Google Lost $500B in a Day. Can AI Mode Pay?

Alphabet had its worst day ever on 29 Oct 2026, losing roughly $500 billion in value after search revenue growth slowed to 12%. Paris Childress and regular Friday co-host Simon Young of Question.Marketing debate whether Google can put ads in AI Mode without losing trust, what happens when authorized bots click ads, closed gardens vs open agent protocols, shrinking AI Mode citations, ChatGPT as a search and ads channel, and murkier measurement.

Paris: Everybody, welcome back to our special Friday edition of The GEO Show. partner in crime, Simon Young, is here with me as always to help me digest and understand and debate some of the top developments that have been happening in GEO and AI search throughout the week. Simon, first of all, how are you doing? How was your week?

Simon: Busy as always. you can obviously see that the interest in what we do is growing exponentially day by day. the number of people talking around, you know, GEO, AEO, everything that, in terms of AI, and I think one of the topics we're gonna cover today is, what Google said.

Paris: Yeah I actually have five stories teed up. The first one is about Google, and it, it doesn't look pretty. Let me give you the, the context here before I hand it over to you. the topic is Google's AI Mode money problem. so you and I have both been speculating that AI Mode is is the future of Google We just don't know It's not a question of if it's just a question of when search revenue Google reported their earnings yesterday and their stock is now down 12 which is the biggest single-day loss in their history.

Simon: Just.

Paris: To put that in perspective They lost half a trillion half a trillion dollars in market cap value.

Simon: Overnight.

Paris: And the reason is that search revenue growth slowed to just 12 So last quarter in Q2 year-on-year growth of search the search revenue their core business grew 17 and this quarter it slowed by five percentage points down to 12 and by the way it's still $63 billion but of course the markets are just looking at the pretty rapid deceleration of of the, of the core business search ads and they're connecting that with AI Mode because I think that there is a challenge now of how is Google gonna monetize AI Mode You can't It's conversational and you can't stick you can't stick the ad units the Google the the Google search ad units you can't stick them into those conversations You probably sh can't stick display ads in the way ChatGPT Ads is doing now That's probably not gonna last either the FT reports that shares actually at the lowest point had fallen 13 all the while the ads They're they are testing ads in AI Mode and and that has expanded to 10 more markets.

Simon: Including.

Paris: UK now and Brazil and Mexico they're testing these multi-turn ads I haven't really seen those in the wild yet but I mean I'm I'm in Bulgaria so we're Europe is is usually last to get the newest innovations there there isn't any kind of opt-out so it's still really really early.

Simon: What.

Paris: Do you think about all this And and the question Simon is can Google make AI Mode pay without eroding the trust that makes it useful when ads start appearing alongside organic answers or maybe even blended somehow into organic answers are we gonna Is it is it gonna be a user trust issue Because now there's a there's a really legitimate alternative or several If you don't like Google anymore for their AI answers you just jump over to ChatGPT or you jump over to Claude or even Grok And you know.

Simon: The biggest.

Paris: Risk of losing a user's trust is not the same when you're no longer the dominant monopoly of search the way they were because there weren't really many legitimate switching options If you didn't like s Google Search eh I mean you're not really gonna switch But now.

Simon: There.

Paris: Are many options to switch away from AI Mode if if.

Simon: Problem, your problem you've got is, like you said, is the trust. And, you know, if we start seeing... Well, the minute you start seeing products pushed into results, i.e., you know, shoehorned into a conversation, then I think people will push back against it. I think I certainly would. and it does look like maybe Anthropic will be, as it stands, trying to hold off for as long as possible or become, you know, a source where you can get what you want, but not be manipulated. I think that'll be one of the big things. They are all guarding their, mode or the way that they're working very, very closely, you know, and they're stopping, they're stopping the attribution going through. we've, we've debated it many times already around, you know, do we have agentic sales agents coming into chats, etc. I think we probably do. But that question as to how they make that happen and everybody's happy is... Well, I would love to be a fly on the wall, you know, sat in the very high levels of Google discussing how you make that happen, because I think if they had the answer, they would've done it already.

Paris: Yeah I actually was just discussing this with a a guy named Stuart McClure who's gonna be coming on the podcast in a couple of weeks And what what came up was interestingly that Goo Google has already been using this ad ranking formula for I mean since since the practically the dawn of Google AdWords so maybe close to 20 years now maybe where they balance They try to strike the right balance between the relevant the relevancy of an ad to the to the user's query or to the intent of the user.

