搜索的终结：AI时代的购物将如何运作

The World Wide Web is unhealthy right now. Most of the things on the internet are crap, and we know that they're crap, but they SEO-optimize crap. So how do you decrapify that?

AI agents can direct people to things if people start their purchase activities there.

Like Google, they kind of are a tax on GDP. Consumer spending is a huge part of GDP. They get a percentage of all that spend because they're charging per click. That tax might just shift elsewhere.

Why is this system so complex? What are the different types of purchases where AI can play a role, and what are we hoping to see as we think about the broader market?

You guys have both been thinking about e-commerce for a while. Alex, why don't you talk about what inspired this piece and how this all germinated for you?

Well, I had started a company called TrialPay a long time ago, and I'd been selling stuff on the internet for a very, very long time—even before the internet. I was trying to think, number 1, what happens to Google, because a lot of people have this on their minds. Is search volume going up or down? My own personal experience is that my search volume is going down, but not for commerce—clearly for everything that is not commerce.

That was one thing, but this company that I started, TrialPay, was one of the biggest affiliates in the world. Affiliate marketing is basically: you send somebody somewhere, and you get a share—a commission—if you send them there. It was the oldest business model on the internet, if you will. It actually predated AdWords and AdSense by a bit. It apparently started, apocryphally, in pornography, because that was the world's oldest business model on the internet: how do you track it? That eventually made its way to commerce.

It's all based on cookies and pixels. You drop a cookie on the person's computer, and then on the confirmation page, you have a little invisible 1-by-1 tracking pixel that reads the cookie. That's how you know to say, "Eric sent me the customer."

What we did at TrialPay—we were one of the biggest ones there—was ask: Is that really going to be what powers this new realm of commerce? And is it even relevant for a lot of things? Impulse buys are huge, and with impulse buys, almost tautologically, you're not going to use AI to tell you to buy something. You shouldn't buy anything that's an impulse buy. You go to the supermarket, and you shouldn't be buying Coca-Cola in the checkout line. They actually charge you more at the checkout line than they do if you just buy Coca-Cola in the Coca-Cola section. All of these things are designed to target your emotions, to get you to buy and spend money that you don't want to. That's not going to be AI.

On the other hand, there are these very, very expensive items. You're researching the heck out of them with AI, but there's no affiliate. How do you then commerce and transact? Number 1 was the ontology of commerce, which was very interesting. Number 2 was this whole affiliate thing: Is it still going to be relevant? Because it seems like that's what ChatGPT and others are getting into. And number 3 was just my own personal behavior. I probably use ChatGPT 3 orders of magnitude more than I use Google now, which is interesting.

What excited you to contribute to this piece, and what did you find most remarkable?

There are a couple of really massive consumer markets, the biggest of which might be online shopping. But we've seen relatively few startups trying to take a crack at that market with AI. Even though, like Alex said, there's a lot more opportunity because you now have these really smart LLMs and agents that can help you make better decisions than you could have made on your own, or even make purchases on your behalf. You would think that would create an opportunity for more people to package these into products that they then offer to consumers.

But we haven't seen a bunch of people doing that yet. Part of this piece was to dive into why this system is so complex, what the different types of purchases are where AI can play a role, and what we're hoping to see as we think about the broader market. We hoped that people who were also working in the space would give us a heads-up and let us know how they were approaching it.

Yeah. And you can observe. I always like to observe first because that's objective, and then predict second. Predicting is hard. I forgot the funny quote about predicting the future, but it's very hard to predict the future. There actually is a lot that can be observed, and that's where I think CamelCamelCamel is the greatest site in the world.

We have no stake in CamelCamelCamel, so this is not self-promotional at all. People use this thing like Google News Alerts for pricing. I gave a talk to the Amazon Prime team recently, and they're very, very aware of it because I think it's actually Amazon's biggest affiliate. People say every day, "I would buy this product if it were priced here. It is currently priced here. Please let me know when it is priced here." What will I do with that information? I'm not just going to do nothing. I'm going to buy it.

The consumer is the agent, and this is a very, very inefficient AI, right? If you could complete the entire circle and say, "No longer give information, but allow for automatic action on information," people will do that because we have observed that behavior today. This is the easiest form of predicting the future ever, because you're really just chronicling the present and saying there's going to be 1 additional appendix to the present, which people would do anyway because they're doing it anyway. They just have an easier way to do it.

I think my version of observing was seeing a couple of viral examples of this. Some were really good because the AI found the product perfectly, and some were hilariously bad because the AI couldn't find the product. Teenage girls started using ChatGPT to upload photos of Lana Del Rey at a concert or Taylor Swift snapped in a street-style photo and asking, "What is this hair braid she's wearing?" or "What is this sweater? I want to find it, and I want to buy it."

