Google vs. Amazon vs. TikTok vs. Meta: Welcome to the AI Ad Wars | Growing Ecommerce

Episode Transcript

Mike: 0:08

Welcome to another episode of Growing Ecommerce, brought to you by Smarter Ecommerce. We are your hosts. I’m Mike, head of Ecommerce Insights, and this is Chris, chief client officer at Smarter Ecommerce.

Mike: 0:23

All right, Chris, we’ve got an exciting episode today. We’ve got a few topics in two clusters. The first theme is about these highly automated AI campaigns from the platforms. We’ll talk about PMAX and something going on at TikTok, and then we have a few items around the intersection of e-commerce and AI. One glaring question: is AI a bubble or not? Surprisingly, shockingly, Mr. Bull Market himself, Christian Scharmüller, says it is not.

Chris: 0:59

It is not a bubble. That’s what I can tell you, but I’m looking forward to the debate. I know you’re sharp, but I think I have good arguments on my side.

Mike: 1:07

Let’s go for it.

Chris: 1:07

Let’s start with the debate now.

Mike: 1:09

Let’s warm up.

Mike: 1:12

First, I want to set the scene a little bit. We’ve touched on PMAX and then all the PMAX clones out there. I think Google has been very pioneering on this technology for years, going back to automated bidding. Every single step forward has made it easier for other platforms to pick up where the last one left off. TikTok became a very automated platform in a very short time because they were looking at how Google does this. Google is the biggest advertising platform in the world. Facebook, they’ve all done it. Even LinkedIn has their own PMAX. It’s crazy.

Chris: 2:04

So you call it, rightfully so, the PMAXification of the advertising landscape.

Mike: 2:08

Exactly.

Chris: 2:09

What a quote, mate. Thanks.

Mike: 2:16

There’s a “but” coming here, though. There was this clear trend where Google was rolling more campaign types into PMAX and it was becoming increasingly a black box. Every other platform was headed in the same direction. This is where I still think PMAXification is very relevant as a trend. Google—we’ve gone through this narrowing funnel where there’s a loss of control and degrees of freedom on the advertiser side. Google was taking those degrees of freedom on their side to deploy budget where and how they see fit.

Now, I think we’ve passed through this most narrow point and more campaign technologies are opening up, at least at face value, like Demand Gen, AI Max for search, and also more options and controls are coming back. If we look at Demand Gen, there are channel controls. If we look at AI Max, it’s very customizable in the way you can implement it. And Performance Max, most recently, added the channel report. By the time this episode airs, it might already be general availability. It was expected to go GA by mid-quarter; Google has revised that to the end of the quarter. We see this opposite trend now, and the question is: will other platforms follow that or stay on their current trajectory?

Chris: 3:59

First of all, to share some insights on our relationship with Google: we are one of the biggest partners for Google in Europe, managing a lot of advertising spend. Given that relationship, we’re very close to Google. We appreciate Google because it’s a great company and a great tech platform. However, we are one of the few players who challenge Google on a lot of things—rightfully so.

When this whole PMAX thing popped up, we were very critical because it was a black box. It diminished leverages and freedom for not just the retailer, but also for agencies, SaaS providers, and so forth. We challenge Google a lot, and it’s time to share some roses with them because Google listened to the market and did a lot of things right the last couple of months. If you look at their roadmap, we look at a completely different trend to what PMAX was perceived at the beginning. You have a lot more freedom and control opportunities coming back. I think PMAX is the superior product compared to smart shopping campaigns, and you see the same trends with all these new campaign types.

Google is doing a lot of things right here. I don’t know how many episodes you’ve listened to, guys, but we weren’t too fond of Mr. Zuckerberg. However, if you compare the strategy of Google with Meta, it’s pretty clear that Google wants to leave some control in the hands of either the agency or the retailer, whereas Meta is a black-box, AI-first, no-control strategy. Google listened to the market and a lot of good things are happening here, so I’m pretty positive about it.

Mike: 6:23

I’m about to be critical of Google’s search partner network in just a couple of minutes. PMAX has had many things rectified: there’s now real search term reporting and real negative keywords.

Chris: 6:49

It would be awesome to prioritize them, but that’s another story. Campaign priorities would be great.

Mike: 6:54

There’s channel reporting. Now you can see where you are appearing on YouTube and all this stuff. But what drives me crazy is that there’s no opt-out for the Search Partner Network. Maybe I’ll just go into that right away. Are we ready?

