EP 257 - Google Already Knows Everything About You — And Their New AI Features Make It Worse

In this episode of Near Memo, Mike Blumenthal and Greg Sterling explore how Google’s rapidly evolving AI ecosystem is transforming search, advertising, and personalization. They begin with a major DOJ fraud case involving 15,000 fake Google Business Profiles and discuss whether Google’s platform design enables abuse at scale. The conversation then shifts to inferred demographics, with Mike sharing experiments showing how Gemini and ChatGPT can accurately infer income, lifestyle, politics, and buying habits without explicit opt-ins. Finally, they analyze the biggest announcements from Google I/O and Google Marketing Live, including AI Mode, multimodal search, AI-native ads, shopping integrations, and Google’s broader strategy to deepen user lock-in and dominate the AI era.

The Podcast Deets

1. DOJ Fraud Case & Google Business Profile Abuse (01:32 – 10:05)

Mike and Greg unpack a sweeping DOJ complaint against Premium Home Service and its CEO over alleged fake listing and fake review schemes spanning thousands of fraudulent Google Business Profiles. The discussion examines how scammers exploit weaknesses in Google’s verification systems, whether Google’s platform design effectively enables this behavior, and why Section 230 protections may prevent meaningful accountability. The hosts also debate whether Google Search and Maps could someday face “defective product” legal theories due to systemic fraud exposure.

**2. AI Personalization & What Google Already Knows About You (**10:15 – 16:26)

Mike shares an experiment asking Gemini and ChatGPT to infer his demographic, political, professional, and socioeconomic profile using search history, memory, and publicly available information. The results were surprisingly accurate, including estimates of income, lifestyle preferences, and social class. Greg and Mike discuss how inferred personalization works even without explicit “Personal Intelligence” opt-ins and why AI-powered search experiences are becoming increasingly individualized and behaviorally targeted

3. Google I/O, AI Mode & The Future of Search Advertising (16:26 – 34:09)

The hosts break down the major AI announcements from Google I/O and Google Marketing Live, including AI Mode, Gemini Spark, multimodal search, agentic experiences, and AI-native ad formats. They analyze Google’s “good enough AI” strategy, its advantages from owning the full AI stack, and the company’s growing ability to personalize both organic results and advertising. The discussion also explores shopping integrations, AI commerce, Google Maps implications, and whether antitrust efforts have failed to meaningfully constrain Google’s expanding ecosystem power.

Key Takeaways

• Google already infers enormous amounts about users without explicit opt-ins.
• AI personalization is becoming central to both search results and advertising.
• Google’s AI ecosystem strategy increasingly aims to keep users entirely inside Google experiences.
• AI-native advertising formats may unlock full monetization of AI search.
• Google’s ownership of infrastructure, distribution, and data gives it major AI advantages.
• The DOJ fraud case highlights ongoing weaknesses in Google Business Profile verification systems.
• Google Maps and local search are likely to become deeply AI-driven and personalized.
• Antitrust rulings so far appear unlikely to meaningfully slow Google’s AI expansion.
• AI Mode represents a major interface shift toward conversational and multimodal search.
• Agentic AI and transactional search experiences are rapidly approaching mainstream adoption.

👇 Watch by topic:

• 0:00 - Introduction • 1:34 - The DOJ Fraud Case Against Premium Home Service • 10:15 - Inferred Personalization: What Google Knows About You • 17:00 - Google I/O and the AI-Powered Search Box • 25:37 - AI-Native Ad Formats and Future Monetization

