Google Marketing Live 2025 Exclusive Insider Intel With Ginny Marvin

Episode Description

In this conversation with Google Ads Liaison Ginny Marvin, we go beyond the Google Marketing Live 2025 announcements to examine what they actually mean for performance marketers. We take a closer look at:

• How AI Max Search fits into the future of keyword-based advertising
• The logic behind Google’s “Power Pack” approach: PMax, Demand Gen, and AI Max
• What’s changed in bidding, conversion signals, and campaign structure
• Whether new controls and reporting options meet the needs of advanced advertisers
• Where automation ends and strategic human input still makes the difference

Episode Takeaways

Ginny’s dual perspective as a seasoned digital marketer with two decades of experience and Google’s primary liaison to the advertising community offers a well-rounded view of how the changes announced at GML 2025 will affect advertisers at every level.

If you’re looking for direction in this evolving landscape, the discussion delivers practical guidance on where to focus, which skills to build, and how to position your strategy for success in an AI-driven advertising world.

1. How AI Max Search fits into the future of keyword-based advertising

Ginny explains that we’re experiencing a transformation as significant as the mobile revolution. Search queries are getting longer because the entire nature of how people express intent is changing.

“We’ve gone from the very short two to three keyword phrases - super easy to target those with two to three keywords in your account. But as queries change and evolve, that’s where you’ve seen broad match aimed at capturing those broader search queries.” explains Ginny.

AI Max Search represents Google’s solution to this problem. It is designed as a one-click toggle that adds AI capabilities to existing search campaigns without requiring restructuring.

The system includes:

• Text customization (formerly Automatically Created Assets)
• Query expansion
• Final URL expansion
• Geo-targeting at the ad group level
• Brand inclusion/exclusion control

Instead of replacing keyword-based Google Ads, AI Max acts as a translation layer that keeps campaigns compatible while adapting to modern search behavior.

Ginny explains that AI Max is somewhat of a first step in a longer transformation where keyword-based advertising adapts to conversational AI rather than being replaced by it.

“AI Max for Search campaigns is aimed at helping address this new landscape of query methodology and modalities. This is a really good example of how Google is listening to customer feedback about migrations and evolving their campaign structure” says Ginny.

2. The logic behind Google’s “Power Pack” approach: PMax, Demand Gen, and AI Max

Google’s new “Power Pack” builds on last year’s “Power Pair” by adding a third AI-driven campaign type: Search with AI Max, alongside Performance Max and Demand Gen.

It gives businesses flexible, AI-powered tools that enhance (not replace) human strategy, allowing advertisers to adopt automation at their own pace within a unified system.

The core strategy includes:

• A unified AI foundation. All three campaign types share an AI core but service different goals.
• Options to mix and match what works best instead of mandatory shifts.

“I think the focus is really there’s optionality. You have the options to choose the set of campaign types that make sense for your business. Many people are using both Smart Shopping and Performance Max, or Display and Demand Gen. Mix and match.” Ginny said.

• Rich creative is front and center. Visuals and cross-channel reach are key to engaging today’s streamers, shoppers, scrollers, and searchers.

3. What’s changed in bidding, conversion signals, and campaign structure

Ginny and Fred also discuss changes to key aspects in Google ads like:

• Smart Bidding Exploration allows more aggressive bidding on uncertain queries without affecting high-confidence bids
• Primary and secondary conversion events now directly impact bid management performance
• Campaign consolidation becomes crucial for providing sufficient conversion data to AI systems
• The system can now explore lower-volume queries that previously lacked data

“What matters to your business should inform your structure. That would be if you have campaigns that necessitate different ROAS targets, for example - that will dictate a structure. But if you look across your account now and have five, ten campaigns that essentially have the same goal, that could be an opportunity to consolidate under that one goal” shares Ginny.

4. How well do these new tools help advanced advertisers?

During the discussion, Fred highlighted a common frustration with Performance Max: while it often delivers strong results, much of the spend goes toward low-hanging fruit like remarketing or branded search. Areas that advertisers feel were already performing well, leading to concerns about cannibalization rather than net new growth.

Ginny says that Google’s new features aim to address advanced advertiser concerns, especially around Performance Max transparency but lean more toward visibility than real control.

Their true value will hinge on whether added reporting leads to actionable insights, not just more data.

For sophisticated advertisers, tools like incrementality testing and the Data Manager API stand out. They go beyond surface metrics, helping solve core measurement and integration challenges that support deeper strategy and optimization within automated campaigns.

5. Where automation ends and strategic human input still makes the difference

Ginny emphasizes the fact that fundamentals are becoming more and more important.

“I really just see the fundamentals of marketing are just becoming more and more important. They’re not - the idea of you need to reach an audience. You know who your audience is. You can know what your brand is, what’s going to make your brand stand out. Those pieces of marketing are still absolutely critical.” said Ginny.

This shift means that while AI can optimize bids, generate ads, and analyze data—but it can’t define success or understand your market. Core marketing questions like audience, value, and differentiation still require human judgment.

The conversation also shows that better AI output depends on better human input, making strategic, clear guidance a critical new skill.

“Still to do that, you need to know have the fundamentals - have the right prompts for those kinds of to get the output that you need. So if you’re a more junior person coming into digital marketing, get your fundamentals right, think like a marketer.” Ginny explains.

