Google & OpenAI are FORCING You to Pay: The AI Ad Overhaul | Growing Ecommerce

Episode Transcript

0:00:09 Mike: Welcome to another episode of Growing E-Commerce. I’m your host, Mike, and with me today is Chris. Just Chris.

0:00:18 Chris: That’s our rolling thing now, huh? I always say it’s a pleasure being with you. Actually, it’s more than that. I feel honored. Just the other day, I’m not kidding, I was talking with a lot of our clients and it’s fascinating how many of them don’t just know you, they are in awe. When the name Mike Ryan drops, it’s like, “Yeah, I know him. Is he here? Can I talk to him?” It’s fun. It’s an honor.

0:01:03 Mike: That is completely ridiculous, but I’ll allow it.

0:01:05 Chris: It comes from the bottom of my heart. I know it’s true. You do a fantastic job. We’re super glad having you.

0:01:07 Mike: I appreciate it. I’m just sharing the knowledge that I get from the conversations that we have in here. Let’s have a good episode. Now that I’m blushing, let’s kick it off. Do you want to do another intro? No, I’m good. I can’t survive another one of those. Let’s kick it off with the topic that is going to be coming up more and more in the year ahead: ads in AI overviews and ads in AI mode—AIO and AIM.

0:01:46 Chris: AI has to be in there in order to be a good product.

0:01:49 Mike: Yeah. Here in Europe, AI mode debuted a couple of weeks ago at the time of airing. You found it now, right?

0:01:58 Chris: It was live.

0:01:59 Mike: They announced it and it took probably another week before it rolled out on my account.

0:02:04 Chris: I’ve seen it, but it was on and off. It’s interesting. A couple of weeks ago they launched it in 50 countries and 50 languages. It was a huge rollout.

0:02:16 Mike: Exactly. And I think both of them are now in over 200 more. Basically every language you could wish for is supported. It’s getting pretty big, and now comes monetizing that.

0:02:30 Chris: It’s becoming big, and you have been covering this a lot. The big question is, they have these great products now, so when and how does the money train come? What’s the monetization strategy? We both talked about how, rather sooner than later, there will be ads embedded in some way, shape, or form. It has happened.

0:02:56 Mike: There are some details that we can share here. A lot of things are not known yet, but to kick it off, there are these two surfaces: AI overviews, which are the auto-generated summaries that appear in search. Google has this Venn diagram showing all their search queries and then a subset where there’s no one right answer or where they’re very long, detailed queries. This is what will trigger an AI overview. In my experience, it appears on 99% of the easiest questions.

0:03:49 Chris: Yeah, anything.

0:03:52 Mike: Also, what pisses me off is that I’ve gotten very lazy with my typing on my smartphone when I’m Googling. I don’t need to be precise because they figure out my worst typos. They know what I mean and I can type words in any order and they’ll sort it out. That was the magic of the Google search engine result page, which I think peaked a couple years ago. Now, sometimes instead of straightening out my typos, they’ll serve me an AI overview that’s trying to make sense of the nonsense I typed in. It’s hurting.

0:04:28 Chris: They have to work on that. It’s not good.

0:04:31 Mike: It’s not good. To me, they’re really over-delivering the AI overviews. I still come to Google search for search results. The other surface, of course, is AI mode, which is much more like their version of ChatGPT.

0:04:49 Chris: Going strong, right? Over 100 million active users already. It’s probably more.

0:04:55 Mike: They’ll be pushing it on their domain homepage and everywhere.

0:05:01 Chris: They’re pushing it hard. So these are the two flagship Gemini-powered products directly embedded in the search results page. What’s going on with ads now?

0:05:12 Mike: Ads can serve adjacent to the AI content. For example, it’ll spit out an AI overview, and ads can appear above or below. That’s already happening in every market and language. Ads can also appear adjacent to AI mode content. The next step is that ads will start to appear inside of the AI content. That’s going to be the big thing. Right now, ads within AI content are limited to the US and English on mobile and desktop. It’s in an experimental phase through Q4 2025. We’ll have ads inside of AI overviews by the end of the year in other markets, and we’ll see how long it takes with the AI mode stuff.