Related Links

Department of Justice Files Complaint Against B.E.S.T. GDR, LLC, Doing Business As Premium Home Service, and its CEO Yosef Bernath for Violations of the FTC Act and Related LawsThe Department of Justice, acting on a referral from the Federal Trade Commission (FTC), along with the Attorney General of Illinois and the U.S. Attorney for the Northern District of Illinois, announced today the filing of a civil complaint to stop an Illinois company and its owner from deceiving consumers in connection with the sale of home repair and improvement services inDepartment of Justice Logo Office Of Public Affairs“Premium” Home Service – The questionable businesses of Yosef Shmuel Bernath"Premium" Home ServiceEP 255 - How Google’s Personal Intelligence is Quietly Revolutionizing (Everyone’s) Search ResultsIn this episode we are joined by Garrett Sussman (iPullRank) to discuss his provocative 12-month study on AI personalization. We dive deep into how Google’s “AI Mode” uses your Gmail, Photos, and Calendar to tailor results—and why “unopened emails” might be influencing what you see next.Near MediaNear MediaEP 256 P2 - The Death of Keywords: How Google’s Personal Intelligence is defining the future of Local SearchIn Part 2, Garrett Sussman joins Near Media to explore how Google’s AI personalization could transform local search, reviews, SEO, attribution, and marketing strategy — and what businesses should start doing now before AI-driven discovery becomes dominant.Near MediaNear MediaGoogle I/O 2026: All the news and biggest announcements of the year for Gemini AI, Android XR smart glasses and moreDon’t miss all of the updates coming to Google Gemini, Search, Workplace and almost every other application the company makes.YahooValentina PalladinoGoogle Marketing Live 2026: The Biggest Shift in YearsGoogle Marketing Live 2026 announced major updates to search, ads, creative and measurement. Here’s what’s changing and what it means for your business.Blayzer Digitalblayzerdigital

Interested in sponsoring this podcast or our newsletters please reach out to mblumenthal@nearmedia.co

E-mail Mike

Near Memo Transcript:

Greg (00:10) Hey, everybody. Welcome back to the Near Memo Episode 257 remarkably.

And I'm here as always with the lovely and talented Mike Blumenthal.

Just as today, we will have more guests in the future. Today, we have a packed episode.

We're going to be talking about a blockbuster fraud, review fraud, listings fraud case that Mike has been kind of tracking and is deep into.

We're going to be talking about Google's inferred personalization. We talked with Garrett Sussman a couple of weeks ago or last week, depending on when this showed up, about personalization in a two-part episode.

It's really excellent. You should listen to it if you haven't.

And there's two types of personalization, explicit, where you opt in and you give Google access to much of your data, and inferred personalization or implicit personalization.

And Mike is going to talk about that. Google already knows that.

ton about us, it doesn't really need us to opt into capturing more of our data.

And then we're going to talk about I.O., Google I.O., and some of the search related announcements that came out of there.

We're also going to be talking about at Google Marketing Live, they announced new AI native ad formats, which I think is very, very significant and has implications for the search experience, which we'll discuss.

So Mike, why don't you kick us off with the fraud

Mike B (01:32)

Sure.

Greg (01:33)

fraud issue.

Mike B (01:34)

So last week, the Department of Justice, which is the legal form of legal arm of the government filed a complaint against BEST doing business as premium home service and its CEO, Joseph Bernth, for violations of the FTC Act.

And one presumes that the move from reporting to the FTC, their determination of repeated violations that reached a level of legal standing.

So it got referred to the DOJ, which means that it's a criminal case at this point, accused of creating 50... Well, it's throughout the whole Midwest and we don't know how broadly it reaches, but he created 15,000 fake business profiles with fake reviews over time that use bricks and mortar addresses belonging to unrelated businesses and local phone numbers that were outed to oversee call centers.

The standard...

Greg (02:07)

And this is in Minnesota also, I think.

Mike B (02:28)

You know, MO of these guys and the defendants posted, you know, fake reviews, which appraised their fake business profiles.

And the defendants misappropriated the images of people from the staff of directors of unrelated websites, all the standard stuff.

including, you know, signing people up for subscriptions and, you know, claiming they do the work themselves and then outsourcing it by this sort of, you know, there's a whole black.

market of subcontractors, often Israelis, you working in certain forums on Facebook and elsewhere, where a business like this can go and say, who wants this job, right?

And typically that subcontractor.

Greg (03:08)

Well, is that the primary model here, lead gen for outsourcing to subcontractors?

Mike B (03:11)

That's what he's, this guy is doing. And also though, in addition to a subscription model that, you know, wasn't fulfilled.

they estimated 15,000 fake businesses and $79 million in illicit gains.

Greg (03:29)

Which is which is kind of astounding that there was that much money generated what what what was the duration of the of this I mean did it did it start years ago or is it a fairly recent thing.

