The AI-Powered Paid Media Playbook with Google

Catch the key takeaways

In our recent webinar, Amsive’s paid media and digital strategy experts partnered with Google to break down how marketers can intelligently adopt AI in paid media. With Google AI tools evolving rapidly and advertisers facing pressure to test, learn, and scale responsibly, the conversation centered on practical frameworks and measured approaches..

As AI-enhanced tools like Broad Match, Performance Max, and AI Max reshape Google Ads, understanding what to test, how to test, and how to align with business goals is essential. Here’s what you need to know:

1. AI adoption can be improved with structured testing

Marketers fall into three groups: those afraid to lose control, those curious but hesitant, and those already testing with a framework. Success depends on that third mindset. At Amsive, we take a strategic experimentation approach: test intentionally, set benchmarks, use holdouts, and build plans around business goals—not just media metrics.

2. AI Max enhances search, not replaces it

AI Max adds a layer of optimizations—search term matching, automatically created assets, and URL expansion—to existing search campaigns. Marketers can choose which AI Max features to enable, and testing can be as conservative as turning on keyword expansion only. The key is clarity on goals and robust data to guide AI decision-making.

3. Early results across verticals are promising

Case studies across B2B eCommerce, B2C retail, and financial services show that AI Max can drive lower costs, higher order values, and incremental conversions. But results depend on the scope of expansion (keyword, ad, landing page), site depth, and risk tolerance. Most early wins came from starting small (with keyword expansion) and scaling based on performance.

4. Broad Match use is foundational and requires testing

Broad Match is critical for AI tools like AI Max and Performance Max, but not all testing methods are equal. There are three frameworks for Broad Match rollout, each balancing control and scale:

• Consolidated ad groups for modern search structure
• Separate Broad Match ad groups with tighter controls
• Standalone Broad Match campaigns for full budget control
  Match your testing structure to your industry, volume, and goals

5. Success with AI comes from intentionality, not automation

Being AI-ready doesn’t mean turning on every switch—it means knowing what to test, where, and why. Optimize inputs, not outputs: clear goals, clean conversion actions, negative keyword lists, and meaningful segmentation. In highly regulated industries, start with what’s safe—like keyword expansion—and exclude sensitive content from automation. AI works, but only if you tell it what matters for your campaign goals.

Dive into the transcript

Introduction

Joseph Sharp (00:05):

Alright. Hello everyone. Thank you for joining us. We’re happy to have you as we dive into the most urgent and talked about topics in digital marketing today. After today you’ll walk away with the beginnings of a game plan on how you might learn more, incorporate or dive into AI marketing for your organization. And why is that important? And now, well, AI is rapidly transforming the media landscape, but what’s hype and what works? Good questions.

We’ve got a great panel with us. So let me start by introducing them first, lemme start with Amanda Feld. Amanda is a digital marketing leader from Google. She’s currently serving as a senior strategic agency manager. She advises agencies, clients on how to accelerate growth, how to incorporate Google products and adoption and shaping digital strategies. Next up, Inna Zeyger, our very own Amsive person. Inna is a respected voice within the digital industry and as VP of Digital Media, she leads Amsive’s paid media team delivering performance-driven strategies, mentoring high-performing teams, and championing innovation.

And I’ve said that like four times and I still, those two words together keep catching me. But across the digital landscape and me, I’m Joseph Sharp. It’s my pleasure to moderate today’s discussion. I’m an agency vet, been around for a while, I’ve worn just about every hat that there is to wear in the business. But here at Amsive, I operate as VP client experience and my core responsibilities are to be the voice for our clients internally with our teams and the voice of our agency externally to our clients and partners helping to drive collaboration and innovative strategies.

Just a couple of quick housekeeping things before we dive in. First, there will be a q and a session afterwards, so write down your questions, throw ’em into the chat, we’ll catch up to them towards the end. If we don’t get to ’em, you can always reach out to us to follow up. Also, presentation materials will be available through email afterwards. Sorry.

Why AI Adoption in Paid Media is Critical Now

So let me ask the first most pressing question, why is AI adoption and paid media so critical now?

So let me answer part of it and then I’m going to throw it over to our friend Amanda. So AI isn’t a trend, it’s a foundational shift in our media landscape. So think of it as the printing press, cable TV, streaming audio, CTV. These things made significant changes to the landscape that we knew and were comfortable with. AI is doing exactly that. And because we’re all here with Google and you’re all here with Google and Am IV and Google together, bring platform innovation and strategic know-how. This is what we’re going to talk about today.

So Google has been at the forefront of data quality and marketing innovation since its inception. So Amanda, let’s start with you. In the era, in this era, how has Google’s AI evolved and how should marketers think about its impact today and ensure they’re ready to utilize it properly?

Amanda Feld (04:09):

Yeah, again, thank you so much for having me here. As Joseph mentioned, AI is not completely new. It’s been around for years in different capacities. And as you can see on this timeline with PMax DemandGen, AI Max, it’s not a new concept, but its capabilities have been rapidly growing. So we’re unleashing the full potential of search and those who are not adapting are going to have an increasing difficult time catching up as new innovations keep rolling out. So as it relates to search ads, AI is allowing us to again reach people in moments that didn’t exist before.

So things like AI Overviews didn’t exist and AI Max and in ways we couldn’t before. So keyword predicting what a person needs when they need it, and we’ll get into all that. But it’s important to note that not only is paid search different than it was 10 years ago, the way people are searching is also different.