Simon: Along.

Paris: With the bid with the with the auction bid the willingness to pay.

Simon: Per.

Paris: Click And and they're blending the two of those things together to come up with an overall ranking score So it clearly if you're the highest bidder and you have the highest or the most relevant ad gonna get ad position number one and then the next one gets ad position number two and number three And this was their this was their formula for ranking.

Simon: Ads.

Paris: In in the ad positions And if it if that can translate to AI it might they might be able to pull it off But there's one big difference here which is that in in the Google Ads experience you have multiple ads So ad rank is just whether you're gonna be shown in the position number one or number two or number three But in a sponsored agent environment it's probably just a winner-take-all situation where propose to the user that we invite one.

Simon: Bot.

Paris: One sponsored bot into the conversation sponsored by TaylorMade I mean we always keep keep going back to the golf clubs.

Simon: Yeah.

Paris: So it's and it's it's not you're It's not gonna be multiple I don't think there are gonna be multiple bots coming in saying you know creating a bidding war right in front of the user It's just gonna be one and so that makes things a lot harder actually.

Simon: But so put, put, put this into the equation, right? And this is a real-life example that one of my customers did this week. He was going to Hong Kong. he's looking to book a hotel. He's got his flight, but he wants to find somewhere to stay. He asked the agents, "Go out and find me the best deals and negotiate the deal for me," and he ended up... So the problem comes is if there are multiple advertisers across multiple surfaces, and then we end up, end in a world where we've got bots talking to the bots to negotiate, surely those bots are clicking the ads. They're interacting with advertisers. Where, where does it stop? You know, because that's the problem. He ended up getting... He, he got quoted originally £1,100 for his hotel for the week, and it... the bot got it down to 850, got him, got him a poolside, view, and got him a, an upgrade on his room. So.

Paris: Amazing.

Simon: Days. But I think the answer probably is we're gonna have so much more control than you can ever believe, and I think that's problem number one for Google because no- you don't have to engage with the ads, do you? You, you... However, my argument with Google Ads has always been that, and most advertisers would probably agree, that the conversion rate is nowhere near as high as it should be, and.

Paris: Mm-hmm.

Simon: Goes missing. Let's put it that way. Just for clarity, I've had a Google MCC account for 20, 20-plus years, I still run a million-plus dollars through my Google Ads account. Not mine, but my clients' every quarter, right? Not that I manage much of that anymore. It's just I've got a lot of legacy clients in there. But when you look at the amount that gets spent and is not spent correctly, which, you know, there is... There's a whole industry c- of bots clicking ads, right? And there always was. If you go back to the very start of Google Ads, people spun up loads of money sites, AdSense. People were clicking on the ads to get the ad revenue because, you know Google Google couldn't control it, or then all the services sprung up to, "Well, we can stop the bots clicking, and we can identify," and then you'd have to claim from Google that this amount of traffic.

Paris: Fraud thing Yeah.

Simon: But how do you then stop that in agentic world? Because I'm sure that'll come again.

Paris: A really good point because yeah these these were click click fraud are clearly bad bots but we're talking about good bots the the bots that have actually been that really represent real users that have been authorized to go out and do research and and in the case of your friend more than just to do research even negotiate the best deal And so those those good bots and that's not click fraud I mean.

Simon: No.

Paris: Would be very wrong to classify that as click fraud Is a legitimate bot representing a real user who's been authorized So they go out and they probably go out and do a lot of searches and they click on a lot of ads to get the best offers.

Simon: And.

Paris: Then and then maybe they they reach out directly to to those hotels and they might be negotiating with the hotel's bots And are those hotels really aware As you said wasted spend is is massively prevalent a-all all across Google Ads I think those hotels really aware that maybe a big chunk of their ad spend has come from clicks not from humans but from by bots not in the form of click fraud that they would wanna claim back but actually authorized good bots And then h that just changes the whole that just changes the whole game really because.

Simon: It, is it true to say that, and I'm, I, I think I've heard this, I don't know that it's fact, but that 50% of the traffic on the internet at the moment is bots or more.

Paris: Oh I've I've seen that already reported that it's more? it's more than 50.

Simon: Yeah it it's never gonna go it's never gonna go back to That'll keep increasing yeah I d- I don't see how they stop it. I don't, h- honestly, no.