Mike B (03:40)

That's the interesting thing to me. it best I can figure it started in 2022. There is a website called premium home service dot info, which is a critique site that details abuses as far back as 2022, 2023. hit mainstream, uh, regional press in 2024 and 2025 in Minnesota and got fairly broad coverage at that time.

Um, and Finally, the AG there, I think in Minnesota, sued these guys.

And I went and looked at the filing and interestingly, some of the businesses mentioned in the filing were still live on Google, which speaks to this whole issue.

This has been going on for four years. don't know. Four years.

I don't know how much Google knew or didn't know, but given the detail on the website that critiques this business.

Greg (04:25)

for years.

Mike B (04:34)

One presumes that it was reported extensively to Google.

Greg (04:34)

Well, what?

At least two years ago, it seems like Google was on notice.

Mike B (04:41)

Yes.

They were on notice because two years ago, it hit mainstream media in Minneapolis.

So for sure, at least two years ago, although I assume it's longer.

Yet only recently Google contends via the DOJ that they are in fact taking these down.

But when I went to look, I still found them.

and in none of the press coverage is Google criticized, which I find fascinating because in effect, even though they are protected under section 230 against abuses of their product, they're not protected against defective product that by design allows this behavior, which is where I think they are.

Greg (05:21)

Well, Google, other than the YouTube social media context, Google has not been hit with a defective product lawsuit yet.

Mike B (05:30)

Right, and so it's a novel theory that hasn't been tested.

Greg (05:34)

Well, you could argue that rampant fraud like this, unpoliced by Google, creates a defective product.

That's what you're effectively saying.

Mike B (05:43)

I'm saying that by design, their product allows this to happen.

They assume that they will catch it before it causes a serious problem and they don't.

And even though there's no incentive.

Greg (05:53)

Well, because there's no vetting on the front end.

Mike B (05:59)

There's not enough vetting on the front end.

They make it really easy on the front end to create a fake listing.

This guy figured out how to create 15,000 fake listings.

And then the subsequent vetting is not agreed, you know, heavy enough to filter out this character.

And, and from Google's point of view, had 15,000. So that's like, if they have 300 million business locations, that amounts to 0.0005 of a percent, very small.

from their point of view, it's essentially below the radar and it's dispersed geographically.

So it's below the radar, but it

Greg (06:32)

Well, is right.

So this is an illustration of Google's general attitude, who knows some of those subcontractors could have done legitimate work.

Maybe some of those jobs were competently done, but $79 million in fraudulent work it doesn't matter to Google.

Google doesn't care that

Mike B (06:49)

Media.

Greg (06:53)

that consumers are being harmed to the tune of $79 million because for them it's a blip.

It's a tiny fragment of their overall corpus of business listings and 15,000.

They are just very cavalier about this and it's all Section 230 because they can be.

If there were liability here, Google would be on this much more quickly.

And there really needs to be some change in the law.

I mean, beyond the defective product novel legal theory, there needs to be some change in the law that tackles this kind of situation.

I mean, you can't identify isolated problems. Like some individual creates some fake business or something.

But when you've got systemic a systemic problem like this or something that spans enough people and geographies, Google has to be responsible to do something.

Mike B (07:50)

Right, but what happens is a dollar here, a dollar there.

Soon it adds up to real money, which it does in this case.

But Google, over the four years that this has happened, every year, they've proclaimed how good their fraud detection is and how many businesses they've prevented from participating in fraud.

And so on the one hand, they hide behind 230, on the other, they proclaim their competence.

And then reality is, that real people and real businesses get hurt.

Greg (08:20)

Right, exactly.

I've been thinking a lot about alternative rating systems to traditional reviews. And no, no, no, no, no.

I like this business and this business and this business. No, that's not what I'm talking about.

Years ago, I don't mean to launch into something different, but years ago, just as a quick digression, years ago,

Mike B (08:29)

The Stirling system.

The Sterling rating, this is like, Zagat.

Greg (08:46)

businesses that had difficulty competing with review volume, even in the early days, in the early 2000s, were trying some of, there were a couple of attempts to use other signals to rate businesses.

Like, I don't know what it was back then.

Mike B (09:02)

Well, Google used it, what was called a Bayesian average early on.

