AI learns to buy. Customers learn to doubt.
This year, a buy button showed up inside the chatbot. You can now finish an entire purchase without ever leaving the conversation, and the revenue numbers behind it are absolutely real. Except the company that built it has quietly started walking it back, and it turns out almost nobody is actually checking out. This week, I've got two builders who look at that exact gap, the loud launch and the quiet retreat, and read it in completely opposite ways. Hi, and welcome back to the Allowable HQ.
Speaker 1:This is Assistant to the CMO. I'm your host, Lexie Meskouris. And the idea here is pretty simple. Every week we take whatever just happened in AI and marketing and we actually pull it apart. And I don't build any of this stuff myself.
Speaker 1:So each week I sit down with two people who do Will Nash and Harjot Singh. Between them, it's honestly over twenty years of building tech and brands from the ground up. And the best part is they've got no horse in the race. They don't sell the things we cover. They just see them differently.
Speaker 1:And this week, the story is AI moving in on the actual buying. OpenAI put a checkout right inside ChatGPT. The ad money showed up fast. Google's leaning hard into AI doing the advertising. And then in the same breath, the pullback and a pretty loud wave of people saying they don't trust any of it.
Speaker 1:So that's what we're getting into. Will, I'm going to start with you because you're the one who thinks this is a signal. The buy button, the ads, all of it. Real shift or just a good launch?
Speaker 2:Good question. So, yeah, I do think this one's developing into a signal. And we we touched on a version of this a few weeks back, but the thing that's changed is just the sheer scale of it, you know, the revenue ramping up that quickly. That part's hard to wave away. But and this is the bit I'd want a CMO to actually hear, that the real shift isn't really the checkout.
Speaker 2:The interesting question underneath all of it is whether SEO and what people are now calling AEO or GEO, whether those are the same thing or not.
Speaker 1:Hold on. You've gotta give me that one because half my audience just heard a new three letter thing and tensed up, AEO in plain terms.
Speaker 2:Yeah. Fair. So search engine optimization, SEO, that's the the decades old game of getting yourself to rank on Google. AEO is answer engine optimization, which is, you know, how you show up when somebody asks ChatGPT or Claude a question, and it just gives them one answer. And Google's line broadly is that they're the same discipline, you do the same things, and you'll be fine.
Speaker 2:I And think that's starting to look wrong, actually. We're seeing real differences, and they have a a genuine knock on effect for how a marketing team spends.
Speaker 1:But how though? Because if I'm a CMO right now, that's the part I actually feel in my budget.
Speaker 2:Right. So the old game, a big slice of your discovery budget goes on, effectively getting your URL onto the list. You pay to be mentioned on some high trust site. You get the link. You measure the clicks coming back.
Speaker 2:Attribution, honestly, that part's not even that hard. The traffic still lands on your site and you can see it. But the shift is, with the answer engines, it's not really about being on the list anymore. It's about what's actually being said about you, the the semantics of it. So you stop caring quite so much about getting the URL mentioned, and you start caring about the content of how you're described, which pushes the money toward, you know, PR, earned coverage, paid features, that whole world, not clicks.
Speaker 3:Yeah. See, this is why I come in a bit differently, to be honest with you, because I I don't disagree with the discovery stuff, the AEO points real, but the headline this week, the buy button, the checkout inside chat GPT, that part to me is mostly noise right now. Like, genuinely, no one buys like that. People are super nervous buying like that.
Speaker 1:Wait. Noise? They put up 100,000,000 in ad revenue in about six weeks? That doesn't feel like noise from where I'm sitting.
Speaker 3:Yeah. No. And that revenue is real. I'm not saying it isn't, but think about whose interest it is to talk about that number. Right?
Speaker 3:Super, super in chat GPT's interest to push this thing hard so they can stand up and talk about the revenue. That doesn't mean people are actually checking out. You think about how you'd feel, like, actually confident enough to transact, hand over the card, finish the whole thing inside a chat. Most people aren't there. It sounds like a really fun experiment to me, but I I just don't see how it moves the needle for most businesses yet.
Speaker 2:And to be fair, I'm not that far from you on the checkout itself, Harjot. I'd I'd separate the two things. The discovery shift, that's, I think, happening now. The native checkout, that's more of a a get ready thing than a today thing.
Speaker 3:Right. But here's where I think it actually gets sharper. You know? It really depends what you sell. So if you're a Shopify shop and you sell something that's a genuine impulse buy, like, I don't know, you sell dustbin bags, something that's cheap and it's super obvious what it is, a tenner, £20, £30, then, yeah, sure.