Paris: Is that trend will never reverse just just go and look at your Cloudflare data I know for our sites I mean it's a small sample but I mean we have we have five times more visits from bots than from humans It's it's not even close So and now when you start to considering that shopping the the next domino that's gonna fall here that that more and more bots are gonna come to e-commerce sites to shop then yeah it's it's not.

Simon: Surely, all the marketing directors, CMOs, and everybody out there are panicking massively at the moment when, you know, half your traffic's gone. But, you know, if you're doing it right and the, you know, you're being cited, you're being recommended, and we're gonna talk about the citation piece in a bit, I think. But if you're getting recommended and people are coming through, the conversion rate on those clicks is 10 times higher.

Paris: Yeah Right Well let's transition then into the next topic because that does have to do with agentic commerce And the the topic is agentic commerce closed gardens versus open protocols And this is a rapidly developing story It's a really fascinating one to watch because the the the future rules of e-commerce are really being rewritten right now and people are taking pos hard positions Amazon remains their stance remains closed We are closed for any bot that is not our bot and they're they're sticking they're sticking to it hard They just reported I think also as part of their earnings call that four hundred million customers used Alexa for shopping over the past twelve months and that is i Alexa interactions for shopping has grown by five hundred and seventy percent so almost six X their CEO Andy Jassy he referred to outside agents as still not great he said I think he was just being politically correct I think he I think he really.

Simon: Liked, yes.

Paris: And they continue to block you know the headline agent that they continue to block is Meta Muse They will not let a Meta Muse agent into in through the front doors of Amazon So so that's the first one and then you have Google Google went in the o opposite direction and their stance is open protocols and Google-run surfaces And this is something I reported my in my solo show earlier today Google has launched Connected Apps platform is gonna bring outside apps into the search into search and into Gemini via MCP And they've already got fifty-plus apps I guess the early partners are already live So this UCP this this universal I think it's called Universal co Commerce Protocol direct checkout and agent payments protocol So Google is saying yeah they're basically saying yes to agents but we need to have the right rails in place We need to have an open protocol system so that we know it's safe And and then you have two others Visa Visa has said f they they have fifty AI partners that are already in production and they're also trying to to develop payment rails for any agent and they just wanna make sure that it's safe and that when an agent comes in an agent has to It has to basically prove its identity and it has to be authorized that they they represent The identity of the user that they represent is authorized basically So Visa is saying Yeah we're gonna let agents make Visa card transactions basically and pay with Visa And then last but not least we've got Shopify Knowledge base for AI agents are moved into the admin of Shopify So Shopify is all in on agents They're saying Let's merchant merchant tools are gonna be available for any agent and they see this as incremental growth If we let humans and agents shop then there's gonna be more shopping overall That's Good, for everybody and that's where it stands right now So can Amazon withstand I mean Amazon is basically going against the whole rest of the industry here Are they big enough and powerful enough to pull it off.

Simon: Good good question. if, if you look at, if you look at the way money works, so... And I mean in terms of... And I'm, I've done... Well, maybe people don't know, but I have done an awful lot in the crypto space, and when you talk about decentralized finance and how exchanges work, it's all based on a percentage of a transaction. So maybe we end up in a world where, you know, the agents just take a very small percentage of everything that they deal with, which.

Paris: Mm-hmm.

Simon: Be surprisingly Amazon's model, right? Because Am- Amazon typically takes, let's say 18, 15 to 18%. Somebody might correct me on that, but I believe that's roughly the figure of all the transactions that go through the platform from the retailer. So surely that creates a race to the bottom, and it's not hard to imagine a scenario where if you take the world of dropship- drop shipping as an example, you know, what's stopping someone going to Alibaba or Temu or whatever it is, and using an agent that is just plugged into every... Like you've said, Google's scenario, where it plugs into every single source of a product worldwide and says, "Right, Simon, you're looking for X. Here's 700 different stores that have got it, right? And the best price is this, and this one is trusted, and it will deliver in three days. Just say go and it'll be there," right? So if you look at that scenario, maybe Amazon doesn't sur- survive. It's not something that I'd considered prior to this conversation and you bringing it up, but there's got to be a, a scenario where that could happen.

Paris: Yeah so you're saying Amazon won't survive by being the low-cost leader because there are.

Simon: There's too much...

Paris: Other.