And what a Bayesian average is when you have a new business and you don't have a lot of reviews, they would go out and look at the average for your category.

And if you were over it, if you were over it, they would reduce it.

So there are even statistical methods of doing it.

Greg (09:06)

Well, no, no, no. Yeah. Yeah, Well, that's interesting. That's interesting.

That's not what I'm talking about. So like today, you could use visitation, right?

How frequently is foot traffic going to the business if it's a storefront? You could use business's credit rating.

You could use transaction data. I mean, there are a number of other data sources that could be brought to bear on this question of which businesses are reputable, which businesses are popular.

and to sort of augment or, you know, even replace reviews.

Mike B (09:47)

Right, and Google does that to some extent.

They target high distress categories and reviews in high distress categories get removed much more frequently than in other categories.

And new businesses get reviews removed more. So they do it in a very broad way to do that.

Greg (10:05)

Yeah. Right. mean, this is topic for maybe another conversation. But let's move on to your inferred personalization experiment.

Mike B (10:15)

Sure. Right. Well, you mentioned that we had a great conversation with Garrett Sussman last week and the week before about Google's feature, what do they call that?

Personal intelligence, which allows Google to see into your Gmail, your photos, calendar, et cetera.

And Garrett set up a great experiment to see how much that influenced outcomes. So that made me curious.

I've been using Gemini pretty regularly for the last six months and ChatGPT for the last three or four years.

And I was curious during that period, what had it surmised about me?

I had asked it both personal and professional questions, used it to edit our podcast and to, you know, to market research and those sorts of things.

I went in and I asked it, if you... search the open web and combine with information of Gemini memory, which I believe is default on it is default on with ChatGPT just the memory function, not the personal intelligence, which you have to select on.

if you combine that with Gemini memory, as well as my search history, can you expand on what you think my personal political and social profile is?

And then I said, went on to say, classify me as a standard marketing persona.

what can you infer about my class wealth income? and for my age and for my annual income and give me a summary.

It's amazing how much it knows. got my salary spot on, my annual income spot on.

It got my buying patterns that, spot on. got my critiques of government spot on, my professional relationships.

so it did a great job of categorizing me beyond any thing I would have thought given the somewhat random nature of my inquiries, right?

I might've asked about, raising vegetables in my garden. might've asked about biking, those kinds of things.

For example, I like Specialized bikes, which are high end bike and well-made, well-designed. And I like iPhones high end.

And they inferred from that, you know, both income and predilection to buy higher quality, less frequently, more reliable sort of stuff.

Greg (12:30)

Yes, you have two bikes in the background behind you.

Mike B (12:33)

I do, one's my exercise bike and one's my bike I am now riding into work.

I just got a new Specialized bike. My wife bought it for me for our anniversary.

Greg (12:42)

very nice.

Specialized as a brand as opposed to a niche bike that has a specific capability or something like that for those in the audience who are unfamiliar with the brand.

Mike B (12:48)

Right, that's correct. But I wanted a commuting bike that, a commuting e-bike that weighed under 40 pounds that had a very long battery duration and Specialized is the only company that made it.

anyways, I am classified as making.

Greg (13:01)

So that sounds like a query for AI mode. Yes, go ahead.

Mike B (13:06)

So I, you know, my income is classified as roughly 200,000, which is spot on.

I'm a new agrarian analyst. In other words, I am critical thinker that likes to grow my own food, et cetera, et cetera.

And, uh, I social classes, upper middle class, professional managerial, uh, with high cultural capital capital characterized by stealth wealth that values utility, durability, and experiences over flashy status symbols.

You know, I drive a 10 year old car. and my age, it got the range, right.

But in the end, underestimated, I don't know if it's glad handing me, maybe it likes to be sychophantic so maybe it me I'm younger than I am.

Greg (13:45)

Yeah, you sound and act much younger than you are.

Mike B (13:49)

It said that too, but then it also ended up giving me an age in like the mid sixties or something.

Greg (13:53)

Well, maybe your enthusiasm about bikes suggests to it somebody who's younger, who's more active or something. Yeah. Right. Right.

Mike B (14:01)

Right, because I have investigated longer 200 kilometer treks that imply that I'm younger.