So if you’re using some of those old search strategies, they don’t translate as well to the way that people are acting now and you’re missing out on a huge group of people that your competitors might be reaching. So for example, multi-step reasoning is a big thing we’re seeing for complex questions. For example, show me the best yoga or Pilates studio in New York City and show me info on their intro offers and walking time from Chelsea.

That’s a lot of questions and it used to be those were separate questions somebody would ask Google. Now people are asking them together. So using strategies like broad match will allow you to capture that leader sale for these users who are starting to do some multi-step reasoning and making sure that you’re adapting to that. And then another example is a user might ask a question, how can I bring my dog on an airplane?

What’s needed? Instead of just having the answers of, well, here’s the airlines requirements for carrying a dog on the plane. The user can then be served an ad for a carrier that fits the size required for under the seat because they no longer have to ask that question. With the data being put in, the system knows that the next question they’re going to ask is, oh, okay, I need this carrier to be this size. What are those carriers? Where can I find them?

You’re making sure that you are finding the user in the moment they need it, and sometimes that explicit question might not even be asked, which is why you need to put the data in. So if I had to boil it down to one word, it would be data. If your data’s accurate, if you’re being using it in your advertising strategies, then you’re setting yourselves up for success.

Companies don’t get paid on the number of clicks on their ad, they get paid on sales or leads. So making sure that you’re using that data accurately. So last thing I’ll say is if you’re going to drive from New York to Canada, New York bias in New York, you’re not going to put Canada into Google Maps, you’re going to put something in like Whirlpool Golf course in Niagara Falls eventually. Either way you’ll get to your destination, but one is a lot more efficient because it has more data to work with. So similarly for your ads campaigns, they will work, but to be more efficient, you want to make sure that you’re putting in the most accurate and robust dataset possible so that you can get the results that you’re looking for.

Advertiser Reactions to AI Evolution

Joseph Sharp (07:06):

Wow, thank you Amanda. And there’s a lot there, right? How we’re using the channels, how consumers are using the channels, how we’re using the channels to market through, what’s the best way to go about it? Inna you absolutely have your ear to the ground on media innovation in your conversations, how are you seeing advertisers reacting to the consistent constant always on evolution of the media landscape?

Inna Zeyger (07:35):

I feel like this might be my favorite question and probably something that I’m sure Amanda, you’ve heard a million different clients ask about too or just express a lot of concern. So I can share what I’ve observed from my experience over the last, I would say two years where this conversation has just really evolved and come on the forefront. So we’ve had dozens of client conversations and the reactions to AI and paid media I feel like tend to fall into two big buckets. And then there’s this third mindset that I think we’ll talk a little bit about is where the real opportunity is.

So the first really has been, I don’t want to call it this, but I’m going to call it the fear and control group. These are clients sometimes really experienced ones who feel a real loss of control, right? They’re used to dialing in every lever manually understanding exactly what keyword triggered what conversion, seeing performance broken down to the decimal.

And I think that’s how we’ve always approached it too. But now with ai, it can feel like the machine is making decisions behind a curtain and that could be really, really uncomfortable. And I have a deep empathy for this mindset because that instinct to protect brand integrity, to be accountable for each scent of spend and to avoid surprises, it really does come from a place of care, not necessarily, but it can also paralyze progress if we don’t shift how we define that control.

The second bucket that I’ve seen in terms of conversations has been that these clients are really intrigued, right? They’re asking thoughtful questions like how does the system learn? How can we feed it better data? What happens if we test this asset, right? These teams aren’t all in yet, but they’re open. They want to understand the levers they want willing to test.

And that willingness is super powerful. That curiosity is really the gateway to that transformation and to being AI-ready. There’s also sometimes this element of I need AI right now or else, but not necessarily knowing where to start. And we hear those marching orders often too. The third mindset is where we’re operating at Amsive, and we’ve kind of have seen that bucket growing, which is this more strategic experimentation mindset. It’s asking what are our actual business goals? How can AI help us unlock new signals, scale or efficiency to hit those goals? How do we test in a way that’s measurable?

So that’s where we live. We live in this sort of third lane of structured testing. So we come in with frameworks, not theories. We don’t just flip the switch on, we set up controls, hold out benchmarks in a plan. So this third group or this third grouping, we’re seeing it growing. And the marketers with that mindset, they’re really the ones that are going to get real lasting results from AI adoption, not because they trusted the algorithm. I know trust the process isn’t always the answer, but because they understood their strategy deeply enough to teach the machine what mattered.

Google’s AI-Enhanced Marketing Tools

Joseph Sharp (10:30):

Yeah, I think from my conversations with clients, I will second, third, and fourth that right? And my perspective is sort of a growing, evolving perspective, but it’s exactly what you were saying. Those companies that lead with curiosity and test and learn their way in are the ones that are really going to find that longevity and success. All of AI isn’t right for everyone, but some of AI you’ll find what fits for you, for your clients, for your brands. So I definitely a hundred percent agree with that sort of framework.

Amanda, I do want to shift back to you because central to AI marketing are some of Google’s tools that are leading in the AI enhanced marketing landscape. So given everything that we’ve just sort of touched on, what are some of the sensitivities around AI adoption and the evolution of some of the tools? Can you give us a little bit on how Google is continuing to be a leader in the marketing landscape and what is this AI Max thing that I keep hearing about and why does that matter?

Amanda Feld (11:51):

Yes, this is definitely a safe space. So you might’ve heard the term AI Max and not wanted to ask what it was. No problem. We’re here to tell you. So AI Max began the global beta rollout around May with Google Marketing Live. It’s not a brand new campaign type. It brings the best of Google AI to your existing search campaigns. So it’s a suite of three settings that you already know layer on top of your existing search campaign. So there’s search term matching which captures demand based on intent and not just exact terms automatically created assets, things you already know from Max or DSAs, and then URL expansion.