Simon: Got too much margin in it. Like, all, all of the services like your Uber Eats or Uber or... And I'm not talking in terms of buying a taxi on, you know, through an agent, but what, what I'm talking about is the model for most of those businesses that are there to sit on top of an industry, then they take 15 to 18%. And I think we, we definitely end up... Maybe there is... Well, there has to be a race to the bottom when agents can negotiate.

Paris: Ja Yeah That's a that's an interesting one And also who owns the customer relationship when the agent is the one doing the buying You don't even really know who the You don't have the the the brands the stores th they lose the direct connection with the customer.

Simon: But then.

Paris: Can't build a relationship.

Simon: That's only Amazon though, right? So I don't, I don't deal with, in theory, the seller on Amazon, I deal with Amazon, and Amazon does the support. My relationship is with, with Amazon. If I want a refund or I wanna send something back, goes back to Amazon. So what's stopping a scenario where, you know, I'm dealing with Anthropic or Claude? Claude's shipping me something, "Oh, it's arrived, it's damaged." And Claude goes, "Yep, just send it back," or, "Here's your refund."

Paris: Mm-hmm.

Simon: Know, depending on what, what sh- surface or what knowledge base or what store, agentic store or whatever you wanna call it, that the seller has. You know, the producer becomes in control then to, you know, I have X amount of product, I'm selling it through whichever channels I'm prepared to open my doors to, presumably,.

Paris: Mm-hmm.

Simon: You know, the whole world or just my local territory or whatever. But providing a bot can come in and say, "Oh, Simon's got 500 of this product," and I'm getting pings saying, "Well, we're getting offers. We're getting offers below your price. Do you want to accept?" Maybe it starts, you know,.

Paris: Yeah.

Simon: Claude is negotiating with me to drop my price and going back the other way to try and maybe make its extra margin or give- get the deal 'cause you would think that the agentic, you know, the sales bot would then want to get the best deal for the, for the customer. Whoever gets the best deal.

Paris: It's a total race to the bottom as you said Simon because every merchant will be forced down to their lowest acceptable profit margin on every single transaction hypothetically And they're gon they're just gonna get s squeezed to the breaking point How low can you go before you just say Sorry I just cannot go any lower I'm not making any money here And I don't know I guess that's that's gonna be maybe a boom for for Alibaba or you know Chinese-made products but the stuff The people are gonna be have a hard time protecting their their margins I.

Simon: May- maybe China has just won then because it's gonna be where everything gets made, and then obviously shipping cost would then determine whether you can then buy it. Maybe then... You know, I, I was talking to another customer of mine who, he, he makes air conditioning units in the UK. but no one can beat him on price because he ships such a large volume of components from China and then assembles them locally. So, you know, the- but the, the actual product itself, you can buy it cheaper in China, and you could order it, but the, the shipping cost from China to get it here would negate the, you know, benefit from buying it locally.

Paris: Only reason why your your friend is probably s still able to be in business right Cause the shipping costs make it prohibitive Yeah.

Simon: Yeah.

Paris: Well let's keep it moving This is a this is a great debate but we've got a few more to cover So let's go to the next topic which is fewer citations and the fight over who gets them So some some big moves Again we talked about AI modes a advertising existential crisis that's looming At the same time the AI answers are showing fewer citation sources and that means that every citation slot then becomes worth more And what we have seen and what I reported on also this week is that AI Mode according to data by The Search Spot AI Mode is now surfacing per answer Cite average citations in AI Mode went from 16 on average down to four so 74 fewer Now I don't know if this is if this is really a trend if it's still kind of early but that that's major So fewer spots so does does earned media attract more than So this is basically now ear earned media because at the same time another another piece of information that's important here is that something like 85 are not the brand's website They're they're third-party sites basically So it's actually 84 of AI citations come from earned media or third-party media And and journalism's share fell from 27 to 22 So you have fewer spots You have vast majority of those spots are earned media So what's ha what's now happening to GEO Is it becoming almost entirely an earned media game with fewer spots that and is that gonna become a pay-to-play kind of thing Or what do you make of all this.

Simon: Well, the, the pay-to-play piece is really interesting. You know, I don't... Do we end up in a world where you just pay to end up in the-- being recommended, and then the whole thing is advertising, like we've just talked about? You know, effectively, you know, you're just getting sold all the time, and all the answers come from advertisers. I don't think that would suit any of the models. in terms of citation, though, do you think this... I'll throw this one back to you a little bit. But they may, may well not be showing as many citations, but the, the fan-out query is obviously there and growing, I believe. But then showing a citation may be reducing just because they want to give you the perfect answer with no distraction, right? We want-- That was Google's, Google's original promise to everybody when all the- all those years ago was to give you the answer, to give you the resource and the answer as quickly as possible with as least effort, right? And if you could...