So anyways, we don't need Personal Intelligence to gain this level of knowledge for Google to understand demographically everything they need to understand about me.

Just if you used Gemini

Greg (14:18)

Well, so one of the features of Garrett Sussman's research experiment was to create different personas, fake personas that had different income levels, different job titles, different genders, men and women, and a range of different elements.

And what he showed was that this is all AI mode.

What he showed is that the results, the brands recommended were different based on the persona.

So somebody like you who has a taste for more expensive this or that is going to see, according to this experiment, the higher end product, the more expensive product, not the run of the mill product.

Now, years ago, as recently as

Mike B (15:00)

But I would add that as a complement to his experiment, it is clear that you don't need to turn on personal intelligence for this to happen.

Greg (15:06)

Right. That's the fundamental point. That's the fundamental insight is that I have not signed up for personal intelligence because creepy, ew, you know, I don't want to give Google any more of my data than it already has, but it doesn't matter.

That's the point of this. Google already knows everything. the more that it gets you, I mean, the more that it gets you to engage with AI, the more that it's capturing intelligence about your preferences, your behaviors, your histories, and it's going to utilize that in search results.

later on, which we'll talk about probably in ads, now one of the things that is so striking now about where we are and where we're headed in search is this personalization idea.

Until pretty recently, I don't know whether it's a year or... couple of years, Google denied that it was doing any personalization at Search at all.

The only personalization you got was location. You you got different, different, yes, you got, you got different restaurants for Buffalo, New York versus Oakland, California, whatever, because for up.

Mike B (16:02)

Which is highly personal, I mean. or different coffee shops for this block versus that block.

Greg (16:12)

Right, and that's just a feature of location. That's not a feature of your preferences or your income or your behavior.

Now, where we're headed is...

Mike B (16:20)

Although they did talk about that in, do you remember the hot pot days from 2010?

They talked about personalizing based on your preferences, which they never did do.

Greg (16:26)

Well, that's. Yeah, never took off.

we're now, Google I.O. was this week and Google Marketing Live, which is now their kind of companion ads event.

And there was a whole bunch of stuff that they talked about too much for us to discuss.

But in all of these announcements, there is an undercurrent or an explicit aspect of personalization.

So what did they announce at Google I.O.? They announced a new personal agent, Gemini Spark, which is yet another freaking stupid AI brand that people have to think about.

They announced an AI powered search box, which Liz Reed, who's the head of search at Google said is the biggest change to the search box or to search in 25 years.

And what that is is a box that will expand to accommodate your query length and will thus encourage longer queries.

Also, you can upload documents and photos and other stuff as you could with Gemini now into the search box and get Google search to respond to it.

They also announced a generative UI, sort of dynamic UI that is based on, built on some intrinsic coding capabilities.

And this is AI mode. So you can... ask very specific questions and Google in some cases will create a kind of mini app or a graphical experience that either explains or illustrates whatever you're trying to get at.

They also, yeah, right.

Mike B (18:01)

and then coding with interface design as well, and with the ability to go directly with the output to the App Store.

Greg (18:09)

Yes, mean, of the, yeah, so the AI-powered search boxes is really sort of the biggest single thing that came out of it.

They also, There were just too many things going on.

Mike B (18:24)

Well, just let's deal with that.

Let me just make a comment about the search box, which is that it's a way for Google with a single box to parse between a high intent, low funnel query and a mixed intent high funnel query.

In other words, whether they're going to give you a local result or whether they're going to give you an AIO result in a single field.

And so that is significant in terms of ease of use, mitigating buttons and cleaning up the shit that has gotten there.

Greg (18:45)

Yes, and that, yes, it's. It's yes, it's a, it's a kind of Omnibox thing, which will drive you either into traditional search results for now or AI mode.

And they're really pushing people into AI mode, more and more, and they're more deeply integrating AI overviews and AI mode, which has been going on for some time.

Mike B (19:09)

And it allows for multimodal input. if you're exploring what kind of flower is this, you can drop it right into your main browser and it will deal with it.

You don't have to go to Lens in the app.

Greg (19:12)

That's right. Right.

Mike B (19:21)

disjointed experience.