So again, it’s toggling on these suite of settings that work best for you and your company that are added on top of its search campaign. And like Inna said, AMSTA is already doing a great job utilizing the strategy and there are some people who are more hesitant to do it, obviously.

Again, it is newer, you’re not sure. So I heard this great analogy, I’m going to repeat and see if it sticks with everybody. I was going to put up an image, but I’m pretty sure we all know the hostess snack Twinkies with the golden cake and the vanilla cream filling. Apparently the filling used to be banana flavored. And then during World War ii, the boats were needed to do more important things and they can no longer import the bananas.

So they switched it to vanilla. And at first there was pushback because that’s radically different. And then people realized it was good, and then eventually the war ended and the boats were available again, but nobody wanted to go back to the banana flavor because they were happy with what they had. Now, so similarly, the same thing is true. Now things are different. We’re not saying that it’s the same, but it is an evolution and we’re probably going to look back to this time and say, oh my God, I can’t believe we used to have these endless exact keyword lists all the time.

But it is a test and scale and it is a reason why Amsive has built out this framework of how to do it because different people have different comfort levels and we need to make sure that you are working within your comfort level without falling behind. So back to AI Max for a second. In order to run AI Max and other of those AI driven ad products, you do need to make sure that you have the foundation set up. So as I mentioned before, and I’ll probably belabor the point, the rest of this session data is key as well as broad match.

So making sure that anything you’re using within the ai, you are using creative, the data should be your North Star. What you want to achieve needs to be clear so that way you can set your strategies up to achieve that. So the last thing again is that AI Max is on top of search. So it’s not necessarily a brand new thing, but it is a new way of reaching your consumer.

AI Max Testing Results and Case Studies

Joseph Sharp (14:30):

It’s an additional tool in the toolbox of the ecosystem. So we’ve got these, this idea that this concept that we can be smarter by layering these things on, but what does it really look like in the real world? And it’s new, but we’re a premier Google partners, so we take advantage of the opportunity to do some things. So I know we’ve done some testing for a few clients in, do you want to touch a little bit on what has been tested and what we’ve seen work thus far with AI Max?

Inna Zeyger (15:06):

Yeah, 100%. So this is very, very, very much the new. So some of these outcomes are very, very preliminary. It’s been rolling out, and don’t worry if your client teams have not been talking to you about it. They probably are, and we’re looking at ways to roll that into the testing that we’re doing. But I would love to walk you through three different case studies. One of them is two, well there’s two of them that are focused on eCommerce, and it’s sort of like almost this tale of two cities. And what we’re seeing so far is that the results are varying very significantly by industry and approach.

But don’t worry, we’re going to also spend a lot of time at the tail end of this webinar talking through what those frameworks are for introducing AI Max into the mix. OOH, that primed and being able to assess what is brand safe for you, what is the right levers to pull?

So talk through one of the very first case studies and as Amanda mentioned that there’s a couple of options there, right? It’s an expansion, it sits on top. So you have the option of using keyword expansion. You have the option of using ad expansion as well as landing page expansion for the most part, for a good chunk of our clients that we’ve been testing on. We’ve just started with a very conservative approach of doing keyword testing. And we also don’t have a huge volume of data, but some of these are just typically just starting off with one to two campaigns.

So the first one you’ll see is a B2B ECommerce client, and the control is just the regular old, boring regular campaign. And the test is with AI Max enabled, in this case it’s been keyword expansion, it’s been running for a couple of weeks, and so far we’ve seen that there was an 8% lift in orders at a 15% lower cost per order.

But the key metric here that I’m really looking at is that we’re having a 39% higher average order size, which is great. So we’re getting lift across the board and I think what’s happening here is AI Max is finding higher intent, higher value queries that just we weren’t targeting manually, but again, noticed that we started very conservative with keyword expansion only.

So we move into the next case study, which is our B2C eComm one. That was the one where we started off with a full expansion. And when I say mixed result, I mean mixed results. What you’ll notice here is in the first couple of days of launching, you’ll see this huge spike in cost per conversion, which like, oh no, that looks very, very scary. We had actually launched with all three turned on landing page ad and keyword expansion enabled, and that really spiked our cost per conversion for that particular one.

So we actually pulled back. Now typically would say you want to give these things more time, but again, these have been very controlled experiments. But also I wanted to show you something that wasn’t an immediate success story with testing out AI Max. But we did start seeing that once we pulled back on landing page and ad that it started to normalize a little bit and now we’re not too far off from our baseline versus the control group. And then lastly, I want to talk through a very, very, very different vertical.

I know we have a lot of folks who are in highly regulated spaces, so there is sort of that larger fear of losing control mostly because of regulation and rules around what you can advertise, how you can advertise. But we also started testing AI Max in a financial services client. In this case, we wanted, again to start quite conservative, focusing in on just the keyword expansion, seeing what we could capture.

And this one, again, in a very short window of time, started seeing pretty solid results, 18% increase in conversions, we saw a really good drop off in cost per click and just an improvement in cost per conversion. What I’ll say here is, and what has been surprising in the very, very short time that this has been running and available is that we’re seeing strong results even at relatively low levels of media investment, which is quite different than what you may have seen in the past with things like Performance Max or introducing broad match. So I think one of the takeaways here is that you don’t need a massive budget to start testing AI Max effectively.