When it worked, it worked really well because it often found the product perfectly. It would say, "Hey, this sweater is $5,000. You, as a 19-year-old girl in Missouri, probably aren't going to be buying this. Here are some alternative, maybe less expensive options that look the same that you can buy."

That age demographic tends to be a really early predictor of all sorts of consumer behavior, which is why this was probably going to be happening more and more—from the research side of things all the way to making purchases agentically when prices are right, like Alex was mentioning.

Alex, do you imagine a world where there's dynamic custom pricing to the extreme, where we're looking at the same thing on Amazon, but it charges you more because maybe I'm cheaper than you, or you have more money?

Well, people have tried this a lot. The problem is that this is probably a very smart world from an Econ 101 perspective. How do you capture consumer surplus? Consumer surplus is great for the consumer; it's bad for the producer.

Apparently, Delta is doing this a little bit, or they were trying to do this. There are poor man's versions of this, such as charging you more if you have an iPhone than if you have an Android phone, because iPhones are more expensive. You've basically communicated that your elasticity of demand is different from somebody who has less money.

I think you're probably going to run into regulatory challenges with that. Or certainly, you'll run into very, very high levels of unpopularity with your customer base. But people have tried this. Generally, it's hard to get away with.

Right. Let's reflect on previous platform shifts before getting into this one. Right now, e-commerce is 16% of total retail sales. If we were talking 20 years ago and predicting what percentage of commerce would be e-commerce, we'd probably think it was much higher. Why hasn't that been the case?

It turns out the demand curve is different for immediacy versus non-immediacy. Even though overnight shopping is pretty darn cool, instantaneous, 1-second shopping is: "I need toothpaste right now because I'm going to bed and I want to brush my teeth, and I just ran out. Oh, there's a Walgreens over there. I'm going to go there and buy toothpaste."

Amazon's awesome, but getting the toothpaste at 7 a.m.—I don't have demand for that. That's not part of the demand curve. There's a demand curve for real-time toothpaste. That's part of it.

The other part is just: I'm bored. What do I do today? I know—I will go to the shopping mall. There's the experience of doing that, and that's a little bit more impulsive. But then there are these long-term, considered purchases, or what we talked about in our piece as aspirational purchases.

Maybe I'm going to gawk at that Rolex a little bit more and, oh, I just got my bonus—maybe I'll go buy it. But it's all part of the experience. So I think those are, broadly speaking, the two.

I've seen this. I'm on the board of a company called Wise, and the product is sending money. It turns out that the market for sending money where it is received in real time is just much, much different than the market for sending money where it is received 2 days later, because sometimes the 2-days-later sending thing is just less in demand. And again, Amazon has proven this as shipping has become faster.

Once upon a time, when you go back to the early days of e-commerce, you would get something in 2 weeks. So it's almost not surprising that the curve has just continued to expand. It's almost surprising that it's only—you said 16%. Yeah.

That seems very low. I think it's higher. So I'm not doubting your research on the numbers. Here's why I think it's higher: I think there are a lot of behaviors where people do research online and then purchase in person, especially for big sorts of purchases.

Sometimes I'm like, “Hey, I need a new laptop. I'm going to do all of the research on Reddit, Instagram, or the Apple website, but then I'm going to go into the store and feel like, okay, what is actually the difference in the MacBook Pro's weight versus the MacBook Air's weight?” I think there are a lot of those sorts of things. Buying clothes is another great example.

I live in SF, so a lot of people will just order a ton of clothes, try it all on, and then send a ton of it back because there aren't a lot of big stores near us. But I grew up in Oregon, and there it doesn't make a ton of sense to order a ton of clothes online and send them back. It's inefficient when there are so many clothing stores that are a 5- to 10-minute drive from you.

A lot of people will research where to go, what specific items they're looking for, or what style they're looking to buy online. So I think it might be that 16% are fully transacted online, but even in a lot of those other purchases, there is some sort of online research component.

Well, this is actually the hardest thing related to this topic: attribution. It's the bane of everybody's existence. How do I allocate attribution for Justine's MacBook sale?

The most pervasively corrosive business model on the internet, I think, is this last-click attribution. You allocate 100%—part of it was that I read this post on Reddit that inspired me, part of it was that I saw this really cool ad at the Super Bowl. You could do this piecemeal, which is probably the more accurate way of doing it, but it's not exactly deterministic.

The thing that feels deterministic, which is actually incorrect, is, “Oh, whoever sent me the click last is the one that I should reward with the spoils.” A lot of people who don't understand correlation versus causation fall into this trap. This is the business model that I hate the most in the entire world.

Take Honey, for example. What is that doing? You're already on the webpage, about to purchase, and then it's like, “Do you want a coupon code?” “Why, yes, I do. Why would I not want a coupon code? Ten percent off. Click here.” You click here. What does it do? It redirects you to an affiliate page, puts a cookie on your machine, and redirects you back to the page that you were just on. Then it actually steals that attribution.