So they've consigned.

Greg (19:22)

Right. Sort of in general, in general, they were a ton of things discussed, tons of announcements.

there was this kind of, for me, there was this kind of quality of clutter about the whole thing.

Google is throwing tons and tons of stuff at the refrigerator, a kind of shock and awe approach.

You know, the entirety of IO is really primarily directed at investors, secondarily at developers and, you know, tertiarily, if that's a word, thirdly at, you know, the consumer population, the user population.

Well, competitors and investors are sort of the same group in a certain way.

Yeah, they're saying, they're saying to investors, hey, we are now in the driver's seat of AI.

We are, we are now ascendant.

Mike B (19:56)

You left out competitors, which I think was a key target.

Well, they're saying it clear enough that the New York Times understands it finally, but it's been clear that they've been in the driver's seat for the last two years.

And this just makes it, this puts a real sort of emphasis on that.

Greg (20:18)

Well, but let's talk briefly about why that is. You and I have different opinions about this, but is Gemini the best AI product on the market?

The answer...

Mike B (20:28)

I don't think it's the best. think it's goog enough, as it were.

Greg (20:31)

Yes, it's good enough.

Google has elevated it to the place where it's good enough to retain usage.

Certain things, right, and they talked a lot about it being faster. They're speeding it up.

And that's a big issue for AI because AI answers often take a while to populate. Google has multiple monopolies.

It has some great data, right? It's got a better sort of local graph.

Mike B (20:37)

It's convenient, it's fast.

So.

Greg (20:57)

data set shopping graph than other people except for Amazon in in that isolated case.

And so it's brought all of this stuff to bear to basically put enormous pressure on ChatGPT, which is about to have its IPO apparently.

And they have the full stack, the full AI stack from chips to distribution.

Mike B (21:17)

And what that full stack means is not only can they deliver this Google enough AI answer more quickly, they can deliver it more cheaply because they are not paying rent to Nvidia, right?

And they're not paying rent to electric company because often they own the data centers and the electric powering it.

So through vertical integration, they're able to drive speed, quicker responses and much lower costs.

And in the end, that's what defines

Greg (21:30)

Right. Right, and they.

Mike B (21:46)

you know, success of their future monopoly.

Greg (21:48)

They talked about bringing down the cost of AI responses.

Ironically though, notwithstanding that, they just are reducing the amount of tokens that are available to their Google One subscribers.

So they're constraining your ability to engage with Gemini, bringing that down.

I don't know what the precise numbers are, but I got to notice about that this week.

To serve you better, we're going to reduce your available tokens.

So we'll do a dedicated show on the implications for both search and local of the announcements in Google I.O.

But they are driving more and more people into AI mode. They are leaning on AI mode.

They did not replace the traditional SERP with AI mode, which many people were thinking was going to happen.

And I've been arguing they're not going to do that until they figure out the ad problem, until they can really monetize AI as well or better.

And this week.

Mike B (22:41)

But this is a step in that direction in terms of integrating the interfaces so that users feel totally comfortable moving from traditional search to Isearch without skipping a beat.

Greg (22:46)

Yes. That's right. well, Google has had Gemini as a standalone experience, which they are maintaining.

They've had many front doors. They have Gemini. They've got AI overviews. They've got AI mode.

They've got traditional search. They've got the assistant, whatever their voice stuff, which is now just Gemini.

And they introduced this thing, Gemini Spark.

Mike B (22:59)

Google app.

Greg (23:15)

which is a new agentic feature of Gemini.

It's a subset of Gemini. It's not a standalone app.

And this was the other piece that I had neglected to mention that was really significant.

In an interview a few weeks ago, Sundar Pichai was talking about the future of Search.

And he said, I could see Search being this kind of orchestration layer for agents, right?

It's going to be agentic. And so what they announced was Spark and other agentic features of search.

So the idea is the first one being informational, sort of Google alerts on steroids, informational agents where you can track and follow certain kinds of things, financial data.

Right.

Mike B (23:55)

If my airplane price to X location gets low enough, buy it for me.

Greg (24:01)

Right, except there's no transaction yet, but transactions are coming for sure.

And this is another strategy for them to keep you in search, on search, give you more capabilities in search.