The other thing is that we’ve seen better early outcomes with keyword expansion versus add and landing page expansion. This does not mean you don’t use these features. There are tests that you could do to start rolling in some of them, but be mindful that your site does need more depth and sufficient content for landing page expansion to work really, really well, right?

If your site is two pages, it’s not going to have enough to pull from, but if you have a lot of depth of content, you’ll have better success with that. For more regulated industries, start with keyword expansion or only again, add and landing page expansion can introduce some compliance risks, but that’s something that you can test iteratively once you make sure that the content is solid and more importantly, out of everything. Again, these are single campaigns or really small sets of campaigns that we’ve tested. Some of these outcomes have been really, really positive.

So I want you to think about how much incremental performance you can achieve at scale across your entire account if you start testing that. So the pattern we’re seeing is that AI Max works, it’s still really early, but approach matters and it’s very much dependent on your industry, your risk tolerance, and just your account maturity.

Joseph Sharp (20:53):

Yeah, I think a couple of really, really great points there. What is your risk tolerance lead with curiosity? Like we said at the very beginning, let’s walk our way into it. And for the most part there will be some brands that are very interested in jumping in head first and wanting to run, but I always, and I’ll talk about this again a little bit later, I always say test intelligently know what you want to get out of it before you go into it, and that will set the stage for learning and success in the future.

So we all love a good case study and I actually don’t think I’ve seen a bad case study. So the relevance with these are, it’s new, we’re testing, it doesn’t take much money, it doesn’t take as much time to start seeing some positive results in these experiments that we’ve been running, but there’s not enough to say that this is really truly a pattern yet. So they’re not necessarily truly typical results. So what should brands be doing right now to stay competitive?

Inna Zeyger (22:00):

Oh yeah, this is the fun part, or at least because we’re about to throw quite a bit of content and frameworks your direction, which is I think what we’re all looking for. So what we need to talk about is not necessarily just the AI piece, but what are the frameworks for getting your accounts AI-ready? But I want to be so crystal clear, these are tool sets, not strategies.

So these frameworks show you the testing structure, but the strategy behind how you roll it out, which campaigns get tested first, what success looks like for your specific business, for your vertical, when you know it’s time for the next rollout, all of that lives in your data and needs to be rolled out strategically so you don’t disrupt the funnel. That’s usually very top of mind for us. So there also have to be guardrails in place.

So whenever we run any sort of testing and rollouts, like how can we roll back changes if things don’t work and retest in the future? Because you saw in some of our AI Max examples, not every test is immediately successful. Some need more time, some need larger budgets to identify what works, and that’s just a reality of any sort of AI testing.

Last thing I’ll say before we start getting into the frameworks themselves is it’s completely okay if your whole account doesn’t have every bell and whistle or modern search structure AI tool applied, that’s not the point. The point isn’t to use these tools just because they exist. The point is about being intentional, knowing what to roll out, where when, and how to measure it. And that’s what makes you AI-ready, not just flipping a bunch of switches.

Joseph Sharp (23:44):

And I think you had mentioned that we have three frameworks to walk them through. So we’ve first a testing framework for broad match.

Inna Zeyger (23:53):

I know everybody is so tired of hearing about broad match, I’m so sorry, but it does need to be talked about. I’m sure Amanda feels the same way as well. But here’s the thing with broad match at this point, it is foundational. It’s your entry point into showing up in AI ads Overviews and being able to scale your account. We know that exact match is no longer exact, right? That’s just been the case for a while now.

And what we have here really are three approaches to testing and introducing broad match into your accounts organized from where we start with the most control because we also, again, are empathizing with those folks that don’t want to lose that control. And the beauty is is that you can apply this campaign by campaign or by sets of campaigns based on your performance data and just your threshold for change and your risk tolerance.

So what you have here is a slide that outlines these for you, and I’m sure you’ll be able to take this away once we’re done with the webinar. So let’s start with going from the least amount of control, which is giving Google the reins, right? So this framework really is using Google experiments and kind of looking at the truest form of modern search structure, which is consolidating all of your match types into one ad group. So having broad phrase and exact altogether other scenarios where you’d use this.

If your account campaigns are primarily running on exact and phrase, whether that’s in the same ad group or spread across the campaigns, this is a way to safely introduce the consolidation. Just using that straightforward Google experiment. And what you’re trying to get out of it and understand is the impact of introducing broad match along exact and phrase in the same ad group you’re answering how does broad match behave when it’s consolidated into the true modern search structure?

Does it add value? Does it have daily targeting? Does it cannibalize performance or expand reach? And there are losses of control when you do this. You can’t control how much budget gets allocated to broad versus exact or phrase. You can’t have match type specific negative keywords. All these are running together and you have to really rely on the algorithm to identify which one is the best to serve at the right time. When does this work really well? High volume scenarios, right?

You have a lot of conversion volume, you have a lot of traffic like think a B2B software company with 200 conversions per month or an eComm brand hitting the ROAS targets who want to test if consolidation can unlock that additional scale. So if you have enough conversion data, the algorithm can optimize effectively across all three match types. We’ll move on into the sum control.

It’s our middle ground, which is also using Google experiments, which just gives you more control. It’s the understanding that giving you the understanding of the standalone performance of broad match in the same campaign. The first one was within the same ad group. This one’s the campaign gives you more levers. You can add exact and freeze match terms as negatives to the broad match ad group. You get some leeway and bidding as well on the ad group level if you’re using target TPA or TROAS.