What's funny is, if you talk to a lot of the marketing people at these larger e-commerce companies—and Amazon is very smart; that's why they don't do any of this stuff—they're like, “Oh, our best channel by far is Honey. They're growing so much.” Or RetailMeNot—that was the original one. It went public, had a big valuation, and it was just theft.

But again, how do I figure out how to do attribution? This is only going to get more complicated in the AI world. It might be the same thing where Justine researched on Reddit, saw the Super Bowl ad, did all of these things, asked a question on ChatGPT, and then clicked purchase. It's actually incorrect for Apple to say, “Oh, ChatGPT, we owe you the entire credit for that. You drove the purchase.” No, it didn't.

Yeah. It's part of it, but it didn't drive the purchase. Figuring out and disentangling attribution is very, very hard.

Let's reflect back on the category as well. It seems like the big winners have been at the aggregator level—Shopify or, obviously, Amazon—and the individual big brands, like, I don't know, Allbirds or Casper. It seems like they were quick to get a lot of revenue but didn't become durable businesses in the same way. They didn't get better as they scaled. Why don't you reflect a bit on the category in general and why it's played out that way?

Ultimately, if it's a one-and-done transaction, you don't really make the product.

Like, you know, Casper didn't make the mattress, right? There's probably some OEM in China that made the mattress, and they put their little logo on it and called it Casper. Then they're just buying traffic on Google and Facebook. So, actually, Google and Facebook were the real victors there, more so than anybody else. People are like, “Wow, mattresses, that's a really good category. I should do that. I'm going to go to Shenzhen or wherever, slap my logo on it, and undercut them on price.” That's what always happens. You can ameliorate this to a certain extent if at least you have recurring billing. Think about what Dropcam did, if you remember that. It was an e-commerce product, but at least it was attached to a subscription. Now there are 9 billion cameras that all do the exact same thing. That category has arguably gotten worse even as the category has expanded, or as demand has expanded. But at least Google owns Nest, which bought Dropcam, so they probably still make a lot of money on that category. Whereas if I bought a Casper mattress 5 years ago, I'm still sleeping on it, and they have to find new people to sell that mattress to. Meanwhile, the original factory that was making the mattresses is now selling the exact same mattress to 5,000 other manufacturers. It's just not a good business model. So, in general, just being a commodity reseller of products—and I think this is the problem—a lot of people will say, “Oh, Casper is its own mattress. Allbirds is its own shoe.” But they're not normally making these products. There's somebody else making the product. It's almost obvious what happened during Internet 1.0. The long tail of commodity resellers went away because location no longer mattered. A lot of what really drove retail for the longest time was that, in just any town in Oregon, there was this store. You could drive somewhere else, but that was really far away, so of course you were going to go to this store. Now, with the internet, you can go to any store. If there are 5,000 stores that don't make their own products and they all sell the exact same shoe from Nike, that doesn't make sense. You should either go to Nike directly, or you should go to the one store that has the fastest shipping, the best service, or whatever. The long tail of commodity retailers basically started dying. And we saw this play out. The first-party commerce experience is not that much better either. Yeah.

Then they're just buying traffic on Google and Facebook. So, actually, Google and Facebook were the real victors there, more so than anybody else. People are like, “Wow, mattresses, that's a really good category. I should do that. I'm going to go to Shenzhen or wherever, find a factory, slap my logo on it, and undercut them on price.”

That's what always happens. You can ameliorate this to a certain extent if at least you have recurring billing. Think about what Dropcam did, if you remember that. It was an e-commerce product, but at least it was attached to a subscription.

Now there are 9 billion cameras that all do the exact same thing. That category has arguably gotten worse even as the category has expanded, or as demand has expanded. But at least Google owns Nest, which bought Dropcam, so they probably still make a lot of money on that category.

Whereas if I bought a Casper mattress 5 years ago, I'm still sleeping on it, and Casper has to find new people to sell that mattress to. Meanwhile, the original factory that was making the mattresses is now selling the exact same mattress to 5,000 other manufacturers. It's just not a good business model.

Because there's no barrier to entry. And if there's no barrier to entry, that normally doesn't work out great for the one-of-N, where N is quite large, producers of—or nonproducers, but just marketers of—the product. It works out great for the consumer in capitalism, but it doesn't work out great for the individual marketer.

I think, especially with true consumer products—and I would consider a mattress more of a utility product—shoes like Allbirds or makeup are very trend-based. Especially with the internet, nothing stays that hot for that long.

Allbirds is the big shoe one year, and then the next year it's the retro Adidas that everyone's going back to. Now it's the On Running shoes. I was watching the Bama Rush sorority TikToks this year, and every single girl had the On Running shoes, whereas last year they all had the New Balance cool look from Japan.

If you're Allbirds, that's a problem, right? You can't capture all of the trends. You have your one SKU, or multiple SKUs across one style, whereas Shopify and Amazon can ride whatever the trend is and have demand come to those individual SKUs.