Episode Transcript

Fred: Hey, my name is Fred Vallaeys. I’m here in studio with Ginny Marvin, also known as Ads Liaison. Ginny goes way back in the industry, so it’s really good to have her here in studio, especially now that we’ve just seen all of the updates from Google Marketing Live. Ginny, thank you for being here.

Ginny: Very excited to be here, Fred.

Fred: Ginny, people know you as Ads Liaison, but one of the things that’s really cool about you is you’re not just a mouthpiece for Google. You’ve actually done a lot of PPC and digital marketing. For the people who only know you as Ads Liaison, give them a little bit of background.

Ginny: I’ve been a digital marketer and in digital advertising for two decades now. I’ve been agency, in-house, consulting, and was also the paid media reporter at Search Engine Land for seven years right as Enhanced Campaigns came out - for a flashback - and editor-in-chief there the last couple of years. I decided I was ready for a break, something new, and then Google called with the idea of this Ads Liaison role, which is meant to be a bridge between our advertiser communities and our internal ads teams to help bring feedback in and knowledge out, help answer questions, and help people hopefully use Google Ads tools better and more successfully.

Fred: That makes sense. It’s a great role and thanks for all you’re doing. Since you brought it up - Enhanced Campaigns - let’s go there for a minute. What did Google learn from that experience?

Ginny: That’s such a great question. For context for those who were not around during that time, it was Google’s response to the shift in consumer behavior to mobile engagement, and marketers were lagging behind consumer behavior. The move then was to have everyone completely recreate their entire search account structure to be able to reach mobile, desktop, tablet all in one campaign. It was needed to get marketers to go mobile and understand that consumer behavior.

I understood the disruption that Enhanced Campaigns caused in terms of all the work and mechanics and testing that needed to go into it - as we all remember, it was a lot. I think Google - I obviously wasn’t at Google then, but my guess from what we can see now is that the way new campaigns and new tooling is being introduced now is aimed at being less “strip it all down and rebuild it.”

Fred: Make the transition easier. It’s an interesting historical anecdote because we’re at that juncture again. It’s not mobile this time. People joke that it was “the year of mobile” every year for six years in a row until it was really the year of mobile. People were very resistant in the beginning because they like doing things the way they’ve done them before.

Ginny: To be fair, it was real when you looked at your conversion reports and campaign reporting and saw desktop is what’s converting. It’s not converting on mobile. So why should I be spending time and effort and money on mobile? I can understand that, but it’s that lag of - are you not converting on mobile because your site is not optimized for mobile? Are users starting on mobile and converting on desktop, but that mobile piece plays a really integral role in the journey?

It’s the same way where we are now. Consumers are really leading the way in terms of what their behavior is, but it’s harder for marketers to see what that impact is on their bottom line and to start adjusting accordingly to be able to capture that new demand, new ways of searching or scrolling.

Fred: To be very explicit here, what we’re talking about is the invention of the chatbots and the large language models and how that’s fundamentally causing a shift in how search is done.

Ginny: And also things like Lens and Circle to Search on Android - multimodal search is a real thing and growing. What do queries look like in that context? Much more complex. How are you adapting to be able to capture new types of ways of searching and querying and engaging with search, YouTube?

Fred: As advertisers, we’ve been conditioned that the keyword is the main thing that carries the intent of the consumer. And not only keywords, but very short keyword phrases.

Ginny: We’ve already seen that evolution and being able to capture different types of queries, longer queries. We’ve gone from the very short two to three keyword phrases - super easy to target those with two to three keywords in your account. But as queries change and evolve, that’s where you’ve seen broad match aimed at capturing those broader search queries.

Fred: I know some people play a drinking game - they take a shot whenever Google says a word, or the more fun one is people do a burpee. Can we all do a burpee every time we hear the stat about 10% of searches are new? And it’s 15%, right?

Ginny: It’s 15%. It’s still the case. This also speaks to the fact that these queries are getting longer, more unique because people no longer have to express it in two or three words.

Fred: I think the 15% query can feel repetitive, but if you take a step back and think about the fact that 15% of new queries are still new every day, and it’s literally been like that for a decade - that’s pretty wild. I’m surprised it’s actually not bigger now in the day of generative AI.

Ginny: I think we’re headed in two directions - either seeing those very long conversational queries, and I don’t know if we still call them queries or we start to call them prompts. There’s also AI Overviews in Google which most people have seen at this point, but now there’s also AI mode, which is another shift more towards that generative conversation. It’s not even just like “here’s my long rambling prompt to get something,” but it starts with that and then it goes deeper and you start asking follow-up questions. How do you turn that back into a two or three word keyword? You really can’t.

Fred: One of the things that the teams are really looking at is how to infer intent within those conversations. That will necessarily change the landscape of how we think about our traditional way of targeting intent, which has been anchored in a query and in a keyword, and that will necessarily need to evolve again along with consumer behavior.

For advertisers who hear that, obviously take a look at what happened in mobile. Get on board with that change. But should we be worried that Google’s basically going to get rid of keywords? Are we going to get rid of campaign structures and overnight we’re going to have this new thing that works in the new world and everything we’ve done for the past 20 years is gone?