And I apologize for getting super into the weeds, but these are the important guardrails and levers that we can pull here. And again, this is really good for clients who are more cautious about this, like financial services who need more compliance control or performance focused B2C brands that just want clean broad match data but need some more flexibility with how it rolls out.

And then the last but very not least is the standalone campaign. And this is the one that gives you the most control, but it doesn’t necessarily translate to understand how these three match types interplay with each other like the other two do. Setting up a standalone campaign just means you take an existing campaign, take all the exact and phrase, make it broad match, add exact and phrase as negatives to your broad campaign. And what this does is it gives you the highest elements of control, right?

It gives you budget control, bidding control lets you determine exactly how much you want to spend there. And this is really ideal for enterprise retailers, just enterprise clients who need to allocate very specific budgets, multi-brand accounts where different product lines need different strategies. It’s also perfect when you have limited budget and don’t want to disrupt existing performance. So I know we went through kind of bit with some of those frameworks, but again, there’s different scenarios there.

Broad Match Optimization in the AI Era

Joseph Sharp (28:22):

So I have a question that came up for me as you were going through that and with Broad Match being sort of more of an audience-first approach versus a exact match, what does optimization really look like when you’re using a broad match?

Inna Zeyger (28:39):

There are so many, and it’s not just with broad match too, and I think we have to look at it holistically as well. One thing I’ll just say around optimization as a whole is that your budget, your conversion volume and search volume really determine which of any of those tests are viable and which will give you those meaningful results. So there’s no one size all fits approach with any of the tests that we’re pulling together. But in terms of optimization, the best way that I can describe it now and how it shifted is it’s an exercise of exclusion instead of inclusion.

Because optimization has fundamentally changed. We used to focus on picking exact keyword, selecting placements, micromanaging match types. Now we’re focused on exclusion. And the reason behind that is because the platforms are designed to go broad by default, right? Exact match is no longer exact.

You might think you’re bidding on a specific query, but in reality you’re bidding on an intent cluster and platform is making a lot of assumptions for you. So instead of looking at it as how do I make it as specific and exact as possible, it becomes how do we shape what the system learns, how it spends, where it scales. And the answer is we optimize the inputs, not the outputs. What that means very, very practically from a day-to-day management perspective is like something with Performance Max means you’re actively testing asset group segmentation.

We don’t just throw all the assets into one bucket and hope the machine figures it out. We structure campaigns around different funnel stages, audience and 10 signals, product groupings if it’s e-comm and then we test creative inputs within each, and we use Google experiments to do that. Other things around exclusion, building and refreshing negative keyword lists, we have to do that a lot more now.

Actively mining queries, just not relying on some broader filters and feeding the system with clean conversion actions. Because if you’re optimizing to the wrong goal, AI will so happily go ahead and chase it. So I’ll end that part with AI without exclusions or audience signals as sort of walking into a restaurant and saying, bring me food. You’ll get something, but it won’t necessarily be what you need, what you want or what you can afford. So AI will take action always. It just doesn’t care if it’s the right one unless you tell it. So yeah, optimization still matters.

Joseph Sharp (31:14):

Yeah, it’s just different, but it’s still a critically important piece of the funnel. We always talk to our clients about yes, we are in the platforms all day every day doing touching, learning, optimizing, it’s just understanding that it’s a little bit different for these tools. Right? So next up we have Performance Max for PMax. Can you walk us through a framework there as well?

Inna Zeyger (31:42):

Yay, more frameworks. So again, and just before we jump into PMax, so you can look at all of these frameworks that we’re talking to through as iterative, right? It could be a place to start with broad match piece, you can overlay Performance Max and introducing that. And later on when we talk about AI Max, that’s another way to get an entry point or it could also be iterative for you to get started. But with Performance Max, fundamentally what we want to get across and understand with PMax is how much more we can unlock in terms of incremental revenue or conversions, right?

I’ve spent a lot of time auditing accounts recently where they’ve come in and we’ve seen that PMax, there’s been a sort of guidance to really, really push PMax, where PMAX is almost completely some of the more traditional search shopping campaigns. But the challenge there is that you’re losing out on all the finer controls, all the strategic levers you used to have.

And I don’t think that’s necessarily the direction that we should be heading into just yet. So we look at PMax as a compliment, not a replacement, but we need to know which levers you have available to understand first what incremental reach performance or conversions it might drive for you, but doing so in a very, very safe way. So there’s really two frameworks for testing.

There’s UPLIFT testing, and it’s really for understanding incremental performance. And it’s like, hey, if I add PMax into my campaigns alongside what I’m already running with search and shopping, what is it going to give me? You’re testing if PMAX can add value on top of your current strategy, and it gives you real data points on running that kind of test on incremental reach and conversions. There’s a lot of clients who don’t want the risk of testing it on its own by disrupting their current performance.

And this is a great way they can give you that information safely. And then the second one is really for figuring out how much budget to reallocate from your traditional campaigns to PMax. If PMax is driving performance for you, if you’ve introduced it, you kind of wonder understand what data points do you have to make decisions on budget allocation? But again, I don’t want to look at this as entirely a replacement strategy.

We have to be data-driven and methodical, and these two frameworks are really good ways to get those data points and move forward with proven information to decide how you want to allocate budget that aligns with your business goals. So yeah, I would say when to use each uplift is when you want to understand your incremental value by adding those in without risk for upgrade is when you need more budget allocation guidance.