I think that's going to be an interesting challenge in the age of AI, too, because you could argue that AI agents can direct people to things if people start their purchase activities there. That could be an opportunity or a challenge for single-SKU retailers. My guess is it'll still end up being more of a positive for the aggregators.

Also, I think it's going to be very hard for AI to, for lack of a better term, inculcate demand.

Yeah.

Which is how do I know that the On shoe is cool? It's like, well, I need to see that Bama. Like—

It's just—

I'll send you some.

I know. I meant, like, I need you—I mean, it's a metaphorical “I,” right? It's like, once I see, “Oh, wow, I should have that, too. I'm in a sorority. I want that shoe.” So, it's very hard for AI to do that.

So, which is why the utility part of it is, “Well, I know what I want. Now buy this for me.” That seems like a no-brainer, because that's a lot of what Google does. I respect the hell out of that company, but they kind of are a tax on GDP.

Yeah.

Right. A lot of GDP happens. A lot of that is commerce, right? A lot of consumer spending is a huge part of GDP. Where do you start that spending journey? With that nice little search box? And then they get a percentage of all that spend because they're charging per click or per impression or per action. So that is somewhat imperiled—like, that tax might just shift elsewhere.

Yeah. Let's flesh out the piece now. Let's get into what are some of the things that are going to be taken away from Google? What are some of the things that are going to stay? I also want to get into the different kinds of consumer spend as it relates to e-commerce. Maybe do you want to start with you?

Yeah. I think Google has been the canonical premium business model forever. They built a better search engine. Everybody knows that started in 1998, and they were the 47th search engine or something. “This isn't going to work.” But it was just so much better because of the way that they linked it.

It really goes back to research, which is kind of like the H-index but for finding things. Everybody, when you search for “bagel,” hyperlinks to this one site. That must have a high PageRank; let's go show that first.

Most of the searches when Google started—because commerce was actually quite nascent on the internet at the time—it was all free. It was all kind of information. I remember using Google when it first came out, and it was, “Ah, this is so much better than HotBot” and all the other things out there. It was all free, non-monetizing.

They eventually basically copied the Overture business model, which Bill Gross came up with. This was an Idealab company that eventually became part of Yahoo. This is why Yahoo ended up owning part of Google, if you know the whole history.

The entire thing that made Google this giant $2 trillion company was AdWords. The cool thing about it is that there are a lot of premium business models where it's like, “I don't want to pay for it.” Here it was premium, but having relevant paid search results alongside organic search made the search better.

If I'm searching for “tennis racket” and somebody hadn't figured out how to PageRank-optimize and everything else, or SEO-optimize, it was very useful for them to be able to show ads there. Those ads wouldn't show up unless people clicked on them, because the relevance was never preordained. It's like, if people click on it, it's relevant. If people don't click on it, it's not relevant.

So Google has always been premium. That's kind of bearing out right now. It's still freemium. You search for lots and lots of things with no intent to buy, but every now and then it's your default behavior: “I wonder about X.” You go to Google. Or sometimes you don't even go to Google; you go to Safari, because Apple makes tens of billions of dollars a year by sending all of those searches to Google.

What is currently happening is they are starting to lose some of the free, but not any of the premium. They're losing some of these informational queries, like, “Who won the Oscar in 1977?” That's not a monetizable query, but that's what you're going to want to know. You're just going to ask ChatGPT.

People are doing this right now. ChatGPT has, I think—what is it?—800 million weekly active users, a huge, huge number. That's what they're doing with it, because they're not buying in ChatGPT. We know that because OpenAI is trying to build commerce. Clearly, they haven't built it yet, so they're not buying directly in there.

But for the premium—the premium part of the premium—that is happening in Google still. How do I know that? Well, I can look at their financials, and the numbers are still going up, but we also know that search volume is actually going down. So what are they losing if they're not losing revenue? They're only potentially starting to lose some of the free searches.

Maybe they're directing some of those to Gemini, but I think that's unlikely. I think right now what's happening is people are using paid-for Google, with no changes at all; they're just going elsewhere with AI for free.

I think part of why that's probably been happening is that all LLMs—but I'll use ChatGPT as the example, because most people use it—had this really unfortunate and annoying problem of hallucinating around product recommendations that basically everyone experienced if you tried to use it for that.

I think you have this grand idea of, “Okay, I want to buy a pair of leggings, and I'm going to search on Google or Amazon. I'm going to get the highest-ranked pages, but what I really want to know is: I'm doing this specific type of hiking, and this is what the weather is going to be like, and I want to know specifically for my needs what is the best legging. It might not be the best overall legging, right?”

A lot of people—especially young women—tried this and thought, “Great, I'll go to ChatGPT. It can take all my information in natural language. It can make a recommendation. It can spit out products.” Then they would find that a lot of the products it recommended did not exist, or previously existed but did not exist in a current form, or that the amount they were charging was way different from what it said.