Ginny: No. AI Max for Search campaigns is aimed at helping address this new landscape of query methodology and modalities. This is a really good example of how Google is listening to customer feedback about migrations and evolving their campaign structure.

For those who are not aware, AI Max for Search campaigns is new - it’s in beta now, rolling out starting soon, or maybe yesterday depending when you watch this. It has essentially a campaign setting in your search campaigns to one-click toggle on to enable a suite of AI-powered features. It’s text customization, which used to be called Automatically Created Assets, query expansion, and final URL expansion as well. Within that suite, there’s customization to enable or disable those particular pieces, and then there are additional controls.

There’s geo-targeting at the ad group level, which is interesting for a number of use cases. For local businesses to be able to reach people both based on their geo-location and their geo-intent at the ad group level, but also thinking about keyword lists signal. You can put the geo-location at the ad group level even if you don’t have the geo in your keywords, giving that signal for the keywordless function. You can do the same with brand. Brand inclusion and exclusions are also supported.

It’s also compatible with campaign experiments, so you can enable this in a campaign experiment and test and see what the uplift is. On the reporting side, you’re going to be able to see the search terms queries that come from search, and you’ll also be able to segment and look at the assets that were generated and the landing pages that were used as well from AI Max.

Fred: Asset reporting is being expanded. Search terms seem to be coming to more places like Performance Max.

Ginny: Yes, a lot of parallel efforts. The asset level reporting and more metrics are coming to Performance Max search. You’ll be able to see your asset metrics for your RSAs and that includes cost metrics, and also for your responsive display ads in display campaigns.

Fred: What do you hope people will do with AI Max?

Ginny: I think it’s great how we’re launching it. There’s that first level top-level control - turn it on and off - and then you get all the details below it. The important part is you enable it within your existing search campaign. Unlike Enhanced Campaigns, there is no need to rebuild or restructure. I think you can start to learn from what you’re getting, and there’s always opportunities probably to do some account consolidation and campaign consolidation if you’re not thinking about that already. Apart from that, you can start just turning on AI Max and testing it within your existing campaign structures.

Fred: Let’s go on a little tangent here. Campaign consolidation - we had a really interesting conversation yesterday with some of the product managers. Many of these were under NDA, we can’t really talk about it, but here’s one that we said we really need to talk about. That was to do with conversion events - primary, secondary, what goes into bidding. I think that’s one of the fundamental things that people think about when it comes to campaign consolidation. How do you get more conversions grouped together so that the system has enough to learn from?

But then we’re talking about account level, campaign level, MCC level conversions. What’s the guidance on that to get the right information into the learning system to do the bidding correctly, but still allowing you to do campaign structures that correlate to your business goals?

Ginny: I think certainly you want to lead with your business goals. What matters to your business should inform your structure. That would be if you have campaigns that necessitate different ROAS targets, for example - that will dictate a structure. But if you look across your account now and have five, ten campaigns that essentially have the same goal, that could be an opportunity to consolidate under that one goal.

Fred: So you consolidate those campaigns, then they can have a single conversion event associated to that campaign, or you can set it at the account level?

Ginny: Yeah. But I think there are obviously other reasons to segment campaigns, but if you are looking at it from a goal-based standpoint, I think that’s the first lowest hanging fruit of being able to think about consolidation.

Fred: A lot of this consolidation would happen because you want to group things together, consolidate the data, the conversion data. But what about an SMB? You don’t have that many conversions. You don’t have enough primary conversions to even drive the bid management system to where it needs to be.

Ginny: I think that’s where you want to be looking at what’s your next lowest conversion event that is still giving you the signal about what’s quality lead, probably. We’re not talking about a sale, probably talking about lead gen. What’s the next lowest conversion event that gets the volume and signal of quality for you to be able to track, measure, and feed back into Google?

Fred: If I remember correctly from the conversation, there are types of conversion events. You need to set that correctly and not just pick one off the shelf. It actually does matter. And then you have primary and secondary. Neither one of these is necessarily going to be used for bidding, but it tells Google basically what’s a macro conversion and what’s a micro conversion. So what’s one of these intermediate steps? That’s good signal, but it’s not the thing we want them to do. That should be your secondary.

Ginny: And then there’s a way that you can say this primary is also the one that I want to use towards bidding automation.

Fred: What I thought was fascinating was that this is complex and there’s a lot of nuance to this, but set those secondary ones because Google wants to analyze that if there is lack of enough primary signal. So the way that you set this up actually will impact your bid management performance.

Ginny: It’s just another example of the more information that you can give Google about what matters to your business, the more the system will understand and be able to optimize accordingly.

Fred: I’ll give you my take on it too because I think initially these things might be built and the Google engineers are like, “Yeah, wouldn’t it be great if we had 15 levels of information?” and then they ask the advertiser to put it in, but initially they don’t use it. Maybe they only use two or three. But over time as machine learning gets better, as they have the ability to build more of these features in, they actually start looking at it. So my point would be don’t get discouraged by initial lack of success from that. Do take care to set it up the way Google intended because at some point that is going to come to fruition.

Ginny: And I think also a good takeaway from that conversation yesterday was Google can do a better job of educating on that. So stay tuned.