Paris: In the cleanest in the cleanest way possible.

Simon: Yeah. And if we can.

Paris: Google started int I mean before Google had ads even you know that was actually that was the perfect Google experience really.

Simon: But surely if we got to... If you got to a point where there was no search engine and it was just an AI response, as in an agent, and you got the perfect response every single time without doing any further s- searches, that's utopia. That's where we want to be, or that's where Google would want it to be. it's a customer service or answer satisfaction race to say, you know, "Am I able to serve this person?" And then we get into... See, the, the thing where I think we go is hyper-personalization, so it then learns about you to the extent where, you know, you and I might ask the same exact question about a similar product but get different answers.

Paris: Yeah,.

Simon: Yeah it's a very complicated one. But yeah, the point is get recommended rather than cited.

Paris: Yeah It could be that Google has already figured out through AI Mode that people aren't really paying attention to those citations and they're not clicking I mean th there's there's a lot of data on clearly on the referral traffic is is I think we talked about this last week was 0.6 or 0.7 of Google organic traffic So people are not clicking and they're and probably not really paying attention So then why not just make the put more focus on the answer itself remove remove the distracting clutter of these citations reduce it down from 16 down to four I mean they're probably just running another test I don't know if this is gonna stick long term but this test pr is probably in response to some type of a learning that we've been putting too many citations in there and it's it's not helping the user at all It's not adding to the user's experience and they're not they're not clicking.

Simon: No, I could 100, 100% go with the it is a distraction and... But do you... You know, maybe it is people like you and I that we're looking that closely at the answer that comes up, that we're going, "That's odd. There's no, you know, there's no citation link there. Why, why isn't there?" and it could obviously be a test. These things change day to day, right? Like, it's changing definitely week to week, things are.

Paris: About some some update even next week about this The personalization thing is interesting too because I think that that is the the increasing degree of me-memory I guess it's the really the the ability for these models to consolidate all this memory and to use that for hyper-personalization So it just progressively gets better and better and more personalized as you accumulate more memory I think th-th-this is actually what is ushering us into this new age of personal agents that are able to be proactive because at some point I I the I just know so much about you Simon that I can start to predict what you might need I can remind you of things that you might forget And we we talked about before we started recording we talked about the fact that neither of us can now even keep up with our Grok bots because there's just there's just too much fi incoming fire coming at me I I I might I could practically spend my whole day just trying to keep up and read all of the all of the Grok bot chats coming at me and approving everything and sometimes having to re-log in here or reauthorize over there but basically we've shifted from us being the proactive ones going in and entering a prompt to now we're we're more the ones that are- getting bombarded with with proactive agents And I think that is because of personalization If I if I know so much about you I can say Hey Simon yeah it is probably it's time to update your running shoes or maybe time to take a l a look at a new pair of golf clubs or But the productivity And then I don't know do you I mean do you still need to go to Google anymore I mean if we become mostly more in reactive mode to these agents instead of proactive mode that that changes the entire game you.

Simon: But the worry, the worry then is you move into a world where are you then being influenced by what the agent is then feeding you? 'Cause of course you are. 'Cause we're all... That is the case already with all-.

Paris: We just become puppets of that agent maybe huh.

Simon: Well,.

Paris: They're our puppet masters.

Simon: I had a, I had a... Prior to the last UK general election, I had quite a lot of young people working for me, and they were all between... I class them as betw- well, they were between 18 and 23, and I spoke to them and said, "Well, you know, how are you, how are you voting?" And I knew already that they were all gonna vote for a particular political party. And then I said to them, "Well, why is that?" You know, "Well, that's, you know, that is what I see, that's what I believe in." I said, "Do, do you see anything else on your social media feeds?" And they're like, "Well, no." And you're, you're effectively then being trained. I'm not saying... Do you see what I mean? The, the information that's coming at you is clearly...

Paris: Yeah.

Simon: So whether the agents end up doing that with us, I would probably... You know, I was... There was, there was a post yesterday on LinkedIn, someone talking about not writing AI slop, and they are, you know, vehemently against AI, only write human-written content, etc, etc. And I put a comment on saying, "Well, you might as well go and buy some tinfoil and wrap your head in it." You know, because if you wanna join the tinfoil hat brigade, you, you're just not gonna win. You c- cannot possibly get past the fact that AI will, will beat you every single time.