That drove a lot of people who experimented with it back to, “I'm going to return my searches to Google or Amazon and wait until ChatGPT figures out this commerce thing.” My take is that people are probably going to return, because, as we know, OpenAI is working on commerce. They're trying to integrate it more into the experience and have actually real, relevant, up-to-date information on products.

Google will probably be at risk of losing some queries, but I totally agree with Alex that we have not really seen that behavior at any sort of scale today. Well, I would say the biggest problem right now for the internet at large is that it's unhealthy.

I remember talking to John Lily, who was the original CEO of Firefox back in the day and kind of an early web stalwart. The internet—or the World Wide Web—is unhealthy right now. The reason why it's unhealthy is because so many things that used to be the open internet, when it really was DARPAnet, then ARPANET, then this internet thing that people started using—but only really researchers and stuff like that—were all just on the open web. There was no concept of a walled garden.

I mean, search has already been fractured, by the way. It didn't happen with ChatGPT. If you want real-time search, you go to Twitter or X.

Right? You want search for your friends, you go to Facebook. None of that is on Google; you can't search Google for stuff that's happening among your friend group. That's walled off there. So you have all of these different walled gardens. That's unhealthy part number 1.

Unhealthy part number 2 is just the commercialization of the internet, which is not bad. I'm a capitalist. I like commercialization. But so much of it—if you look for “What is the best sneaker?”—who are the people writing content about great sneakers?

In 1995, if you had a blog, you just hosted it on your own site. You set up Apache on your own server that you racked yourself, and then you did it for the love of the game. Affiliate links provided the monetization model, but they really polluted the internet that was still open, because so much of it is like, “Top 10.” A lot of these top-10 sites are out there. “Top 10 running shoes”—you know what that is? That's top-10 affiliate revenue to me. I pay somebody in India to write garbage and then SEO the heck out of that to make money.

Contrast this with the pre-internet era, where there was a publication still around today called Consumer Reports. The really cool thing about Consumer Reports is that it was the only publication that refused to take advertising. It was entirely subscription-based. The idea was that you could trust the actual reviews.

They would do things like— they were the Ralph Nater of consumer products—“This thing is terrible. Don't buy this blender, or we'll chop off your finger,” or, “Do buy this thing.” They would really review everything. We kind of need that, and that entire business model just went away.

Craigslist killed almost all of traditional media. Maybe they deserved to die; maybe they didn't. But they made money from 2 things. They made money because they had a monopoly on information.

They charge for ads there, but a big part of the monetization model was the local classifieds. All of that went away, which is why newspapers have been dying. You could imagine a newspaper having a do-gooder thing where it’s like, “Oh, let’s review all the blenders,” and obviously not show blenders that cut off your fingers. That’s bad. That whole thing went away.

The summarization of the open internet is tough because there’s less open internet than there used to be as a percentage of all the content being generated. A lot of it is walled off, and the stuff that isn’t walled off is just pervaded by junk. What we talk about in the piece is that you can’t turn shill junk into honest analysis.

Mhm.

So, I don’t know how we solve that. No matter how good it is—no more hallucinations, everything is awesome—

But most of the things on the internet are crap. They’re crap, and we know that they’re crap, but they’re SEO-optimized crap designed to earn affiliate commissions, and summarizing that crap is not helpful. So how do you decrapify that? That’s quite challenging.

I think, honestly, what I’ve seen in terms of the channels where you see the least crap is actually video. As a creator—if you’re a creator now—due to the death of traditional media, there are now creators who go out and review 10 different running shoes. They’ll specifically make it very clear in their video: either this is sponsored by this specific brand or, obviously, the better ones are completely nonsponsored, but they get ad revenue from Google, from YouTube, from people watching.

When I want an honest review—someone who has looked at 5 different blow dryers for this sort of hair—I will go to an unsponsored YouTube video. Those often have a lot of views because there are a lot of people having similar queries. But my sense is that Google doesn’t treat that information the same way because it’s video, it’s not skimmable, and they’re not automatically generating transcripts for every video. So that information does not appear in traditional search.

Right? And I think now we’re starting to see some companies say, “Hey, look, we should turn all of those high-quality videos into transcripts that an LLM can then read and review and use to make recommendations.” But that doesn’t seem to have hit the traditional Google part of the internet yet.

Yep, I agree with that.

Yeah. The New York Times recently bought Wirecutter, which may be an example of what you’re talking about.

Well, but I think that’s yes and no. Everything’s an affiliate link. Is it really true? It’s so suspicious that almost every item they recommend always has an affiliate link. Isn’t that odd?

Yeah.

Right. Does that mean it’s a sampling-bias thing? Is that true?