Performance Max vs AI Max: Understanding the Differences

Joseph Sharp (34:43):

Okay. So we’ve just between broad match and PMax, we’ve thrown a whole lot of information at folks. Amanda, again, I want to bring you back in from a Google perspective, and Inna had mentioned that PMax isn’t necessarily a replacement, it’s a compliment. How are you seeing it activated through an overall media mix complement, replace? What types of brands are using it in these types of frameworks as well that you’re seeing?

Amanda Feld (35:15):

Yeah, I think one of the questions we’re hearing a lot is what do PMax and AI max have in common? How are they different? Obviously they both have the word max, but if that was my whole answer, I wouldn’t be that anymore because that’s not helpful. So beyond just both having the word max, they are different campaigns.

So AI Max, which is the newer one, aims to compliment PMax, not cannibalize it. So AI max campaigns enhance, like I said, your existing search campaigns. They are search only runs across all of Google ads channels. So search display, YouTube, Gmail, discovery shopping, there’s a lot. They ensure that you’re able to reach the user wherever they are. And AI Max is ensuring that you’re able to most effectively reach your consumer within the confines of search. So you should be using both in order to get the most out of your strategy.

You do want to remember that with both of them you’re adhering to best practices. So for PMax, are you using video and not just images? Are you using horizontal and vertical so you can show up in all the opportunities that make sense.

Spoiler alert, you can use AI to help you with reformatting videos. But just making sure that whichever method you’re using, you are adhering to best practices, but they should be done to compliment each other. One is not cannibalizing, the other one is not in place of the other. They are here to do different functions.

So as long as you’re aware of what the purpose of each strategy is, you’ll be able to analyze with Amsive, okay, is it accomplishing this goal? Should we be increasing it? Should we be expanding it? How does that work? Again, laddering up to a specific goal for your company.

Joseph Sharp (36:48):

Got it. So the short version of the answer is it depends and let’s talk about it and figure out the right mix for you. Exactly. Because really that’s really, there isn’t a one size fits all and it’s not necessarily the most complicated answer in the world, but let’s not just a plug and play, let’s talk about it. Let’s find the right fit for your needs, understanding what we want to get out of it in the end. Right.

So we’ve touched on two of the three frameworks that you mentioned before. AI Max is the one that we haven’t touched on. And Amanda, I think that was a great segue in how Max and AI Max play together. So can we walk through the AI Max framework?

Inna Zeyger (37:32):

Sure. And I would say this is probably the easiest one out of all of these. And as you saw, we’ve been running testing and across a number of accounts across different verticals. And this is something that just started really rolling out globally now. And really what we want to understand is how much incremental volume and efficiency can AI Max deliver when we let it optimize our search campaigns? So really have two approaches here, which are really, really easy.

So with Google experiments, having AI Max on versus off and really using an experiment to split the traffic between your campaign with AI Max enabled and disabled. So you have to make decisions to, and these are some things that you can later test iteratively, but you have to make decisions on whether you want it to be just keyword expansion or if you want to introduce landing page or add assets.

But a safe place to start is with keyword expansion. So what you’re trying to get an answer out of this type of test is which AI Max optimization should I enable and how much lift did they actually deliver? So I know we walked through a number of case studies, again very early on that demonstrate some of those early results.

And then there’s the other way, which is if you’re very gung-ho and you just want to opt in is just turning it on and it’s taking an existing campaign and looking at it more from a pre-post analysis, of course it’s less controlled but faster to implement if you want to test AI Max without setting up those experiments. And we have plenty of clients who are very gung-ho and want to do something like that, and also clients who want to do it and with more of a guarded framework too.

But yeah, it’s really just about which one is going to give you the cleanest data and have enough volume for statistical significance. You can use the manual if you want faster insights because it’s just turning it on. But again, the last thing I’ll say on that is that what we saw in our case study is it can deliver significant performance improvements, but it’s not guaranteed. Some tests are immediately successful.

That’s why having proper measurement and rollback capabilities is essential. And I would say just with anything that we’ve talked about, a good way to look at this type of rollout is just having a separate testing budget dedicated to it that doesn’t operate on the same KPIs, maybe some of those other campaigns or how you holistically view media because that’s a really good safe way to be able to test many, many things and be able to allocate enough budget to get those outcomes and results that you’re looking for.

Joseph Sharp (40:29):

Awesome, thank you. So with the three frameworks, and I think we have at least the start of an idea of why AI adoption is absolutely critical, maybe even more than the start of an idea, we know why ai AI adoption is so critical right now. This is the way that the industry is moving and it’s been moving this way for a little while. We’ve got the three frameworks to consider.

And then you sort of touched on it just now, but what advice for our friends in the more highly regulated industries and or industries where they’re hesitant about ai, what advice would you share with them in terms of the approach, which we’ve talked about approaching with curiosity, but what advice would you give to them? Would you share with them from the experiences that you’ve had thus far?

Inna Zeyger (41:22):

Sure. So adopting AI does not mean adopting all elements of ai, nor is it a requirement. I think I mentioned earlier, you don’t have to have every single bell and whistle enabled with AI Max. You can start off with just keyword expansion, not add copy, your landing page expansion if you specifically need to preserve the messaging that you’re putting out there because it is highly regulated with some of the other tactics, you can make sure that you turn off some of those AI optimizations like in PMax and make sure that the ad copy remains, the ad copier, not include landing pages in it as well.

But I think it’s really about figuring out and making sure that you’re leveraging the AI features that work behind the scenes like algorithmic bidding or some of that keyword expansion in order to still move forward but still protect the brand.

