted what you have in the site. But here it will be directly selected what is in the site and combined with what the user is looking for. Accordingly, the rating of the ad will be higher and, accordingly, the price per click will be lower. I expect so far, but Google is such a corporation, which has already been noticed for the fact that they are there to inflate the price per click, so it is already a separate story for this story. But at any rate we’re going to look at it this way for now.
Another advantage will be that it will use keywords of a broad match type. But unlike DPO-shka, which matched similar queries on the basis of fact, on the essence of the keyword. That is, in fact, if we sold elephants, DPO-shka, if it is not filtered, could get into “Photo of elephants”, “Big elephant”, “Elephant toy”, and we will pass, for example, statuettes. Or, for example, even “Going to the zoo on an elephant”, there could be such requests. Then here we are promised that AI-check already approaches more intelligently in terms of selecting this semantics, in terms of the sales funnel. Remember that Googlers all perceive the block of search campaigns and performance together with the merchandise at the bottom of the sales funnel. There a little bit higher in the hierarchy is Demand Gen and even higher is YouTube.
How many calls and sales will I get by ordering contextual advertising from you?
So there is a lot of hope. We’re going to be testing right now. If you’re interested, by the way, tests, how cool it’s going to be or not, drop me a line in the comments, I’ll make a separate video, because we’ve already launched some of this functionality. We’ll see how it works because it’s really interesting. “There are no analogs”, as they say, so you have to use it, test it. Therefore, the quality of semantics selection is likely to be better.
System learning AI
Since this is an AI-ka, it will need time to learn how to pick the right semantics. So, you and I are now, by including these functionalities in our accounts, helping this AI to learn. I’m all for giving AIs the ability to learn from our data. They don’t have the ability to test it in some kind of simulation, right? It’s going to be done on our advertising spend. But it’s probably now going to be massively trained on that data and probably by the end of, I think, the 25th year, it’s going to gain momentum in terms of getting the semantic core right, which is the keywords that you’re going to be shown on. Plus pumping up the skill of generating those ads. Why do I think that? Because if you ask the GPT chat room for a list of keywords for your niche, it will pick up, it seems to be correct logically, but purely system-technical for Google Ads not quite suitable keywords. Therefore, most likely it will be something of average quality, but if you optimize it all correctly, monitor the effectiveness, adjust bids, budgets, then, most likely, you will be able to teach your AI Search to generate the right traffic faster than anyone else in your office.
The only thing that’s nuanced is that you need to keep an eye on it, because AI-ika is out of beta for now, it will be available for all accounts in the third quarter of ’25, so they promise. At least there’s definitely prufs here somewhere. Yes, “it will be available for all advertisers at the beginning of the third quarter of ’25”.
Where and how to enable AI Max
Where to enable this AI thing? The most interesting thing. See, you go into your advertising cabinet, go to “Ad Campaigns,” here’s “Search.” You can see the search icon here. I select, point to my campaign icon, click this gear. And right there in the campaign settings, there’s a very huge block called, I mean, it’s very hard not to notice it, “AI Max Search Performance Campaigns.”
It’s this AI Max Search. It offers here to enable this whole block. There are additionally two more such enablement features here.
What to customize and what to pay attention to
“Text modification”, that is, use text from the website, landing pages, ads and provided objects. You need to allow you to see how much your website gives correct AI-ca information. You might find resources on improving simply even for the same Pmax.
Final URL: direct traffic to the most relevant URL. I don’t have any of the URLs excluded here yet. We’re doing a special test to see how well the system can do this. But, ideally, just like for AI-ka, exclude some pages that you don’t want it to take – texts or targeting. This is usually “Return Exchange”, “Blogs” yours, a bunch of different information that is not particularly commercial, so it’s worth excluding. We’re not excluding it for the test yet.
And, of course, the exclusion of brands. Should brands be excluded? I would exclude brands right away for this type of advertising campaign. Because if you have a powerful, such a strong online brand that generates a lot of sales on its own, you should still exclude it here. Why? Because there will be a very big temptation for AI-ers to get all the super-class sales just from a branded campaign. Is it worth interfering with that? No, better to exclude it, forget it.
By the way, you can include brands here. Ads will be shown only on the search query that match your keywords, mention the selected brands. This is such a nice addition that can also be utilized. Why not? So pay attention to that.
I mean, this is the kind of customization block that you need to do. More such a super exclusive in the settings so far I do not see what you need to do. This is already in test mode. Betting strategies do not offer us to change anything here yet. For now, everything remains in this format, as it is.
Your task is to keep an eye on the search query block. We make sure to check the block of search queries. Search queries need to be cleaned up, cleaned up, because you can trust AI-ki, but it’s better to see how high-quality or low-quality words are. In this format, I see the queries, they seem to be good, but you have to sit down to figure out if everything is going to be like this or not, because still, you know, trust, but double-check. So this is the format we have.
Summary of AI Search advantages
Once again, let me remind you what the advantage of this AI Search is. Collection of low-frequency targeted semantics with buyer intent, as opposed to just targeting on similar semantics. Because, for example, if you were selling just dresses, DPO-shka could pick up black, red, of course, taking into account the context of your site, but the disadvantage of DPO-shka was that it was not suitable for a very volatile product range, which was constantly changing. Plus it was not suitable for those sites that were not with enough history that they were indexed enough.
Here, however, there are already such circumvention mechanics in place, so it’s not surprising that I would call it an AI-tooled DPO, but that’s what they’ve called it so far. So we’re launching it, we’re looking at it. They are still emphasizing that there will be very good quality ads for these specific queries. I call them long-tail queries – that’s when there’s a query with more than three or four keywords, regardless of the number of characters, four, five, six words. These are things that don’t get pulled up, usually, in the Search Queries block, but they’re also conversion queries for us.
Conclusion
So that’s the story here, we’ll keep an eye out. We’re just running a test. We will now see how well or poorly it works. Maybe I will share these results if you are interested, because we still need to run this tool with you so that AI-ka learns, because delaying the task only delays the realization of this thing. They can also artificially force us to do things. As you remember, this was the case with Smart Shopping, when they forcibly removed it and created just PMax. So it can also happen with search campaigns, because now even in search campaigns, keywords that have a fairly small volume of search queries, i.e. low-frequency keywords, are just automatically suspended by the system.
So we have no choice, we have to use it. Монополист также диктует свои условия, но, как говорится, когда везде AI, AI, AI, AI, AI, везде оно все присутствует, надо идти в ногу. Plus view it as a powerful boost to your account, because if you teach that AI agent inside to generate what you need, you’ll only be praising it. In fact, if you teach it to generate the right results, it will only be for the better.
Well, that’s it for now. I hope this video was helpful. Such a review of a novelty product that’s on Google. And see you in new issues.
