---
title: "Podcast: Will consumers adopt agentic commerce? (with Michael Morton)"
url: https://adtech.lol/articles/9bb07e9d-c22e-40fc-a53e-1f4f307c12cb
source_url: https://mdm.co/podcast-will-consumers-adopt-agentic-commerce-with-michael-morton/
source_site: mobiledevmemo
author: "Eric Seufert"
published: 2023-10-01T00:00:00+00:00
tags: ["agentic commerce", "e-commerce", "consumer behavior", "AI", "digital ads"]
publisher: AdTech Radar
---

# Podcast: Will consumers adopt agentic commerce? (with Michael Morton)

> Discussing the limitations on consumer adoption of agentic commerce with expert Michael Morton.

## Transcript

**Eric Seufert:** Welcome to the MDM podcast. I am your host, Eric Seufert, and I am joined today by **Michael Morton**. Michael, good to see you. How are you doing?

**Michael Morton:** Wonderful, Eric. Thanks for having me.

**ES:** On the occasion of our podcast recording, I get to wear my 1963 Seamaster because I thought you would appreciate it.

**MM:** I like that a lot. I have fallen down the vintage Omega hole in a bad way.

**ES:** It puts a hole in your bank account, but this watch was very meaningful to me. It was the watch I bought for my wedding. The reason I wanted to buy it is that it is one of the watches that Don Draper wears in Mad Men. It is a really beautiful piece, although it is a little too small for my wrist.

**MM:** Vintage watches are all smaller. In the newer era, they made the watches larger. I need to use the special event occasion to pitch to my wife why I should buy more watches. People buy them around their kids’ births or weddings. I need to do this more often. I am wearing my faithful Speedmaster today.

**ES:** You cannot go wrong with the Speedy. I have a Speedy as well.

**MM:** You are new to the podcast. We have spoken a number of times and met in person at the event you hosted in Boston. Please introduce yourself to the audience.

**MM:** I am a publishing sell-side analyst. I cover consumer internet marketplaces. Michael Nathanson and I are doing digital ads together under the holistic umbrella of consumer internet.

I fell in love with the aspect of writing for public consumption and debate. Something about my personality likes debating with people. I worked at a firm called Bernstein Research, then I worked at YouTube on the strategy team for a brief period until Michael Nathanson called me. I have admired Craig Moffett and Michael Nathanson’s firm and research for my entire career. When Michael Nathanson calls and gives you the opportunity to write research with him, it is the easiest career decision one could make. That was April 2022 that I left YouTube to join, and we are approaching the four-year mark of my initiation covering marketplaces and consumer internet.

**ES:** It is a dramatically different time than when I initiated at the end of 2022. We were on a COVID hangover for e-commerce, and there was no such thing as consumer-facing AI. Now, this is unquestionably the fastest pace of change I have seen in my entire career.

**ES:** Commonly, when I say agentic commerce, I mean third-party, independent agentic commerce, not on-platform agentic commerce. As of today, I am also carving out travel. Really, what I mean is agentic e-commerce. Am I wrong about agentic commerce being a mirage?

**MM:** First, I have to respect that title. As someone who writes research titles for a living, they matter more than I think people would expect. I think that was a great title for the subject matter. It is also important here just to be a little nuanced about when we are talking about agentic commerce and what we are discussing, because I can see myself being dragged on X as a Luddite unless I am clear about it.

We think AI is dramatically changing the product discovery funnel already. I think it is going to compress the funnel and help consumers find products faster in a more efficient manner. This is going to have big implications across the e-commerce and ads ecosystem. Where I have found myself disagreeing with the agentic bulls, and where I find myself in your camp of the mirage aspect, is that I do not think consumers want third-party autonomous agents buying things for them.

The way that we tried to approach it in a recent report we published avoided the technological debate, because it is easy for someone to dismiss that right now. If I try to highlight some product shortcomings, the quick response in technology is always that today is the worst it is going to be. That is a very reasonable answer and has historically always been true, so do not bet on tech standing still.

What we wanted to do, and where I find myself generally agreeing with you, was to think about how the average consumer behaves on a monthly basis with their purchasing activities. I love taking shots at Wall Street people and model monkeys myself, and Silicon Valley too. I think in my career, I cannot recall an example of where Silicon Valley and Wall Street—the people building and investing—are extrapolating their lives and their personalities and viewpoints onto the general population in such a mismatch as ever before.