Aligning AI Tools with Business Outcomes

Joseph Sharp (42:29):

Got it. And I guess I’ll throw this out to both Amanda and Inna. How are you guys seeing brands that you are working with now actively aligning to these larger business outcomes or sorry, aligning the pieces to get to their larger business outcomes? We’ve got our groups of clients internally, and Amanda, you’ve got a little bit of a wider breadth of experience. Are there hallmarks? Are there things that you see certain brands doing to align these pieces that we haven’t touched on yet?

Inna Zeyger (43:08):

Can do. I’m very good at rambling. Alright. So for the clients where this is successful, it’s about measurement and holistic thinking, and I can’t stress this enough. So you cannot look at any of the outcomes, whether it’s testing broad match or anything else that we’re testing as an island. So whether you’re running Google experiments or standalone campaigns, you have to evaluate the net effect on your overall account performance.

So don’t just look at broad match metrics in isolation. Look at what happens to your total cost per acquisition across all campaigns. Look at your overall conversion volume, your impression shared your reach. Because what we often see is that these tests might have a higher individual CPA, but it’s driving incremental conversions that you wouldn’t have happened otherwise. And so when you look at that blended performance, your overall efficiency actually improves. Other things I would say is you can run incrementality tests.

I mean those are fairly straightforward and very powerful. You want to see how PMax is performing or not performing, turn it off, see what happens. You don’t have to do it across the board, but you can do geo-based incrementality testing holdouts to really prove how effective these channels are, again, without disrupting the funnel. And last thing I’ll say is looking at it from a growth versus efficiency perspective. So yes, focusing on efficiency is great, but there’s another side to the equation.

Growth and scale, these frameworks aren’t just about maintaining what you have, it’s about your ability to grow the business. So if you’re already maxing out your exact and phrase match keywords, if your impression share is capped, if you’re CPCs are climbing because you’re competing in increasingly competitive auctions, broad match can be your path to sustainable growth.

But it’s about comfortably dipping your toes into the AI driven direction that Google is very clearly has been saying, this is what we’re moving towards. And I think you have to ask yourself the question, it’s not about whether AI is coming, it’s already here. Do you want to test and learn on your terms with proper controls and measurement in place, or do you want to wake up one day and realize that the platform and your competitors have moved on without you?

Joseph Sharp (45:46):

Yeah. Excellent. Excellent. Thanks. It was a lot.

Inna Zeyger (45:49):

I don’t mean to scare anybody, but it is how we have to look at it.

Joseph Sharp (45:55):

Yeah, absolutely. Absolutely. Amanda, was there anything that you wanted to add in there at all, or you think you touched on all that?

Amanda Feld (46:02):

The eye rolls. As I say, I can feel the eye rolls as I say, the importance of data, but I think that NInna hit it on the head. I mean, you want to make sure that what you’re doing helps. So if you’re just looking at, let’s say my CPC is going up, okay, but what does that really mean? When you start looking at things like smart bid and conversion values, you want to make sure that you understand the true impact to your business. So maybe their cost per clicks a little bit higher, but now you’re looking at an actual closed lead versus just a lead form. So having that complete picture will help you really move forward with these AI implementations.

Key Success Principles for AI Adoption

Joseph Sharp (46:38):

Yeah, I think that’s really, really smart. And so I think you two have given us a ton to think about. We’re nowhere near done well sort of. But I do want to bring us back to the start of the conversation. You both touched on things that are really, really important that go back to where we started from, which is why is AI so important and why is it important right now? But also testing, learning, and curiosity. So if we think of AI as the engine, then what needs to be in place to drive those results? And I would say based on our conversation first, understand AI enhanced marketing and be curious about it.

We’ve been using that word curious a lot because we really feel like this is the time now to start learning. Even if you aren’t turning it on or turning it on fully, learn about it, ask questions, reach out, find out more, because all of AI might not be right for everyone, but I would promise that some of AI is right for you, and all of it might be right for you as well. But let’s find out second, and this is my soapbox.

When you’re ready to start testing these tools, other tools, the single most important thing that you have to do is know what you want to get out of it before you start. So the single most important thing you need to have is good data. The single most important thing you need to do or understand is what you want to get out of it before you start running a test and just waiting to see the results to understand if you were successful or not won’t actually lead to success. You won’t necessarily have a good understanding of what the results mean. So I talked with my clients a lot about how are we going to design the test so we can learn something from it hopefully, and get business success results out of it.

I think a good test always begins with that understanding. Next up, know your business goals, right? We’ve talked about this before, whether it’s a CPC or a lead, that understanding what that is is critically important because that’s going to help you to understand how to design the test. And Amanda, I think to your point, you were saying looking at traditional metrics may not necessarily be the right answer or the full answer, right?

If we look at just click-through rate or CPC or impressions, that’s not going to give us necessarily the whole answer. And with these tools in place, we can look further down the funnel of metrics and deeper into the business side of metrics. And I think that’s a critical piece as well. Don’t just focus on the media metrics, look at what’s happening, what the impact is on the business. Lastly, critically, critically important, create full funnel alignment.

You started to allude to this and it really as a client lead and working across many, many channels, anytime you’re building a test, anytime you’re implementing a new product or a new channel, making sure that all of your channels are in full alignment. So when you’re running that test, you get clean results, and then when you start to implement, you get clean results that you can read.

Super, super important to make sure that you’re getting insightful learnings right. Okay. Sorry, I’m going to get off my soapbox right now because we’ve talked about some of the tools in the Google toolbox. We haven’t actually talked about video, which is a critical piece of many marketing plans. Amanda, can you give us a peek into how YouTube fits into this ecosystem?