0:06:17 Chris: It’s such disruptive and dynamic times. I love it because although there are so many untackled questions—for instance, how I can make sure that if there’s this new shopping assistant, which AI mode will become at some point, that I get shown? What can I do to be favored by the shopping assistant? What about the ad rank? There are so many questions. I think it might be a threat because it’s unknown, but there are a lot of chances for retailers who know their stuff.

0:07:12 Mike: I think so too. We talked in a previous episode about this intermediate phase and the curve where traffic to websites might decline while AI-facilitated purchases increase. We don’t know how sharp those curves will be. In the first place, Google has chosen not to expose AI overviews on e-commerce queries that much, which is something they can pace. But AI mode is not in their control because it depends on how the user uses it. They face pressure from other platforms who are going aggressively on this e-commerce shopping assistant thing, so they’re going to be forced to do this.

0:07:58 Chris: Forced at some point. As of now, you don’t have any clue if the AI mode serves inventory from PMax. You don’t know if the PMax ad triggered by AI mode works any differently compared to standard search. Google just doesn’t know how this will play out, but there’s pressure on them to try it out. They are doing it smartly most of the time, but it will be interesting to see the different search behaviors. How will the ads perform there?

0:08:55 Mike: We have to be permitted by Google to find this out. Google’s statements about performance are very Wall Street oriented. If you jump into the FAQ, it feels like it’s written for a CFO, not an advertiser. It’s for someone looking for things like the tick rate. Will this have a negative impact on the amount of ads shown or the click-through rate? These are legitimate questions for shareholders and for advertisers who have a business dependency on this channel. The answers are very fluffy, without much substance.

0:09:56 Chris: “Fluffy” is very kind. I want to know what’s going on there. As far as I know, the online retailer can’t even control if they are shown in AI mode with an ad.

0:10:17 Mike: No, you cannot target these properties specifically, and you also can’t opt out. If you are running standard shopping or search campaigns with broad match, you will appear in these properties. Google has sales slides saying the key to winning AI search is the “Power Pack,” which is actually completely misleading. Demand Gen, another part of the Power Pack, does not serve on these properties at all right now. It’s very misleading. The sales slides are very confident, saying the ability to match intent has never been higher, but I think it’s a very challenging moment for Google. These queries and behaviors are new. It’s hard for their algorithms because they don’t have historical performance data. It’s all long tail.

0:11:43 Chris: Search behavior will change. Google implicitly or explicitly educated the user on how to search for a product, and it’s certainly not a natural language-based search query. If you put a query for a specific indoor tennis shoe between 100 and 200 euros into Google, you get shitty results. This will change now, but these search queries will be new to Google as well.

0:12:14 Mike: Long term, it’s a demographic dream because people will volunteer more about themselves and what they’re in market for in an explicit way. It’s like an online consultant.

0:12:31 Chris: Down the line it will be good for the Google end user. The question is, what about the online retailer? How do they deal with that? How can I make sure I can push that?

0:12:58 Mike: We talked about it last episode: first-party data. If someone is creating these very detailed queries with a lot of specific context, they’re doing all the work. Normally, someone might land on your website in browsing mode and use your filters and taxonomy to find their way around. Now they’ll do all of this qualification through the prompt. In order for you to appear, you need as much information density as possible. If they’re asking for a vegan alternative with specific features and Google doesn’t know that from your feed or landing page, you’re not visible. You need excellent landing pages, excellent structured data, and a really detailed feed with as many attributes as possible. All these teams need to work together.

0:14:42 Chris: It’s a hell of a challenge. It’s more important than ever to walk in the shoes of the client when you create content. The type of search behavior will change. A lot of retailers don’t have that type of content ready in a structured way.