Fred: That’s why we’re here. So, you kind of get into AI Max, Performance Max. It was sort of funny because Google at GML last year it was the “power pair” - broad match with smart bidding. That was thrown out yesterday and now it’s the “power pack.” Tell us what’s in that.

Ginny: Last year the power pair we talked about search and Performance Max. Now the power pack that we’re talking about is Performance Max, Demand Gen, and Search with AI Max enabled. Essentially taking the best of the AI enablement across these AI-powered campaign types and using them to drive performance.

Fred: Under search it was the AI Max toggle which gives the AI capabilities. I just want to get ahead of concerns about what about Display and standard Shopping and just say there are no plans to deprecate those, and you’ve seen that we are continuing to actually add support for new features going into those campaign types too.

Ginny: Yes, build a Performance Max campaign, but still maintain a Display campaign, a Search campaign, a Shopping campaign as maybe your primary. I think it depends on your business and what you’re trying to accomplish with those campaign types. I think the focus is really there’s optionality. You have the options to choose the set of campaign types that make sense for your business. Many people are using both Smart Shopping and Performance Max, or Display and Demand Gen. Mix and match. I think obviously the search with AI Max, Performance Max, and Demand Gen are all built with the AI at its core and with the intention of being able to adapt, use the best in technology, and also adapt to where changing contexts and consumer behavior.

Fred: One of the frustrations with Performance Max has been sure it has good results, but you’re just spending all my money on remarketing or you’re spending it on the brand on search, which is easy, and you’re cannibalizing what I already had. What’s Google doing about that?

Ginny: The channel reporting that was announced earlier this month and is in beta will be rolling out to everyone over the coming month, maybe two months. That is specifically due to customer feedback. We are seeing results but where is it coming from? And giving advertisers the insights into what’s performing and what is not, how channels are playing a role within their conversion journeys that they’re seeing, and diagnostics for potential ways to improve performance.

There are a number of different layers within channel performance reporting. There’s the Sankey visualization that you can interact with to understand how each channel is contributing to your performance. Then there’s a more performance table that you’re able to dive more deeply into the metrics. You can download that data, use it in other ways for analysis. And then the diagnostics piece to be able to see specific opportunities for improvement.

Fred: We asked for questions. We have Heidi Sturk asking this is all great. She would love to learn more about the visibility into performance reporting regarding ads shown in AI Overviews and AI mode. Makes me think a little bit about the whole channel reporting in Performance Max and whether AI mode would be one of the things we can actually break out.

Ginny: AI mode and AI Overviews will be grouped within search. There won’t be separate segmentation, but there will be insights in the search terms report.

Fred: Let’s talk for a minute about those search queries because they’re conversational. I think keywords are still limited to 10 words or did that go away?

Ginny: I don’t think it went away. I think that’s still the thing.

Fred: Even if you wanted to, you could not put in that 100-word keyword, nor would you really want to. You’d still hit the low volume. So you have that issue, but what Google is doing, it wasn’t talked about much, but it’s a synthetic keyword. Can you explain what that is and how that fits into this whole longer query prompting knowing when something is AI generated?

Ginny: I think there’s work being done on that. I don’t have a ton to share on that right now. But that’s something that’s obviously being looked at in terms of how can we help marketers infer and understand how they’re showing up in those types of modes.

Fred: Since you can’t say, I’m going to speculate here and maybe make some assumptions and you can set me straight or have no comment. I equate it to what we saw very early on Microsoft and Bing search results. You’d be interacting with the co-pilot and it would literally say next to the co-pilot box, “now searching for blah blah blah.” And that was a distillation of what they thought that longer prompt was about. To me that sounds like it’s the synthetic keyword because at the end of the day the Google system is still built around keywords. Even if it’s a broad match, it has to go back to a foundational keyword to be able to pull the ads into the auction. That’s where my assumption, but you probably see these long prompts being turned into what’s called a synthetic keyword that corresponds and that’s what goes into the auction.

Another question about Performance Max from Mariano Bilski. Are we going to get the same channel selection features for Performance Max as we’ve got for Demand Gen?

Ginny: I’ve gotten this question a lot and understand the why. Performance Max is designed specifically to reach across all channels. Demand Gen is where you can have the optionality to select the channels that you want.

Fred: Speaking of Demand Gen, Google came out with the four S’s yesterday. Do you remember what those are?

Ginny: Streaming, shopping, scrolling, searching. Did I say that right?

Fred: That’s right. Yeah. I don’t think that was the right order.

Ginny: I think it starts with search. Search comes first.

Fred: I mean, it is Google. Search comes first. I thought that was interesting too because that also speaks to why you would want to have a Demand Gen campaign because certainly scrolling is part of that experience that consumers want. YouTube very big, so streaming matters. You’ll see more like the Display campaign is now an option in Demand Gen, and I think you’ll see some expansion on channel opportunities in Demand Gen. But again, it’s really about that campaign type making the most of your visual assets - image and video - and getting driving mid to lower conversion funnel opportunities there.

Speaking of creatives and AI generated creatives, we saw some pretty cool stuff at GML, but a question I hear from a lot of people - here Gabriel Benedetti was asking it, our friend from Italy. To optimize images with AI is great, but if you work in e-commerce, you have a lot of them. Obviously have a lot of products. Is it coming to work with images and AI at scale?