Paris: Yeah All right Simon let's move this is getting ph-philosophical here but yeah echo chamber is self-reinforcing Th-this is social media on steroids though when when your agent becomes your echo chamber I mean God God help us but let's move to the next topic We have we have two more to go I wanna make sure we have got enough time Fourth topic is about ChatGPT ads and and well both ads and organic but ChatGPT basically really growing up as a search and ads channel So here's the context so OpenAI explained that its ad system has an orchestrator and then an auction then a final check and that everything is kept separate from their answers So they're saying there's a there's a firewall between advertising and our objective answers that we're providing to users more on this we've got creative enhancements which now means that ChatGPT ads can rewrite advertisers ads without without even their approval or consent This is switched on by default as of the 30th of October today so just consider that ChatGPT ads can on its own at its own discretion they can rewrite your ads if they think they can do a better job and probably often they will also some ChatGPT ads are appearing through Amazon's DSP so that means that now you can buy ChatGPT ads programmatically OpenAI something called OpenAI Signals version 3.0 estimated this was reported in Search Engine Journal informati-information-seeking conversations let's say prompts has risen to 20 of the total consumer messages coming out of ChatGPT And that's that's basically the the the chunk that you can directly compare with Google searches That number grew well over a two-year period it grew from 16 to 20 So basically it's saying one out of every five ChatGPT conversations total one out of every five is basically the equivalent of a Google search because it is information-seeking That's interesting and that basically put it at one 1.24 billion search-like messages per day which is about 7.5 of Google's volume Now that's still a small percentage of Google volume Google has is is massive but it's significant and it's growing So the the point here to put to you Simon is this 7.5 Of Google is it is it big enough to get a dedicated budget and you mentioned ad-free becoming a selling point of Claude I think that is also really interesting So what what do you see happening here ChatGPT now 20 of ChatGPT's volume is basically Google equivalent is That's a lot That's big and it's 7.5 and growing think that at at Is it time to say that this is a real threat to Google.

Simon: I'd, I'd say if you add all of the other surfaces up, absolutely. You know, culminating in what we talked about earlier in the piece. to say would you add or allocate ad budget to it, have you, have you used the OpenAI ads interface yourself?

Paris: No our team has been using it I haven't used it myself.

Simon: Horrendous.

Simon: Oh, it's like going back in time to, like, in the 90s. It's that poor. You would think, you know, surely they would say to their own engine, "Please, can you just design us, a Google Ads style interface that's as good as or better than?" But no. It's like it's, it's out of the arc. that said, well, there is also the argument, do you build open AI ads into a strategy where you're trying to get listed in ChatGPT? Well, you know, is there a relationship between being recommended or cited in ChatGPT and spending money in their ads? That's a whole nother question. But, I think personally that is true. Google has done it for years, although they would never admit it that, you know, does SEO improve as you spend money on ads? Well, you're getting more traffic for a start, and you're getting people to stick on your page from a more relevant search. Then there is a relationship between people not bouncing off your page and buying product, and therefore your, you know, your website as is being higher in search, right? That would be the closest, analogy you could make. But yeah, I think if... Does, doesn't the public... You know, when they talk about AI, they all still, I think, talk ChatGPT.

Paris: Yeah.

Simon: So you know, you give, you give somebody ChatGPT for the first time, and it opens people's eyes, and they're like, "Wow, I don't need to go back to the olden days of searching on Google." So I think, yeah, the problem is, is Google is panicking. They have to get to a point where, they serve a better answer more quickly and try and retain their customer. But, you know, the more serious users are already using, you know, Claude, Perplexity, use a mixture of all of the engines, right?