So I’m quite skeptical of a lot of these things. Again, the Consumer Reports era was just a little bit different. You might have biases—maybe the person reviewing things for Consumer Reports just hates this one company and is taking out their bias on it. This is always possible, but you would think that, with all these algorithms and everything else, if you could get truly objective feedback, this would be fantastic.

Amazon actually is a giant search engine, and that thing is polluted to crap as well, because what a lot of the sellers on Amazon do is go on a site called AliExpress. AliExpress has changed a little bit with tariffs, but they’ll buy 400 of some gizmo that shows up 6 weeks later for $2 each. They’ll slap their logo on it and sell it for $25.

That actually goes back to this latency point that I was making before: How many people want something 6 weeks from now versus how many people want something tomorrow? A lot of what Amazon was doing was just arbitraging that.

If you search for an item, particularly in consumer electronics—I remember I was looking for heated socks for skiing. It turns out they’re very useful. There are 9,000 different pairs of heated socks, and they all have the same OEM, the original equipment manufacturer. They all have bogus reviews.

Amazon should fix this, but they have no incentive to do so. I used to sell a rock on Amazon. I get five-star rock reviews. Now I switch the SKU from rock to heated socks, and I trade off my five-star review. How does Amazon allow that? Again, Amazon just wants to sell more crap.

So they’re totally fine with this. But, like most things, if you’re willing to wait, you’re so much better off buying on AliExpress than on Amazon.

It’s just this polluted sea of crap. My favorite business model for commerce, by far, is Costco. I think Costco is the greatest company in the world because Costco refuses to sell bad things. They refuse to take a high gross margin.

Why would they refuse to take a high gross margin? That doesn’t make any sense. Why would they refuse to take a high gross margin? You know why? Do you know why?

Because they want to pass that back to customers?

No. It’s because it degrades the value of the membership.

They make money from the membership. So they’ll charge you something like $100 a year to join Costco. If you look at their net income, it’s basically the number of memberships that they have—50-plus million members—times the price of the membership. That’s their net income. Everything else just kind of is a wash.

If you’re making a 50% gross margin on a shirt, they’re like, “That’s too much. You’re fired. You can’t make that much money.” It devalues the membership.

They’ll do crazy things. The hot dog is still $150. They started their own chicken farm because the costs of the rotisserie chicken were going too high. That’s how they run the business, and they refuse to sell anything that they are not proud of.

The generic brand is just as good: Kirkland wine, Kirkland beer, Kirkland shirts. They’re getting sued by Lululemon right now because they made pants that were better than Lululemon’s pants, but much, much cheaper—and they’re actually much better.

So Costco is the greatest thing. That’s why, for everything we talked about in commerce—pre-internet, internet, AI—Costco is immune to all of this. They’re like Consumer Reports. They treat customers incredibly well, and that’s why this company is worth hundreds of billions of dollars too.

People really trust them. My mom has been a Costco member forever, and now she gets her glasses at Costco. Every time I want to go get flights or something, she’s like, “Log in and use the Costco thing,” because she always thinks that Costco is going to have the best option at the best price. And she’s usually right.

That is sacrosanct to them. They refuse to violate that because they can make so much more money if they decided to. Normally, I remember there’s a speech—

—by Jeff Bezos where he talks about this: There are 2 business models.

There’s the model: What’s the most that we could get away with in terms of charging? That’s Apple, right? “Oh, let’s charge $1,600 for this iPhone 25 that we’re going to come out with that has 18 cameras. How about—can we even get away with $1,700?” They have very high gross margins.

There are other companies: How do we charge the least amount possible? That’s kind of Amazon. Let’s have this sea of crap, but whatever. Why would we curate the crap? That’s up to the consumer. We’ll have the reviews and everything else, but they don’t do a great job on the reviews.

And those are the extremes, right? Android and Apple. There are so many examples of this. There’s the premier provider—the Mercedes, the Ferrari, whatever—and they just want to show high-end stuff. Then there’s the mass-produced, low-end stuff.

And then there’s this very, very unique business model that’s very hard to replicate, called Costco, because normally it doesn’t work. It’s like, “Hey, just trust me because I’m the best.” You have to have many, many decades of trust, such that Justine’s mom is like, “I don’t know what it is, but if it’s sold at Costco, it’s good.”

And if you were a CEO of Costco, would you further leverage that trust to do other things, or would that risk the whole enterprise?

I think it risks the whole enterprise, but there’s a lot that they can do. It’s funny—we’ve met with the Costco board, and one of my partners pitched them on financial services. Everybody’s trying to rip you off, right? Every bank is trying to overcharge you for loans or underpay you on your deposits. The Costco loan would just be the cheapest possible.

They’re trying to make no money on that because they make money on the membership. So they probably could expand it quite a bit. But yes, it’s hard. There’s some modernization that they could probably do because it still is this very warehouse thing that closes at 5:00 p.m., and I wish it were open later. Their shipping isn’t great if you want to order stuff, but it is a very, very unique business model that will stand the test of time, and I think it’s AI-proof.