Ginny: In Merchant Center, Generated for You is a new option that’ll come out that will automatically generate images at scale based on your catalog, and you’ll be able to then review, save, and publish those across Google in one click.

Fred: That’s scalability. That’s a big shift from last year when I think it was like a one ad group at a time or one product at a time. But now we’re even thinking about it’s not just generate me an image but generate me the video that goes along with it. So we have it everywhere. What’s Google doing in that regard?

Ginny: It’s integrating Veo into Asset Studio and in Merchant Center. So you’ll be able to generate videos from your assets and publish them and use them.

Fred: Do we have to pay for tokens?

Ginny: No. When they’re built into your ads environment, you’re able to use them and use them in your ads.

Fred: A little anecdote on that. There were some demo stations at GML. We saw a demo where you take a product image and then put it in a context that you can describe. What was interesting is that the product manager on that said a lot of time had been spent developing that model. So it is built on top of what we all have access to, but how do you make sure that the product is not changed, that you maintain the logo in the place where it is on the pants or the headphones? That is really one of the key reasons to look at this Google technology because someone’s already done the hard work of making sure the product is not changed, as opposed to doing it through some other system.

Ginny: We also announced yesterday that the image asset generation is now using Imagen 4, which is the latest model of the image asset generator. The other thing I wanted to mention - we were talking about Merchant Center - is the ability to A/B test product generated product images or two different product images. So A/B testing for product images in Merchant Center.

Fred: A lot of people were very excited about that. Let’s go back to some AI Max questions because that is part of the power pack now. We had Craig Belcher asking he’s played with AI mode since Google I/O the other day. He knows it’s still being rolled out. You see fewer ads along with that. Will the ads be labeled so that the consumer knows, as well as for advertisers who might be doing their own research? What is an ad? What is a rich snippet link? Is there going to be an intertwining layering through the rich snippets on both the AI mode results as well as the snippet references on the right?

Ginny: I think the question is will ads still be labeled as sponsored? Yes. Ads will still be labeled. We still have to meet legal requirements and user expectations.

Fred: That’s a good point. The reason ads on Google have done so well is that Google saw ads as information and not as intrusive things. With AI, sure, we still need to know what an ad is, but if it’s a helpful experience, ultimately the consumer probably doesn’t care if they’ve been directed to where they get the benefit from it.

Ginny: I think relevancy and helpfulness is key underpinning of the entire Google ads system. One of the really interesting things with AI Overviews and AI mode is being able to predict and understand intent even from what would traditionally be seen as informational queries. That’s where you see the examples of us showing product ads in a higher level research query because we can understand that you’re probably going to be in market for a related product or service to what you’re searching for.

Fred: I think this is going to be a challenge for advertisers because if you look at a singular query, synthetic keyword, single prompt, you may say, “Well, that looks like an awful place to show my ad.” But you don’t know what was before that. You don’t know the context.

Ginny: This is so important because it really is a mindset shift of going beyond the “I’m trying to find this bottom end of the customer journey query.” For the last 20 years, I’ve known exactly what that query is. Now that the journey is evolving and it’s more complex - I know we say that a lot but it is - and as AI is able to understand more and ingest more signals about what the consumer is doing, looking for, and can serve ads in those moments that otherwise previously would have been not felt helpful or relevant. So yes, for marketers that’s a mindset shift. When you’re looking at those query reports and you’re going line by line thinking “this isn’t relevant, this isn’t relevant,” taking a step back and understanding that there’s likely more at play for the reason why that ad got matched to that query.

Fred: There’s interactions, there’s nuance in the journey, there’s things you might not know. Now, you as the advertiser don’t know, but the AI often makes guesses too. That’s what it does. It makes predictions with a confidence level and sometimes that confidence level is not very high. One thing that was announced yesterday at GML was smart bidding exploration. I personally thought it was really fascinating because they position it as the biggest improvement in automated bidding technology in more than a decade. But at the same time, people look at it and it’s basically a toggle. You can say for my ROAS bid, you can make it 10 to 30% more aggressive to help me get more conversions. Why is that different from having new broad match keywords, broader matching, or how’s that different from just raising your bid?

Ginny: Great question. This is one of those things that seems minor but on the back end is a really big shift and move forward. Search smart bidding exploration is, as you mentioned, for ROAS if you’re using ROAS targets, lead gen, or online sales goals and broad match, DSA, AI Max. Turning this on, you’re essentially giving the system added flexibility to find more conversions within your existing reach in your existing target. You’re not changing - it’s not changing how broad match is working. It is allowing the system to go after a broader set of queries that you might be missing out on bidding for otherwise because your campaign ROAS target will tend to find the conversions from the bulk of queries that you’re matching to. What smart bidding exploration does is it gives it more flexibility to go after a broader diversity of search categories with your existing targeting.

Fred: So it’s going after more fish in the pond that you’re already fishing in. The query diversity is important because you’re not putting all your eggs in one basket in a sense.

Ginny: You’re really not changing your targeting. But the flexibility will allow you to bid more competitively on that set of queries that you could already match for but are missing out on.