0:15:08 Mike: The crazy thing is that this is what we should have been doing all along. We just relied on consumers doing that labor for themselves once they landed on the website. When people offboard that labor to AI, if you haven’t done it, you’re cooked. We need to wait for machine brains to start optimizing towards human brains.

0:15:42 Chris: And we need to wait for more details from Google directly.

0:15:47 Mike: To recap, there’s no opt-in or opt-out, and no segmented reporting.

0:15:57 Chris: You don’t know actually. The ads are either adjacent or directly embedded in AIOs, and it’s already happening.

0:16:04 Mike: Google will not tell us what the impact is on advertising volume or click-through rate. I think volume is neutral to positive, but click-through rate is probably negative. Right now, we don’t know much about it.

0:16:22 Chris: We know it’s happening and what potential leverage the retailer has. This is the future of search. Our CEO was recently invited to Mountain View and had some great conversations with Google reps. Google is the dominating force with 90% of global search queries, but they are very aware of what OpenAI and Perplexity are doing. OpenAI has made a massive move that will dictate the next move on Google’s side.

0:17:25 Mike: OpenAI has rolled out in-chat checkout, and Shopify and Etsy are integrated as flagship partners. Shopify makes a lot of sense. They’re also launching effectively an App Store, and Walmart and Target have signed on.

0:18:03 Chris: These are huge players. Isn’t it interesting that Target and Walmart are jumping on that boat? This will eat into the marketplace business model, especially for Walmart. What’s the thought process there?

0:18:23 Mike: Long term, it is a threat because those marketplaces serve as an infinite digital shelf. Now that comparison happens in someone else’s place and the checkout happens somewhere else. That transaction data is hugely important. They must see more advantages in cooperating right now than competing.

0:19:19 Chris: Maybe they see the threat and want to take it by the horns. So you can now buy directly in the app?

0:19:58 Mike: Exactly. Right now, it’s still pretty primitive. I think there needs to be a standalone, specialized e-commerce shopping app or a tabbed experience. All that rich media needs to be integrated. For now, you can only buy one product, which isn’t great for basket values.

0:20:48 Chris: But it’s a first step. I’ve been in the business for 15 years, and these feel like the most dynamic times ever. I remember when PLAs and DSAs were launched, but this feels more fundamental.

0:21:19 Mike: I think it’s the biggest thing since smartphones.

0:21:27 Chris: Talking about Walmart, you have some interesting data?

0:21:33 Mike: I saw some figures from September that OpenAI accounted for 32% of Walmart’s referral traffic. That means traffic coming in with a UTM parameter from ChatGPT. That’s up from 21% in August.

0:22:10 Chris: Walmart is so big, though. What’s the overall share?

0:22:17 Mike: I want to give a shout-out to Joe from Marketplace Pulse for writing this up. Is it breaking news or a nothing burger? A bit of both. Referral traffic is only 4% of Walmart’s total traffic because they have so much direct traffic and paid ads. So this boils down to roughly 1% of their total volume. But for other retailers without that massive brand power, referral traffic might be a lot more.

0:23:33 Chris: It’s a sign of things to come. I assume they will enrich this feature so you can buy more than one product soon.

0:23:51 Mike: They have to. Google knows these e-commerce basics and has robust features that could step in, but OpenAI is learning. I could imagine an end state with a universal cart where you buy multiple products from multiple retailers. I’m a skeptic about fully automated shopping agents, but a shopping assistant use case makes a lot of sense.

0:25:02 Chris: It will be interesting to see if this has an impact on Google’s market share. Big move by OpenAI. Mike, time flies. It’s time for the outro.

0:26:03 Mike: This has been another episode of Growing E-Commerce. Thank you for listening. This podcast is brought to you by Smarter Ecommerce. You can learn more at smarter-ecommerce.com. If you enjoyed the podcast, please recommend us to a friend or coworker and give us a five-star review. We’ll see you next time.

0:26:33 Chris: Bye-bye.