Yeah. Justine, why don’t we get into other ways in which AI will change commerce? You outline a few different types of purchases that might get in.

Yeah. So, we looked at the range of purchases from impulse buys—which I think still are the Coke thing, the Coke bottle on the aisle—but often now, for a lot of people, they’re the TikTok Shop thing, where you’re watching a video and something shows up and you’re like, “That T-shirt looks cool. I’m going to buy it.” All the way to really considered purchases, like a house, a wedding venue, or a car, where you’re spending a significant chunk of your income. It’s a one-time or a multi-time thing, and you’re doing a lot of research.

I think both ends of the spectrum are harder for AI to disrupt. I think the impulse buy is harder because there’s no research in advance, and you’re not going anywhere specific to buy it. By its nature, you are making the decision to buy it immediately when you see it. Algorithms will get better and better at targeting you with the shirt that shows up on your TikTok feed and somehow has your dog’s name on it, and you’re going to buy that more than the other thing. But that’s not the generative AI that we’re talking about.

Then, at the most-considered end of the purchase spectrum, I think it’s hard to have that be fully AI, end to end. While you may start doing your research online on ChatGPT, Gemini, or any sort of new AI-native property that shows up, the purchase is so significant that you’re probably going to want to have some sort of in-person experience where you’re seeing the thing, touching the thing, experiencing the thing, and talking to another human expert about it, right?

That means there’s this whole range of products in the middle where we believe purchasing behavior could be disrupted by AI in a couple of different ways. One is obviously the research aspect. I’m trying to find the best handbag. My handbag wore out, and I bring it when I travel all the time. I need the best one that fits a laptop, can fit a big water bottle, can fit in the overhead compartment of a plane, and so on.

If you’re busy, you don’t have a ton of time to do that research yourself. You might ask an AI agent that can watch all the TikToks for you, read all of the Reddit posts, pull in the real consumer feedback, and then make a recommendation. You might want to do some of your own clicking through to look at options, but I would say that in that case, it’s decently likely that, if there’s then a good integration for purchasing, you might do it through an AI agent.

There are also just things you already know you want, as Alex has mentioned, where you want the best price. I think AI agents can do a lot around price optimization. If you always buy a specific type of laundry detergent, an agent can find across the internet where that laundry detergent is best priced. It can also probably know, “Hey, I should scan this daily. If it’s 30% less on this specific site than it usually is anywhere else, and it’s going to arrive in a reasonable amount of time, I should probably just buy this.” You can store an extra box of laundry detergent because, based on what the agent knows about the consumer, that’s worth it to them.

As you move up the consideration stack, there’s another sort of purchase that I think will be AI-intermediated but may have some human impact. Things like bikes or couches—slightly higher-value purchases—or laptops, where you want to feel like someone has taken the time to really understand all of your criteria and help you make the best decision about what you should buy.

This is probably an item that you’re going to be using for years, and it’s important to you that it works, that it’s the best option, and that it doesn’t immediately become obsolete. Today, I think the only way this has happened is that people will go super deep into Reddit threads on the Buy It for Life forum and all of these different places, or they have a brand they really trust, like Apple, and they’re willing to pay the premium.

In the future, I think it’s fun to think about having an AI agent that really deeply understands you, where you can have a more in-depth conversation about that sort of thing. Maybe even a phone call where it’s asking you a bunch of questions dynamically, back and forth, and you’re providing it with the information it needs to go back, do the research, and decide. So, those are some of the things we’ve considered about how AI could impact purchase behavior.

Well, there’s another kind of lens. There are many different ways of cutting this, but does the product that you’re buying have a UPC or not? If you know what a UPC is—Universal Product Code—that’s the little scannable thing. It’s the successor to the ISBN, which is for books.

If it doesn’t have a UPC, actually, a lot of the commerce that has worked post-Internet 1.0—how did Wayfair work? Why did Wayfair work? Well, they’re selling things like bar stools. You’re like, “I want a bar stool,” but there’s no UPC on this. So it’s like, “Well, here’s a bar stool. It fits the right dimensions, but there’s no scannable code.”

If there is a UPC, you can run this little algorithm: “Get me the lowest price.” Pre-AI, you would just run this algorithm on your own, and you’d probably end up at Amazon. So everybody who wasn’t Amazon just got killed, and Amazon did well. I’m oversimplifying a little bit.

If it doesn’t have a UPC, then that’s a different process than if it does. If it has a UPC, then the algorithm that I was describing is exponentially better with AI, because before, you’d have this situation where some people value time more than money and some people value money more than time.

If I value money more than time, I am the algorithm. I will spend the time. I need to find the best coupon, the best cashback site—all of these cashback sites are out there on the internet. Lots of people who value money more than time do this.