Fred: Let me try to give a technical explanation on that. You have a universe of queries for which your ad would be eligible based on your targeting, your broad match, your DSA, your AI Max. Now you’ve set a ROAS target and that ROAS target means that Google’s AI makes a prediction about what is the likelihood of the conversion at the price that we have to pay to get that. Now some of these things it doesn’t have high confidence that it’s making the right decision and it says “well I’m not sure but I think we can’t afford to bid on this thing. It’s just not going to convert at the ROAS that you need.” So it doesn’t show your ad.

Now what can you do as an advertiser? You can say I’m going to raise my ROAS targets or I’m going to lower them so my bids become more aggressive so that I actually have a chance to get those searches. But in doing that you’re raising your bids across everything, even the high confidence things where Google already knew exactly how it’s going to convert. Now you’re also getting that lower return on ad spend and that’s not desirable. You want the stuff Google knows that you already know - keep it at the ROAS you want. The stuff where Google is not certain, that’s where you’re saying “okay you can have 10-30% more to test it out and build a higher confidence level” to say that either yeah it really doesn’t convert at the rate that we thought it needs to so we’ll bring it back and we’ll get rid of it, or they test it out and now you get good results, it’s got higher confidence of what it needs to bid, and that becomes part of your query diversity and you can actually start showing on it.

Ginny: Often time it’ll be the lower volume queries that there might not have been enough data. That’s where it’s also looking at those categories. You’ll be able to see the performance in the bid strategy report and dig in on the unique search categories and it’ll be able to give you a sense of what that expansion is looking like.

Fred: Or exploration, I should say. I’m excited to see what that one is. I recommend people definitely do try that out and give feedback on it.

Ginny: That’s another one that’s also compatible with campaign experiments.

Fred: Use campaign experiments. All right. I think we had a couple more AI Max questions. Natalya Lopez was asking, “Exciting. I would love to know a bit more about AI Max and what the expected best practices are. For example, does it work well with all types of keyword matches or does the campaign need to have broad match?” We’ve kind of answered that, but what about non-broad match? Would it actually do anything with phrase match?

Ginny: No. It’s broad match. It has to have broad match AI Max or DSAs. In terms of best practices, the same with smart bidding exploration - you need to be using ROAS as your bid. You need to have ROAS targets. You need to be using broad match. At least some broad match. Like half of your campaign or so should be broad match or DSA or AI Max for search campaigns. Your budget needs to be unrestricted because you need to be able to give that system flexibility. If you’re using a capped budget, which isn’t recommended anyway if you’re using ROAS or CPA targets because you’re really using those targets to help - that’s your spend management lever. So you’ll see an error or a warning if you try to enable smart bidding exploration with a capped budget.

Fred: Another question on AI Max. Sophia was asking about experiments, but you’ve covered that so you can turn this into an experiment. But she also said, “Is there a specific number of conversions needed for AI Max to work best?”

Ginny: No, but again, I think that goes back to our conversation about looking at your campaign structure and as much consolidation as makes sense for your business to help fuel the AI learning is the best. It’s still the general best practice. I mean we’ve run studies at Optmyzr and we see 30-50 conversions is when things really start rolling really well. But this is not introducing anything different from how that’s worked in the past.

Fred: All right, we’ve covered a lot of ground, but we haven’t talked about measurement, which is exciting. We got “the data guy” - don’t have the data girl t-shirt which I hear they have as well - but Gorav announced a number of exciting things. The ones that I picked up on was the data manager standardization of the alphabet soup. When they said alphabet soup I was like wait, Google’s Alphabet is now making soup, but he literally meant the jumble of acronyms that all have their own API - OCI, ECLID, G-Cloud, Gbraid, all of that stuff seems to be getting standardized into a new data manager API that should be easier to use. And then incrementality testing. I’d love to talk about those.

Ginny: On data manager, I think this is something that is so helpful for businesses of all sizes to be able to more easily bring their first-party data into Google Ads. It’s right in Google Ads. A lot of integrations that just make it so much easier to centralize your data, bring that data into your ads account to inform measurement and bidding.

Fred: And then let’s prove that all of these new things that Google is doing work. So the incrementality testing.

Ginny: Think brand lift, search lift studies that have typically taken a long time and significant budget. We have introduced Bayesian modeling to our incrementality testing. So it allows us to understand and find statistical probability on your incrementality, search lift, brand lift. For example, we’ve lowered that minimum experiment budget to $5,000. So that’s $5,000 for the length of the experiment.

Fred: Can it really work at $5,000?

Ginny: Yeah, it’s great. Try it out. In faster time too. I think as short as a week to 28 days.

Fred: I think they were saying minimum 7 days, maximum 56, ideal at 28 days. Multiples of a week. There was another interesting stat that close to half of marketers don’t use mixed media attribution and incrementality testing. So Google is really trying to make it easier to make these things accessible because it makes sense. Even with Meridian it’s not easy to set up.

Ginny: I think that’s an area that’ll be interesting to see. The thing with Meridian that makes it - is open source. So if you have a current mixed media model that you’re using, you can choose the pieces of Meridian that you want to integrate into your own system or you can build your own mix model from Meridian as well.