Speaker 3:Test it. Why not? But everyone else, you know, anyone selling something with any real value to it or anything bespoke, people are not gonna outsource that final call to the AI. They might use it to learn about your brand, decide you're worth a look, that bit I buy, but then they wanna go and actually see the thing themselves and check out themselves.
Speaker 1:So the AI is allowed to do the homework. It's just not allowed to press buy.
Speaker 3:Yeah. Basically, for most things. Yeah. And the reason the bit the hype skips over is everything after you click. Like, what happens when it gets the order wrong?
Speaker 3:How does the refund work? Who do you talk to? We build these flows. Right? And the friction is always in that post purchase bit.
Speaker 3:Returns, support, just confirming you actually got the right thing. None of that magically gets solved because the buy button moved into a chat window.
Speaker 2:Yeah. And this is where I'd push the other way a little, though, because, you know, I agree it's not for everyone today. But if you're in ecommerce, the thing I'd actually be doing right now is rebuilding your product feed, you know, getting the schema right, making sure the models can actually read your catalog properly, because there's a good chunk of that work that pays off in plain old SEO anyway. But the real prize is as and when the native checkout does come online properly, you're not doing a a massive rebuild under pressure. It's just a switch you flick.
Speaker 3:See, I I hear that and for some sellers, sure, but I'd be careful telling most businesses to go and replumb everything for a buy button that, like we just said, their customers mostly aren't gonna use. There's a real cost to building for the agent before the agent's actually how people shop.
Speaker 1:And there's the actual split, isn't it? Will, you'd build for the agent now and get ready for it? Harjot, you'd build for the human and wait for the agent to earn it?
Speaker 2:Yeah. That's a fair way to put it. Though I would say, even on Harjot's side, the discovery work isn't optional. You can decide the checkout's not for you. That's fine.
Speaker 2:But the the question of what the AI says about you when somebody asks that one you don't get to opt out of.
Speaker 1:Say more on that because that sounds like the bit that actually touches everybody.
Speaker 2:So so here's the thing most people don't realize. There's a fairly known set of sources that the big models lean on really heavily. You know, Reddit's a huge one. Wikipedia, obviously. But also things like g two and weirdly YouTube transcripts, they they over index on those.
Speaker 2:And what that means is if you've got a budget for, say, influencer marketing, the instinct is often to spread it really wide, you know, lots of small micro influencers, And on the Answer Engine side, that gets you almost nothing because the models are so heavily weighted toward the biggest, most trusted sources. So what you actually want, and this is a bit counterintuitive, is concentration. You concentrate your spend on a small number of big trusted targets.
Speaker 1:That's the opposite of what most people get told though. Everyone says go wide.
Speaker 2:It is. Yeah. For this specific channel, and I don't wanna sound too neat about it, but the the models reward the heavyweight sources. So that's where the effort should go.
Speaker 3:Yeah. No. That part, I'm actually with Will on, to be honest with you. The concentration thing is right. It's the same reason the the long tail of tiny mentions just kind of vanishes.
Speaker 3:So we agree there. It's really the checkout where I pump the brakes.
Speaker 1:Good. Because you've each given me one to do list with a buy button on it and one without. Hold that because the next part is the part I really wanted, which is what happened right after the loud launch. So here's the turn. The same company that put the checkout in ChatGPT has quietly started pulling it back.
Speaker 1:The reporting is something like a dozen Shopify merchants actually went live, and now they're shifting toward dedicated retailer apps instead. And around the same time, the trust backlash got loud. A US survey had something like 60% of people saying that just seeing AI in your brand messaging is a turnoff. And a court in Germany held Google directly responsible for what its AI said. So, Will, if even the people who built it can't make the checkout stick, what does that tell you?
Speaker 1:Is this a trust crisis?
Speaker 2:Yeah. And honestly, no. I I don't read it as a crisis. I read it as completely normal for where we are. The the single thing almost everyone gets wrong here is they think the world's already adopted AI, and it just hasn't.
Speaker 2:Not really. I actually looked it up. It's something like 20% of US adults using generative AI on a daily basis, which sounds like a lot, but it means we're still right at the the front of the curve. We're on the way up.
Speaker 1:So when people see the pullback and the survey and they panic, you're saying they're reading the wrong curve.
Speaker 2:Exactly that. And what you've got is really two curves rising at the same time. You've got more and more people just getting used to using AI day to day, and then separately, the people who got there earliest, they start using it for more, you know, including shopping. So a wobble right now, that's just the noise you get at the start of something. It's not the ceiling.
Speaker 2:On this particular one, I genuinely don't think you need to worry too much.
Speaker 1:But don't worry is not a thing I can put on a slide for my boss. What do I actually do with that?