And we'll see more of these agentic tools being integrated over time.

And they've talked about informational agents is going to be, or information agents is the first one and then we're going to move to transaction.

So shopping and travel and other things. And that's quite significant.

But as I was saying,

Mike B (24:32)

particularly when you layer it with UCP, their universal commerce protocol.

Greg (24:36)

Right.

they talked a lot about shopping and the universal commerce protocol.

Now, the UCP, Google's quote unquote open source, I think it's open source, shopping cart and related infrastructure was competing with the agenda commerce protocol.

And OpenAI pretty much bailed on their shopping cart, which was going to be native to their experience.

Now they're just sending people to to third party retailers. We can do a whole session on UCP and the shopping cart, but all of these assets are very significant.

And Google is accelerating the migration from the traditional experience to a heavily AI inflected experience.

And the barrier to that acceleration has been the need to maintain ad revenue, which they have not figured out for AI.

But here this week at Google Marketing Live, they introduced new ad formats. They're AI native ad formats.

Mike B (25:37)

just as a note, I am going to be on a webinar with Darren Shaw next week.

of talking about the impact of AI on Google Maps.

So if you want a deep dive on Google Maps, we will cover it there.

I see all of these elements as able to influence Google Maps, and ultimately, I think they will.

Greg (25:48)

Right.

Well, all going to come to Google Maps. And we had a dedicated podcast on Ask Maps and its significance.

But these new features in search will all be coming to Google Maps with some meaningful implications, not the least of which is personalization.

So the four ad types or five ad types that Google talked about during Google Marketing Live that are now AI native, conversational discovery ads, which are dynamically generated.

All of these are AI mode with one exception. Show up in AI mode.

Highlighted answers, which are AI-generated responses or recommendations that appear in line. They're not at the bottom or the top.

Those are also AI mode. AI-powered shopping ads, which are in search. These are dynamic.

Mike B (26:45)

So hold it, I have this idea about the information ad. Who should I vote for for president?

So that spot can be purchased and will give me the highlighted answer to it in the ad, right?

Greg (26:55)

Well, they will probably not do that for politics. That would be kind of ill-advised. But it's OK.

So shopping ads, which are dynamically generated also, those are going to be richer ads appearing in traditional results.

Those will have product images, reviews, and other explanatory text.

Business agent for leads, is a lead gen ad or a form fill.

Mike B (27:02)

Sorry, go back to what you were saying.

Greg (27:24)

ad, kind of structured ad, and they're going to plug in a Gemini bot inside of that, which is going to get information for the marketer's website and will be able to answer questions.

So people will interact with that ad and get questions answered.

And this was like an ad type that ChatGPT talked about but hasn't implemented yet.

And then they've expanded what they call direct offers, which was an ad unit that they introduced, I believe, earlier this year or late last year.

which, is basically as it sounds, you know, some sort of discount designed to get you to engage with it.

But now what they've done is they've added the universal commerce protocol into the ad.

So you can, you've got a native checkout capability. You click the ad, you can check out and pay within the ad unit.

And, there's a lot more to say about these things, but they're quite significant because this is what has been missing from advertising in, in AI.

The old sort of static ads, even ad types like LSAs or other kind of ads that have been working very well in traditional search are really not adapted to AI and the AI user experience.

And these ads are really the first foray into that.

And Google will learn a lot about how users interact with them and improve them.

And it breaks the barrier around monetization. what that also means is that once they feel confident that they're going to be making money at an equivalent level, or they'll be able to charge more for these ads or whatever.

When they feel confident that they're going to be generating money revenue from these, they then can accelerate their other consumer facing experiences because that will no longer be an obstacle.

Mike B (28:49)

equivalently outrageous level.

Greg (29:08)

the monetization issue will no longer be an obstacle to changing the interface or changing the user experience.

One final thing, one final thing, one final thing. All of the stuff that we said about personalization and their inferences about you and your income and what you like and your preferences will come into these ads.

Google has not made any announcements about these ads, but that is one thing that I am 99.9 % confident at some point in the future

Mike B (29:13)

And the only tragedy in all of this though, well go ahead, sorry.