Fred: I think there’s a number of partners that Google has that will help with that. As far as the data manager API, what I’m excited about is that as the measurement landscape evolves due to regulations and just new technologies coming along, the vision is that you integrate with a single endpoint and you start layering more information through that same endpoint. But it’s not like “hey we got a new technology, you got a new API that you need to connect into.” Because bigger organizations, this is not easy to put a new endpoint in.

Ginny: I know. I was talking to somebody yesterday and having flashbacks of in-house experiences of integrations and the time it can take. This is all aimed at trying to make that entire process simpler, easier, faster.

Fred: I would say that one really matters for lead gen businesses because sometimes you go to GML and lead gen is like, “Oh, you didn’t have anything cool for us.” The shiny stuff, the videos, the images, that’s all e-commerce. But that’s just how e-commerce works. I think there was exciting stuff for lead gen. But you also have to do the fundamental things. When we often look at how many people still don’t communicate conversion value or quality in any way - do that first and then start worrying about the shiny stuff.

Ginny: I think it’s so important. That measurement foundation and data foundation when you’re in lead gen - going back to the idea of what do you know about your business? What are the conversions and the customers that are most valuable to you and giving Google that information to be able to optimize and find conversions that are going to be valuable for your business? That is such a key foundation to make having successful lead gen campaigns.

Fred: What about small businesses? What was announced at GML? Do you think that matters to them?

Ginny: There were a couple on the agentic side that I think are going to be really interesting for businesses and agencies of all sizes, but especially for small businesses. Often times you’ve got one person or the business owner themselves managing their campaigns. The conversational experience in Google Ads - being able to use prompts to set up your campaigns - that’s getting more agentic and then the marketing…

Fred: Before we go deeper here, what does agentic mean to you?

Ginny: It’s the idea of having the systems be able to do tasks for you. That can mean within one system or, as we’re going to start seeing more, being able to do multi-task across different entities and basically get tasks done that you ask it to.

Fred: These tasks are not necessarily limited to things that have API connections. They could literally be there’s a bot that goes to a website, clicks buttons. It could be a bot that makes a phone call to a human to do something. Eventually it’ll be robots that walk to the store to pick something up.

Ginny: We saw at GML being able to tell Google, “Find me this. I like this dress. I tried it on virtually. Looks great. Let me know when it hits this price and then go buy it for me.” So, it’s also about constant monitoring. It’s not about you initiating the task. It’s just doing it for you constantly. You’ve got a little agent in the background that you’ve instructed to do certain tasks.

Fred: This then leads back to another conversation we had with product management at Google yesterday, which was how does the agent know that the price has changed because nobody can check the price every second. That just uses too many resources. So they reminded us it’s about your feeds. Make sure we know when your feed is changed. Make sure we get reliable data from you, the person selling that dress, so that our agents can be pinged and hey, the dress is now at a different price. Go check - is it the price they wanted and then continue.

Ginny: I think that also just speaks more broadly to data freshness - that your website being updated, your feed being updated and accurate, as you’re sharing your conversion data, your customer match lists, all those data sources, just being as fresh as possible.

Fred: For those who didn’t watch it, the demo by Saline Song at GML was great. It was her speaking to an agent. The agent figured out the tracking pixel wasn’t installed. That’s why there were no conversions in the campaign. And then she said, “Can you go and install it?” And it was like, “Yeah, it looks like you’re using Wix. I’m going to go grab the code from Google. I’m going to log on to Wix. I’m going to install it. I’m going to check it.” Did all of that. Done. That’s a pretty cool future. You know, me at Optmyzr, I got my work cut out for me. But we’ll be building in that space as well.

Ginny: I think it’s going to open up a lot of opportunities for so many people. I know we were having this conversation yesterday, but the idea of what does this mean for our roles? What does it mean for marketers? I really just see the fundamentals of marketing are just becoming more and more important. They’re not - the idea of you need to reach an audience. You know who your audience is. You can know what your brand is, what’s going to make your brand stand out. Those pieces of marketing are still absolutely critical.

Fred: I agree. Knowing the right promotion. I think this is also a lot what these signals that Google is giving are about. So, what are the search themes? What are the channels that your Performance Max is running on? It’s not about saying I’m going to exclude or include or just bid differently, but it’s also about why isn’t my offer resonating with this segment? Is it something about the way I’m positioning it? Is it maybe the price? All of those things you can put your strategist hat on and rather than pushing buttons, you tell your agent, “Hey, go and fix this or go and try it out.”

There’s a beautiful video too in the GML where a more junior employee was being talked to by four different more senior people and they were like, “Do this, do that, do that.” And I was like, “Oh my god, in the normal world she could never do that,” but now she handled it in an hour and actually had great results. I know that was a demo video that wasn’t real, but that’s the future vision.

Ginny: I think so. And still to do that, you need to know have the fundamentals - have the right prompts for those kinds of to get the output that you need. So if you’re a more junior person coming into digital marketing, get your fundamentals right, think like a marketer.

Fred: Is there anything else you would advise?

Ginny: I think part of that also is understanding your data and understanding what matters to your business. What are the metrics that are going to matter to your business? Understanding the role that marketing is playing within the broader organization to drive growth or profit is really important.

Fred: Andrew Loll also came into studio before and one of the things he talked about was that after COVID nobody was doing in-person client meetings anymore and they said we need to get back to that and then giving examples of you work or you walk the factory floor or the delivery floor and you just start picking up on these insights about the business that you can take back to marketing and actually do something with.