Chris: 7:15

Let’s keep it moving. We can come back to the first agenda point, but let’s do this now.

Mike: 7:21

On PMAX, you cannot opt out. Other campaign types, you can opt out of the Search Partner Network if you want to, but that is not possible in PMAX. I’m on a crusade; I want Google to give an opt-out option on PMAX. It’s important to me.

Chris: 7:45

Ladies and gents, if this happens, it was him. You can send your thank-you cards to Mike.

Mike: 7:48

I’m committed. Google has done a couple of good steps here, and the most recent one is that they made Search Partner Network placement data available for shopping and search campaigns. It’s actually been available for PMAX for over a year. My big complaint there is that you can’t segment this data, so it’s just mixed in. But now you can view this for search and shopping, and I’ve been looking at it—it’s not good. It reinforces my view that these are not partnerships anymore. They’ve included everything they can. I’m looking at this placement data and there are websites on here that I can’t even click on because IT would slap me.

Chris: 9:46

I would immediately get an email.

Mike: 9:48

There are malware sites that are not fit for human visitors, let alone to have ads on them. There’s no genuine commercial intent; these are bizarre domains used by hackers and malware.

Chris: 10:17

Why is Google opening this up now? In theory, I could look up every website where my ad was shown. Is there some risk implied for Google? If someone is really digging into these thousands of websites and finding really bad ones… Why are they doing this?

Mike: 10:48

When they first rolled out Search Partner Network placements a year ago, both the placement data and the narrow opt-out possibility came because of incredibly damaging reports. Independent researchers, most famously Adalytics, demonstrated that Search Partner Networks included sanctioned Iranian websites. They caused a US Army ad to be shown on a sanctioned site, and big spenders were pissed. They were going to turn off PMAX. Google immediately had to cool that fire to maintain PMAX adoption. So they enabled reporting, and this is a follow-on effect. People have been demanding this data for years and Google is listening. There’s a risk to them in exposing the data, but there was a risk in not exposing it, too.

Chris: 12:27

Circling back to PMAX features, might I be right that Google is doing this because they saw some stalling in their PMAX adoption?

Mike: 12:46

100%. PMAX adoption has stalled. PMAX share of wallet came at the expense of shopping and search campaigns, particularly dynamic search. This new solution, AI Max, is very similar to dynamic search. They saw it plateauing and needed new ways to maintain high levels of automation. One thing they do is develop new features like AI Max, which brings PMAX to search. Another thing they do is try to eliminate complaints about PMAX so that people keep spending there rather than reverting to shopping.

Chris: 14:00

That’s how it should be. It shows that Google is still listening to and reacting to the market. They saw a stall, understood the reasons, and are now moving in the right direction. It’s good news for us as a tech provider and for every retailer because you can again separate yourself from the pack by being smart. Still some way to go with the Search Partner Network, though.

Mike: 14:46

Yes, I will keep fighting. I’ve got you covered.

Chris: 14:58

What about TikTok?

Mike: 15:08

TikTok immediately jumped right into automation. TikTok Shop has its own campaigns and its own PMAX version called GMV Max.

Mike: 15:34

Most brands and retailers talk about revenue; a marketplace talks about GMV (Gross Merchandise Volume). TikTok Shop is concerned about maximizing its GMV as a marketplace, and they named the campaign GMV Max. It sounds like it’s designed to maximize the top-line revenue of the TikTok Shop marketplace, not necessarily individual advertisers. As of September 1st, they’re forcing everyone advertising on TikTok Shop to exclusively use GMV Max.

Chris: 16:28

You’re literally forced to use that PMAX-ificated campaign type. If you compare it to PMAX on Google, are there differences in terms of freedom or leverages?

Mike: 16:48

I’m not as familiar with the technology, but it’s highly automated. There have been signs from Meta that Advantage+ could become mandatory. Mark Zuckerberg made statements that they basically just want people to look at the numbers go up and trust the AI.

Chris: 17:20

He doesn’t believe in vendors, whether it’s a SaaS provider, an agency, or even in-house teams. He believes in AI end-to-end. Google has a different strategy.

Mike: 17:56

I wrote about this in 2019—a hypothetical end state where these platforms would be fully automated, like a hamster in a cage with a drip feed. You can’t really competitively differentiate; all you can do is perhaps spend more. That is the risk of where Meta could go.

Chris: 18:32

The end game probably will be that the AI knows best which ad is best for me as an advertiser. That is not a good place to be.