Beyond me saying that the average person does not live like you, let’s go look at the data. The Federal Reserve once a year publishes this data about consumer payment patterns. The average US consumer has 47 payments a month. Ten of those are on bills, and most bills are automated. I do not think that is addressable. Ten go to prepared food, and we can have the conversation today if you are going to tell your agent to order pizza or if you are going to go on DoorDash. So ten on bills, ten on prepared food, six on gas stations and transfers to your friends.

You are left with 21 payments. Let’s just say 20 to keep it easy. If we want to be really generous and say that half of those are e-commerce penetrated, even though we know e-commerce penetration is not that high, you are left with ten purchases that the average consumer makes per month. The point I am trying to make is that the average consumer is not overwhelmed with this crushing burden of purchasing decisions that the Silicon Valley tech bro or the Wall Street dude think. It is just not what the average person experiences on an average day.

On an average day, they spend close to an hour on social media and two and a half hours on television. The average Whatnot user spends 95 minutes watching live commerce. I am skipping the technological debate and trying to think: is this what people really want in their life? Is this what they are really trying to have optimized for them? I don’t see it.

**ES:** That is such a good point. A very good friend of mine just bought a place in San Francisco, and it was like Last Saipan in Vietnam. He got this place right before OpenAI did a big secondary and Stripe did a big secondary. He is constantly sending me these listings and telling me how far over asking they went. He sent me this house that looked like a house; a free-standing house is already quite rare in San Francisco. It was in the Inner Richmond, which is a decent area, but it went for $15 million, and it was $7 million over asking.

This is play money. If you work for one of these big labs, this is play money. The USD is play money. There is this extrapolation of the concerns of a very isolated, peculiar group of people being made to commerce broadly that just defies logic and reason. It is very frustrating because I have chosen this hill to die on. I engage in these discussions on Twitter, and I just want to ask people where they are drawing their frame of reference.

To your point, e-commerce penetration per the Census Bureau is around 16%. You are already talking about a fairly small pool. Is this a problem for people to manage with an agent through automation?

**MM:** Intuitively, it feels that way to me out of the gate. When I look at the data, it is even more compelling. To be fair, two years ago, we started writing the research when people were not talking about it. The silverback gorillas of the investment community, the old-school guys who have seen a lot of cycles, immediately came to me and said that directionally I was correct, but the consumer moves way slower than this. Retail behavior in particular is really sticky.

I can empirically show you that product discovery is better, multiple times better, than fumbling around through traditional search and the sponsored product carousel at the top of Google. That is inevitable. If you look at the percentage of AI-sourced traffic sent to the big e-commerce websites today, it is still less than half a percent. We are years into this now. Industry participants have told me that number is undercounted today because it is not tagged appropriately. Even if it is multiples bigger than that, it is still smaller than I would have expected it to be two to three years ago.

**ES:** I made this point in a podcast I did a couple of weeks ago. We are four years into ChatGPT. Are we really that early? Look at the user base numbers. If you slightly relax the definition of a chatbot to include AI overviews, you have two billion monthly users. We are talking about a billion weekly actives for ChatGPT. Even if on a timeline standard we are very early, the adoption curve has been much steeper. A significant proportion of humanity uses these things, and we haven’t seen them fundamentally deteriorating any of the core pillars of the economy.

Yes, there are challenges in early-career stage hiring that are attributable to AI, but what we do know is the economy hasn’t utterly collapsed. People are saying we don’t know anything yet, it is too early. No, we do know that four years in, the economy hasn’t actually materially deteriorated and is holding up quite robustly. Shouldn’t we update? You can’t just keep changing the timeline. Four years ago, I said in ten years, but now I am saying in ten years it will be this fiery hellscape. I think you need to update your priors here.

**MM:** Sam Altman did a recent interview where he said he was surprised by the inertia in the economy. I was also surprised by the inertia in consumer behavior on this specific new tech change. I study consumer shopping behavior for a living. There is more inertia in this than we all expected, unquestionably.