Ginny: I think we don’t talk enough about - talk to all your cross-functional teams. That’s sales - they have obviously usually your marketing sales teams are tightly integrated but talking to individual sales reps. What are you hearing? What are the kind of conversations you’re having? What are the objections that you’re hearing from customers? Those are just such valuable nuggets for marketing fodder.

Fred: And they’re valuable nuggets, but we’re all too busy to get them. So that’s where you look at Gemini, you look at Vertex, you start building systems, you pick a tool that already exists that listens to your sales call meetings, that listens to your customer support meetings and summarizes it, says these are the most important trends. This is where we’re seeing a shift, things that people now talk about they didn’t before. That really informs the direction of a product company like mine, but any company can really benefit from that.

Ginny: Being able to take those insights at scale. I think that’s great and I do still also think those individual conversations can be super helpful as well.

Fred: We’re talking about people talking to people. It’s also often said people buy from people. So where do influencers fit into all of this?

Ginny: Great question. We have not talked about the Creator Partnership Hub that was also has will be rolling out and when it is available in your account it’ll be under tools. The whole point of this is to make it easier for businesses of all sizes to be able to find and partner with YouTube creators and influencers.

Fred: That’s cool because the other points I was having conversations to people about yesterday but it’s about authenticity and as so much becomes not necessarily inauthentic because it’s generated by AI but there is a level of - if you couldn’t bother to write it why should I bother to read it - whereas if it’s an actual human influencer who’s spending some time making a video using the product talking about why it matters to them there’s still something there that connects at a more authentic level for humans so it’s certainly going to become more important.

Ginny: I think so too. Obviously, creators have the authenticity, they have the trust, they have reach within their audience. In that hub, you’ll be able to basically search for - you can search for videos that may already mention your brand from creators and reach out to partnership with them to boost those. And you can also find creators relevant to your business, keywords or genres and then look at their reach, their demos, pricing potential information to partner with them. And then once you do reach out to them, partner, link their videos in your account, then you’ll be able to see both paid and organic performance results from those videos.

Fred: Let me take you on one more tangent here if you can. Will Reynolds from Seer and I, we have a massive disagreement on the future of landing pages. I’ve said and you don’t have to pick sides here - but I said landing pages are dead and I really believe that because with agents and chatbots, the consumer is having such a great experience chatting to them, figuring out what they want to buy. Why do we even still need landing pages? What am I missing?

Ginny: I’m not going to pick a side on that. I still think there’s value in a landing page, but I also, to your point, think that there are going to be experiences beyond the landing page.

Fred: And it’s interesting because I do believe landing pages will still have a place for deeper purchases and higher consideration things. But honestly, if I figure out this is the running shoe I need, why do I still need to go and look for the place to buy it? Why do I still need to go through your site to check out? That just makes no sense. It’s not a good user experience.

Ginny: No, I completely agree with you. I think the shopping experience we just saw at GML, that’s a great example of just show me. But it’s Nike. Just do it.

Fred: I think those experiences will continue to evolve and inform what a website looks like. One interesting thing that then came out in conversations as well was, okay, so you talk to your agent, your chatbot. You tell them, I have a pronated foot. I have this shoe size. This is the brand I like. And you get to the point where it’s like, okay, this is the shoe for you. We know what you need. And then it hands you off to maybe a landing page, maybe another site, and now their bot comes in and says, “Fred, tell me about what you need.” I’m like, but I just went through that whole thing. So, how do you pass that on in a privacy sensitive manner? I don’t think anyone has - and I’d love if you have an answer, great. But I don’t think we have answers to that. But I think we’re at a point here again where go along for the ride, find a trusted partner like Google. Don’t assume they have all the answers, but be willing to sort of come along on that new journey as the way the world searches and finds things is shifting. And find those partners who are there with you with the same underlying goals because nobody’s going to tell you how to do it. You have to help us figure it out.

Ginny: And it doesn’t come without points of friction obviously that in some cases that’s going to be a necessity to have points of friction. But yes to your point I think start exploring, start experimenting, start researching, and have an open mind and think about what the consumer experience is, what they’re expecting, and try to meet them where they are which I know is a sort of flat thing to say but I think is what needs to drive your strategies moving forward.

Fred: Well, I think you’ve brought us back to the key point here, which is things are shifting. Try to do the right things. And hopefully I know people will have learned a lot from all of your answers. So, thanks for being in the studio here with us today and sharing all of your expertise and all the work you do as Ads Liaison across all the social media platforms and facilitating the meetings and making sure the product team at Google knows what to do. So, thank you, Ginny.

Ginny: Thank you.

Fred: And then obviously if people want to find out more about you, Ads Liaison on all the social medias.

Ginny: Yeah. Actually I’m Ginny Marvin on LinkedIn and Reddit and then Ads Liaison most everywhere else.

Fred: Great. And with that, thank you all for watching. If you want to see more content like this or next time we bring Ginny in studio, please subscribe to this channel and also put in the comments if there were other questions because I know Ginny’s very active and we’ll take that to heart and try to answer those. With that, we’ll wrap it up here. Thanks for watching and we’ll see you for the next one.