YouTube and Video in the AI Ecosystem

Amanda Feld (50:37):

Yes, I know we are getting close to the end, but we cannot not mention YouTube at all, even though we are already sliding it by talking about it so quickly. But YouTube is still the number one in TV and streaming. There’s over a billion hours of YouTube daily on TV screens, which is actually twice a daily watch time of Netflix. We also have YouTube select shorts ads that are viewed longer than TikToks according to Media Science report. They’re also more personally relevant. So YouTube is huge.

I will not be covering YouTube in nine minutes, but there is a lot that can be done across the funnel, not just upper-funnel. And AI plays there as well. So AI generating assets means that you have the ability to create more assets. You can get into moments that are culturally relevant faster. You can generate horizontal and vertical images.

You can have multiple ad variations to test in the system with the proper data and creative, we’ll be able to find which combination works the best for the rich people. So making sure that you do have an understanding of what YouTube has to offer, where does AI fit into it?

There’s the creative side, there’s a targeting side. Brand performance is this term. We’re hearing of branding and performance together. YouTube’s no longer just upper-funnel, so where and how does it work for you? And then how can AI improve? That is a great conversation that you should make sure that you’ll be having with your teams.

Joseph Sharp (51:58):

Yeah, absolutely. Absolutely. And again, full-funnel alignment I think is really, really important. We really are getting close to wrapping up here. I know we’re coming up close to time and I don’t know about you all out there. I have a ton of thoughts that I didn’t have before and I actually work with on a daily basis. I also have more than a few questions for Inna and Amanda for my clients as well. More conversations to be had Inna, if there’s one thing to take away from this, what is it for you?

Oh, you’re on mute.

Inna Zeyger (52:53):

Oh. And of course Microphone decided to stop working when I asked said so I’ll be very, very brief in that case. These are just frameworks. They need a strategy to be put behind them and more importantly, your data to support how we roll these things out. So these are conversations we have to have and there’s no one size fits all approach, but that’s something that we’re more than happy to have more conversations about.

Q&As

Joseph Sharp (53:22):

Alright, cool. I would say I would add to that don’t experiment blindly talk to a professional test with structure and care. So lastly, if we didn’t answer your questions during the conversation, and if we don’t get to them during our q and a, which is coming up in just a second here, please feel free to reach out to us. You can see how to contact us right here on the screen. So definitely do not hesitate to reach out. And I say let’s dive into your questions.

Leih Boyden (54:06):

Yes, the first question’s for Amanda, we know that consumer behavior is changing on Google. How should we interpret that? What does that mean for your Google Ads strategy?

Amanda Feld (54:19):

Yeah, I think a lot of things are kind of buzzy, but you don’t really know what it means. It used to be that people were doing one thing at a time. I would sit down and watch TV or I would look something up on my computer. Today, consumers are seamlessly and simultaneously moving across four behaviors. So you’re searching for something while you’re scrolling on your phone, while you’re watching tv, you see an ad and you start shopping it. So leveraging AI effectively is how you can keep up with the pace of change.

Things like PMax do let you go across different formats to meet people where they are. And like I said earlier, people are even searching differently. So they’re using a lot of questions into one question or the algorithm understands what they’re asking before they even ask it. So making sure that you understand today’s behavior and how people are acting will allow you to better reach them and resonate with them because reaching them is half the battle and then resonating is what you probably care about the most.

Leih Boyden (55:17):

Thank you. The next question and final question for Inna is how can healthcare companies specifically, can they use all of AI Max in their industry? They can’t necessarily use automated ad content. Their ad copy has to be approved prior to use. So how could healthcare companies use all of AI Max and stay compliant?

Inna Zeyger (55:44):

Gotcha. I think we talked quite a bit about the keyword expansion for AI Max. That’s one component. You can definitely get started with some of those other elements and this is where I think we need AI Max to do a little bit more testing with. And I definitely would lean on Amanda for more feedback on how some of the information is pulled. I would say there are a couple of things that you could do. We talked a lot about exclusion.

So for example, for landing pages, if you don’t want certain landing pages to come up or be used in AI Max, you can exclude those. So that way it’s only what you’re controlling that’s getting pulled in for ad copy. I think that one might be definitely more of a challenge. And I think there is another beta that’s coming out, right, Amanda, that may account for some of that, but I don’t want to speak on Google’s behalf.

Amanda Feld (56:49):

Yeah, I think that generally speaking, there are no restrictions. I mean, there are the normal brand safety and lead quality and all those restrictions that everybody has, but healthcare, finance, those kinds of verticals are able to use AI Max. I do recommend reaching out to Amsive. My team will partner with them as well to make sure that we understand nuances. So if there’s also a difference between allowed and comfort levels so we can get into that. But the blanket answer is that there are no restrictions for the product.

Joseph Sharp (57:21):

Alright, well first thank you all for joining us. Thank you for the questions. Again, if we didn’t answer your questions throughout the presentation and these last few questions didn’t cover what you were looking for, please feel free to reach out. Like we’ve said, we’re here to help. We’re here to help you. We’re here to help your teams to make the most out of these Google AI-powered tools and the ad ecosystem overall across your paid media investment.

So don’t hesitate to reach out. We’re looking forward to the opportunity to collaborate with you and help your brands grow. Lastly, a very, very special thank you to Amanda and Inna for joining us today and sharing their expertise with us. And to all of you who spent part of your day with us, thank you so much for gracing us with your presence and have a great rest of the day. Thank you.

Inna Zeyger (58:17):

Thanks. Thanks for spending your lunch with us. Bye.