Paris: Yeah Yeah Which leads us perfectly into the last topic and I think we talk about this from some angle every every week but it's about measurement and can can we trust the measurement The whole measurement picture here is just getting murkier murkier and murkier as we go But let me let me read you off a few data points all of which are really undermining our classic sense of security in measurement in digital marketing First and this has now been confirmed that Google is serving poisoned results to rank trackers SEMrush SEMrush confirmed it and and as did Ahrefs that they are actually serving completely fake made-up results to to keyword rank trackers to throw them off of the scent that's point number one Point number two AI Mode ads report only in the aggregate so advertisers cannot isolate AI Mode performance the third point Google's privacy policy this is something I reported earlier this week too Google's privacy policy now lets connected Gmail and Photos data personalize ads So I suppose this has to be opt-in but it's a sign of what's coming Ad personalization based on your Gmail emails and your photos your Google Photos so next up a ChatGPT bug dropped the UTM parameters This was only for about a week They said it was a it was a mistake from 8th to the 15th of October but it still happened and maybe it was intentional And then the last data point which is showing where AI Overviews is going AI Overviews is now at roughly 82.5 of US brand searches So when you're reading your GA4 report and you see organic search that's that's AI Mode actually That's that's actually AI I mean or is it Did somebody who looked at the AI results they looked at the AI Overview they got influenced and got motivated by something there and then scrolled down and clicked on a classic blue link Is that really the same as organic search now So it's just the whole picture is getting murkier if rank data ad reporting referral traffic are all becoming less reliable then what should we as practitioners actually be reporting to our clients and to leadership Simon over to you.

Simon: It's a nightmare, isn't it? Absolute. Like,.

Paris: We-we're backwards in time I mean we're going back to you know Mad Men days.

Simon: Most... So if you take the majority of businesses out there, the- most of them don't have the proper end-to-end attribution built into all of their campaigns anyway, right? Bel- probably SME and below, right? The... If you've got, you know, if you've done it properly and you've got a decent agency, you should be able to tell. But like you say, now results are being, well, faked or influenced with the... It's, it's an actual war that is going on between Google and all of the other services. And like, like you say, if Amazon's not letting Muse in or other bots or, you know, f- ring-fencing itself, panic mode is in effect. And when you say, yeah, UTM links are dropped for a number of days, absolutely 100% that is on purpose. A test.

Paris: Accident huh.

Simon: Yeah, it's a test, right? So,.

Paris: Mm-hmm.

Simon: Yeah, they're, they're all trying to grab the slice of the pie, and I think Anthropic and OpenAI, I think they know that they're winning, right? They just- they've just gotta let people come to them. So Goo- Google, Google, I would be very surprised if maybe by Christmas we don't see an, a fundamental change in the way search works with Google. Like imagine if they just changed the whole interface and you could now choose to be completely... Well, you're getting forced completely into AI, right?

Paris: Yeah So Simon let let me ask you this before we wrap up which we we're gonna have to do in a minute Google stock is down 12 are you buying or are you selling It's their largest single day drop in in the history of Google Isn't this the best buying opportunity ever.

Simon: Well, it depends if you're day trading, right? If you're, if you're day trading, you're probably, yeah, you can do well for the next, let's say, couple of weeks as it slowly creeps back up.

Paris: The market will Yeah. the the overreaction there'll be a correction to an overreaction It'll drift back up a bit maybe.

Simon: Yeah But no,.

Paris: But.

Simon: Not long term. But who knows? Maybe they.

Paris: Yeah.

Simon: Maybe they do the deal with Amazon that they should do or, you know, they do a deal with Muse or Facebook or... Imagine what conversations are going on in the background there.

Paris: Oh yeah yeah I I think you know Amazon is taking their hard stance on anti anti-agent Google is also trying to say Let's build these open protocols and maybe the two of them are actually gonna come up with a a combined solution to say that we will we will allow it but there's gonna be these strict strict rules and definitions for it.

Simon: Are you, telling me that if I open Amazon on my desktop and I put my Claude bot on it, it won't run? 'Cause I think it will in browser.

Simon: Be services that go around over and above all of that and work inside all of these platforms.

Paris: I mean th there are probably gonna be ways to disguise these bots and it'll be a total cat and mouse game for a while But.

Simon: Another,.

Paris: Simon.

Simon: Another interesting chat.

Paris: Yeah yeah I do have to wrap up but great a great great Friday chat as al always with you here And for those listening, thanks for tuning in as always, and see you all in the next one.

Simon: See you next week.

Paris: All right.

Google Lost $500 Billion in a Day. Can AI Mode Ever Pay?

Key takeaways

Can Google make money from AI Mode?

What happens when good bots click ads?

Closed gardens or open agent protocols?

Are AI Mode citations shrinking?

Is ChatGPT becoming a search and ads channel?

Why is measurement getting murkier?

Correction

Links

Full transcript

See how AI answers describe your brand

platform

Resources

About us

use cases

compare us