Mike: 18:42

I’m not an AI hater, but I think for the foreseeable future, human plus AI is better.

Chris: 19:01

Talking about one of the bigger AI players, Amazon: they are blocking AI crawlers. Why?

Mike: 19:15

They’ve been doing it for a bit and even started blocking Google. Marketplaces are highly threatened by AI because there’s the chance that a chatbot itself can serve as a marketplace. Why should Amazon help support the shopping experience on Perplexity or Gemini by feeding their data? They aren’t in a position where they need the traffic from those platforms yet, and they want to stop them from getting big. They don’t want these models training on their fresh data.

Chris: 20:55

Strategically, it’s the right move, though it will hurt business models depending on that information.

Mike: 21:11

Amazon has the luxury of that dominant position. But if you are an SMB or a median e-commerce business, you take a risk when you block these crawlers. You want to ensure you’re visible there, even if it’s a devil’s bargain in the long term.

Chris: 21:49

What about OpenAI?

Mike: 21:54

OpenAI is looking at a commission-based model where if someone uses ChatGPT as a shopping assistant—they’re working on integrations with Shopify and might be starting their own Merchant Center—there’s a native checkout experience. Then a brand pays a commission for sales through that experience. In that case, the AI acts as the marketplace and the brand gets almost nothing. They haven’t really acquired a customer or gained first-party data; they’re acting as ChatGPT’s warehouse.

Chris: 23:20

As Chief Revenue Officer for OpenAI, do you think it’s the right strategy long-term?

Mike: 23:34

For OpenAI, it makes perfect sense. I think Google’s “Buy on Google” feature will come back to life in an AI shopping form.

Chris: 23:46

That’s one of the biggest risks for brands. We’re going down that direction again where you can’t control the customer journey or customer loyalty. Who is positioned the best when these AI platforms act as marketplaces?

Mike: 25:06

Google. These other platforms are experimenting with Merchant Centers and Shopify integrations for the first time. Those things have been around for years at Google. They already have a universal checkout experience; they just have to assemble the pieces.

Chris: 25:41

What about the impact of AIOs (AI Overviews) on the core business of Google?

Mike: 26:16

Quarterly earnings were very positive. Right now, AIOs are a cost center, but Google will be serving ads on AI overviews from shopping, search, and PMAX campaigns. For now, it’s probably neutral for Google. It’s very bad for publishers on the open web because Google is summarizing their content.

Chris: 27:24

AI in combination with paid search isn’t a zero-sum game, but right now it seems that way. Commercially intended search queries remain comparable to 2024 levels. AIOs increase the value of the product, and Google is managing it well. The open question is the monetization strategy.

Mike: 28:22

When you make informational queries, you see an AI overview almost every time. That’s not yet the case with shopping or product-related queries. They are sticking to their classic experience for those. We’re seeing organic product carousels based on structured data, but they aren’t touching the paid listings yet.

Chris: 29:29

90% of all commercial searches still belong to Google. The share of searches with commercial intent stays stable at roughly 15%, which is a good thing for Google.

Chris: 29:58

Last agenda point: Is AI a bubble, yes or no?

Mike: 30:07

Let’s keep it a lightning round.

Chris: 30:30

I believe the stock market reflects what big institutional investors believe. When you look at the money flowing into AI, it’s staggering. If you look at the dot-com bubble, price-earning ratios were 100 to 200, and it wasn’t consolidated. The AI business right now is dominated by big players like Amazon, Meta, Google, and Nvidia. These companies have very healthy financial stability and price-earning ratios. Alphabet is roughly 22; Nvidia is 58. From a purely financial perspective, it’s not a bubble.

Mike: 33:43

I come at this from the technology side. I think we are at a point of diminishing returns on what these models are capable of. They’ll get faster and cheaper, but will they get better on intelligence? I’m not sure. I’m also very bearish on the “agentic” side because of the security ceiling. For example, Perplexity’s agentic browser, Comet, had a massive security flaw where a malicious prompt on a website could cause the browser to compromise your other tabs. AI is like a child—naive and easy to manipulate. When you combine that with the access an agent needs, it is a security risk that isn’t going away soon. It’s too immature to have a near-term future.

Chris: 36:28

It’s still not a bubble, but we’ll discuss it next time.

Mike: 36:33

We’re just beginning to scratch the surface. Thanks everyone for tuning in to Growing Ecommerce, brought to you by Smarter Ecommerce (SMEC). See you all next time.