All of that will be automated, or automated for the benefit of the consumer, if and only if you have something where you have determined the SKU or the UPC. You know, the SKU has a UPC; you feed it in there, and it’s good. If it doesn’t have that, then that’s another lens where it’s like, okay, I probably can’t feed whatever it is into that system.

Again, from impulse to highly considered, AI is going to help you on the highly considered side but not help you buy it. But if it spits out something with a UPC or a SKU, then this part of the AI will just automate that.

Right.

Right. So if it’s a bike, sure, I don’t know what bike to buy. But if it’s a Specialized bike—and here’s the thing: it has a UPC on it—boom. Why wouldn’t you feed it into this part of the system as it gets developed? That’s just going to buy it for you with the best shipping, the best terms, the best whatever.

Right now, the reason why that doesn’t happen—or it does happen, but it happens manually—is that people for whom time is valued more than money don’t do any of that stuff.

Putting this all together, we were talking about how, over the last decade, there haven’t been a ton of net-new, big winners in the space, and a lot of the gains have gone to the aggregators. Why do we believe that, over the next decade, there are opportunities for net-new, big, durable companies? Maybe share what types of companies could exist that we’re excited about.

Obviously, ChatGPT is an upstart to a certain extent, but it’s not Amazon, it’s not Shopify. It’s a net-new company that clearly will have a role in commerce, right?

The question is: Will there be specialized subsegments? And I do think that hyper-optimized companies—I know you very well—like CamelCamelCamel can be independent companies that are probably very, very profitable, as far as I know. They’ve never raised venture capital.

A lot of people use them, and a lot of these cashback sites are the kinds of things that have always been for people who value money more than time. There was a site called Ebates that was bought by Rockutan a while ago. There’s a company in the U.K. called Quidco, which is very, very similar.

You can imagine these kinds of things going much more mainstream and becoming very specialized shopping agents, particularly not on the heavy-research side but on this one little tiny vector that might actually be very big.

Going back to how most companies can’t really figure out attribution, it’s like, “We are going to be the last click.” The last click of the 21st century, post-AI, is going to be AI companies that know how to do this. It might not be ChatGPT, because they’re this horizontal-everything kind of company.

It’s like, “I’m going to give you all of my credit cards.” You’re actually going to figure out which credit card to use for this particular purchase, because this one has higher cashback than that one for this type of good. You’re going to integrate affiliate tracking, where you give me cashback like Ebates or Raku 10en does. You’re going to do all the coupon stuff.

Not all of this will be good for merchants, by the way, but you can imagine it. It doesn’t require a lot to imagine this, because there are already a lot of companies that do it. They’ve been somewhat of a niche space because they only appeal to people—and there are plenty of people who are like this, by the way—who value money more than time.

And they're somewhat technical. So it's actually not just money more than time. My mom might want to use one of these. She would probably value money more than time because she's retired, so why not? But it's just too complicated to use.

If you make it so easy to use, that's the other thing. We talked about how if you make something show up right now, that's going to have a bigger market than if something shows up 5 weeks from now. That makes sense. If you make something so painfully easy to use that it's more of an IQ test, it's like: Do you want to pay less for something or more for something? Everybody, of course, would say, “I want to pay less for something.” But then it's, “Oh, you have to do these 18 things.” That's too complicated. I can't figure out how to do that.

But if it's so easy, I think that's one area where startups have lived, because it's clearly not going to be Amazon. Amazon wants you to shop at Amazon. Amazon, by the way, is somewhat imperiled because it has a giant revenue and profit line item from advertising. You go to the Amazon website and then you click on an ad that takes you away from the Amazon website. That's a 100% gross margin for Amazon. They'd rather sell you that than sell you a product where they have to deliver it, God forbid.

So the best SKU that they sell is the advertising SKU. And that's going to be imperiled if they no longer control the presentation layer, because AI intermediates it. But I think it's this kind of money-more-than-time expansion to the entire universe. There's certainly a there there.

Yeah, I think there's sort of 2. I think there's the consumer side, right? If we go back to my conversation earlier about wanting to have a really in-depth conversation about what bike to get, you could imagine someone fine-tuning a model that is much better, based on tons of conversations between bike experts and people who actually know the right questions to ask, to give you a much better buying experience and a better outcome than ChatGPT could. So that's 1 way that consumer distribution could be disrupted beyond the ChatGPT disruption that will happen.

Then I think there's the merchant side of things. What are the implications if we suddenly have a ton of AI agents browsing your site and potentially even making decisions on behalf of consumers and hitting the purchase button instead of people? How should websites change to make themselves more browsable, easier to interact with, and easier to find what the agent is looking for? What sort of infrastructure do we need on the financial side for AI agents to actually be able to make a purchase on behalf of someone and use their credit card? The entire infrastructure and merchant-facing side of it is probably going to change quite a bit, and I think that will be just as big as the consumer side of the market.

I think it's a good place to wrap. Alex, Justine, thanks so much for the great conversation.

Thank you.

Thanks for having us.