Speaker 2:No. Totally fair. So there's a few bits of, I call it, hygiene that you start now. One, you start treating your US and your EU customers a bit differently because the regulation's genuinely diverging, and I think that only continues. Germany is the obvious example.
Speaker 2:They tend to be much more skeptical about data, and I suspect that rolls into a broader skepticism. And then the second one, which almost nobody tracks, is misinformation about your own brand. You wanna be checking whether the bots are actually saying things about you that aren't true. And then the third, and this one's really, really key, is be clear about AI personas. If a customer's talking to something that looks like a person, you tell them early that it's AI.
Speaker 2:You don't have to stamp every product image you touched with AI. But the moment it feels like a person, disclosure, fast, that that builds trust rather than burning it.
Speaker 1:Though the survey almost pushes you the other way, if 60% flinch at the word AI, the temptation is to hide it.
Speaker 2:It is. And I think hiding it is is the mistake. The flinch is about feeling deceived mostly. So the honesty is actually the thing that calms it down.
Speaker 1:And that's one read on the pullback. But the same story landed completely differently over here. Harjot, your first instinct wasn't trust at all.
Speaker 3:Yeah. 100%. And mine's really simple, to be honest with you. Do not ditch your ecommerce site, like, absolutely not. Your site, that's the thing you optimize.
Speaker 3:Right? It's got your personality. It's got your brand. It's got the feeling somebody gets when they land on it. And all of that, basically, all of it gets stripped away the moment the AI is just serving your product up as a set of facts.
Speaker 3:It doesn't talk the way you talk. It doesn't make people feel anything. So the pullback to me that just confirms it.
Speaker 1:So the walk back isn't the AI failing. It's the AI hitting the bit that was never just facts. Facts.
Speaker 3:Yeah. That's exactly it. You know, maybe in ten years, there's a world where people are so happy with how AI does this that you don't need your own sight. But until then, you keep it. You keep it as your first party thing, and you just you keep watching.
Speaker 2:Yeah. And the the two of those actually sit together quite well, I think. Like, my point's about the trust process and Harjot's is about the brand, the channel you own, they're different problems, but the answer to both is kind of don't hand the part that's actually yours over to the AI just because it can technically hold it.
Speaker 1:And honestly, that's the bit I keep coming back to. Everybody's asking, can the AI do it? And you two are both quietly asking a different question, which is should the thing it's reaching for even leave your hands yet? Okay. Last thing, the bit everyone actually shows up for.
Speaker 1:One thing a marketing leader could actually do on Monday before lunch, no big budget, no replatforming. Will, yours first and keep it small.
Speaker 2:Yeah. So the the smallest cheapest thing, and honestly, everyone should just do this, is go and order yourself, like open up ChatGPT, open Claude, and ask it what it recommends in your category, and ask what it says about you versus your competitors. And you read it properly. You look for where it's wrong or where it's actually negative. And then the important bit, you work out where that's coming from.
Speaker 2:If it's a pile of bad Reddit threads, well, now you know the channel you have to go and deal with. You can't fix what you haven't looked at, and most people have just never looked.
Speaker 1:So my homepage isn't my homepage anymore. The first thing a customer sees is whatever the AI decided to say about me.
Speaker 2:Yeah. Pretty much. Yeah. And you don't control it, but you can absolutely influence it. And step one is just knowing what it currently says.
Speaker 1:And you'd add the guardrail to that, Harjot?
Speaker 3:Yeah. Exactly. You know? So do the audit. That's great.
Speaker 3:Do it. But the discipline bit is don't then panic and go and rebuild your whole business around a buy button. Right? Keep your site. Keep the thing that's actually yours and treat that audit as something you keep doing.
Speaker 3:It's not a one off. It's hygiene, like Will said. You check what the AI says about you the way you check your analytics.
Speaker 1:I love that. So the Monday move is basically go find out what the AI already thinks of you, and then don't lose your nerve and tear down the one thing you actually own. Will, Harjot, that was a really good one. Thank you both. And if you're listening, thank you so much for being here.
Speaker 1:We'll be back every week with something new. Before we go, I have to tell you something. Every voice you've heard today, all three of us, me included, it's all AI. But the opinions are absolutely ours and the judgment behind every word is human. That's the part you can't replace.
Speaker 1:And that's what ALAUDABLE does. We take what you'd write, turn it into a show in your own voice, and we keep the taste where it belongs with a person. And we didn't put a buy button in this episode, and we're not going to pretend a robot can be your brand. We just make the show. So if you wanna hear what that would sound like for your brand, there's a link in the show notes.
Speaker 1:And that's the tea. I'm Lexie Meskouris. This has been Assistant to the CMO, and we will see you next week.