Greg (29:37)

These ads will be more personalized. Not only how they're shown, to whom they're shown, but the ad units themselves, the copy, the product choices, all of that stuff will be personalized.

Mike B (29:49)

So we'll combine intent shoes with demographics, high income, and show me a pair of high income shoes, even if I don't do personal intelligence.

Greg (29:57)

Exactly. If I'm looking for best running shoes for men that I can use on a trail and I live in this kind of environment, based on what it knows about me versus what it knows about you, it's going to wreck.

Mike B (30:11)

Dude, go to Walmart, it's gonna tell ya.

Greg (30:13)

Well... That's right. The point is we'll get two different ads.

The point is we'll get two different ads and the ad copy might be different, right?

The ad copy dynamically generated might be quite different. So if you like off-road biking, that might be featured in the ad in a sense that somebody else who's of a comparable income level, but who has different preferences might not see.

Mike B (30:37)

What I was going to say before, which I think is, that this is happening in an environment in which basically, even though Google lost two antitrust decisions and a court case, well, basically not only has nothing has happened, but the consequences are so minimal that Google has pulled out all the stops and Skynet here we come.

mean, it, you know, this announcement combined with AI.

Greg (30:48)

Nothing has happened.

Mike B (31:03)

pretends a much more invasive reality going forward. We don't have to worry about AI killing us.

We have to worry about Google killing us, right? I mean, it's not the AI itself that's a danger to humanity.

It's the decision makers who control it.

Greg (31:07)

Yes. Well... The, will echo your statement about the antitrust case where Google was found to be an illegal monopoly in search and paid search advertising and display advertising and the.

Mike B (31:32)

And we're giving a pass because ChatGPT was there.

Greg (31:34)

Right, that is the irony, right?

That the existence of ChatGPT put, know, who knows what was going on psychologically with the judge.

The judge was afraid of acting more boldly and the remedies that were issued were totally inadequate to address the competitive environment for which Google was found liable.

And so as you point out, what we have now, is going to be 10 X what we have before because Google now has, I mean, I would make the argument that beyond fending off ChatGPT, Google's primary objective in developing all these AI capabilities is to get more data from you, to learn more about you and to deliver more personalized ads and information to you so that you generate more income for them and They can keep you in their ecosystem.

I mean, this was one of the big features of this IO set of announcements is that Google is taking what is already a very self-consciously insular system there and trying to lock you in further.

They're going to keep you more and more and more within their system.

UCP don't have to go to a third party site to check out.

The assistant keeps you engaged and fulfills all your information needs.

The data that they're sucking, you know, because you're giving it more high intent data and more elaborate data, they know more about you.

They can target you better. They can bucket you better in into a persona.

And all of that is brought to bear on this kind of flywheel of, you know, using Google more and more and more and locking you in. and the

Mike B (33:13)

And on that note, have a great Memorial Day.

Greg (33:16)

No, no, no, no, no, no. I think it's a very serious problem.

And I think there are no immediate solutions except competition. know, ChatGPT is in trouble.

They are stumbling at a time when it would be useful to have them as a counterweight to Google.

Claude has some chance of becoming that, but probably not really, because Google just has the revenues, the stack.

the data, the local graph, the shopping graph. It's just they have, yeah, I mean, they have Chrome, they have Android, they have, now they said they have five properties which have in excess of 3 billion users and 13 or something like that that have more than 1 billion.

And then they have YouTube and it's just insane.

Mike B (33:48)

distribution, Android, blah blah blah.

Like I said, have a good Memorial Day weekend.

Greg (34:09)

Well, all right, let's end on a more friendly note.

Having said all of that negative and critical stuff about Google, this is a really interesting time to be in search.

it is very dynamic and very intellectually stimulating and fascinating to see how this market is evolving.

And that's my attempt to be positive as we head out into the weekend.

Mike B (34:33)

Well, from one neo-Brahmin to the other, I'll say goodbye and thank you for a great episode.

Greg (34:41)

All right. And as I said in an upcoming episode, we will be talking about the local search and broader search implications of all of this with a panel of guests.

So in the interim, have a great weekend. Try and enjoy things to the extent you can.

Live in the moment. Appreciate your friends. Appreciate your family. Appreciate your health. And we will see you next time.