**ES:** Let’s anchor to agentic commerce here. When I was making the point about the economy, we know what the penetration has been and it hasn’t been disastrous for the economy. We know what the penetration has been and we haven’t really seen widespread proliferation of autonomous buying. Yes, there is uncertainty about the future, but we know where we are at right now. If we had a prediction two years ago and that didn’t track, then we can update that prediction. We don’t just get to borrow that prediction and then shift it two years forward.

What do we know about what OpenAI has done? They went through the process. I feel like I am in Groundhog Day sometimes because people are just pretending like this has never happened, like this is utterly, entirely novel and new. It is not. OpenAI tried it, and where did they land? Ads. Talk to me about what we can learn from OpenAI and from what they have already done regarding instant checkout. Is there a generalizable lesson from that?

**MM:** It is funny because a lot of us market participants, especially my friends who are other publishing sell-side analysts, were all pulling our hair out a few weeks ago when this happened again. In fairness, Muse is different; this is the best type of personal agent product we have seen. It syncs nearly as well, so there have been some impressive developments, but we have done this fear and bear case before.

I could just rewrite the front page of some of my notes I published a year ago and republish them, and it will still apply. I think there are a couple of things we can learn from the OpenAI ChatGPT checkout. The first is that focus is critical. At that point, OpenAI was trying to do too many things at once. The consumer is unforgiving, and you don’t get a lot of bites at the apple if you deliver a bad experience.

When checkout was active, I spoke to industry participants who were in publicly disclosed partnerships with OpenAI for checkout, and OpenAI was the bottleneck. They had merchants and offerings ready to be plugged into checkout, and it wasn’t happening because of some restraints at OpenAI. If you are going to do it, you have to be singularly focused on it.

E-commerce is really hard. We seem to do this every couple of years where a big tech platform decides they want to do e-commerce, and then we are all reminded there is a reason there is only a handful of gigantically scaled e-commerce players. Fifteen percent represent over 70% of US e-commerce GMV. It is super concentrated when you think about it because it is really hard.

The second lesson for me is that the consumer is smart. The consumer knows they are not seeing Amazon product listings on there. They are not seeing 40% of the US e-commerce market and inventory, and they know that. If you are selling e-commerce in the US and there are 100 million Prime members, everybody you are serving these answers to has a Prime number. The launch wasn’t as well as it could have been done.

Also, it is a very hard experience to get right. There are some things that you need; you need creative images and ratings. This is a thing I am going to talk about that I worry the marketplaces are undervaluing, but there is a mountain of data that shows reviews and ratings have a correlation with conversion. When you are screwing around right now on Instacart and Muse and you just see products without ratings, I take a lot of faith when I see 5,000 people have rated a product 4.7 stars. It saves me the time. If it was bad, it would have shown up in the numbers by now. So they fell short on several aspects of what it takes to launch a successful e-commerce full-funnel experience.

**ES:** That is totally fair. I agree wholeheartedly that what instant checkout offered is fundamentally different from what Muse and Instacart are offering. I still believe we are going to end up with ads in those products for a number of reasons. They still did end up with ads and they still did shut it down. They clearly took an opinionated stance on where they thought the revenue opportunity was. That is the question that always changes the trajectory of these conversations.

Amazon has 100 million Prime subs, but a lot of the value of that company is captured by that identity that they have at the household level. They said at Amazon Unboxed last year they have deterministic identity on 90% of US households. They essentially can identify your household no matter where you show up on the internet through their DSP. You have seen them partake in these identity partnerships with Disney+, Roku, and Tubi. They are essentially the identity backbone of FAST and of the open internet. There is just immense value in that.

If they take this hard stance, they may change their mind. They may see a different value calculus and say they are going to open up to the agents and pivot. My sense is they can’t, because if they hold the line, the agents have to bend to their will. Amazon has the products and the identity, and they can insert better recommendations that are built around all that data they have around historical reviews, ratings, and returns, and make the recommendations in the agents better. That is what everyone wants. They want more relevant recommendations. That works in everyone’s interests. The idea that you can somehow circumvent Amazon and still win and still provide a functional, useful product to consumers that can do e-commerce, I think is just premised on flawed logic.

**MM:** Another thing I forgot to add to why it failed is fulfillment by Amazon. Another thing that I believe people don’t fully appreciate about why e-commerce is so difficult is that scraping is a bad experience for e-commerce. When ChatGPT did their initial shopping journey, there was a lot of raw web scraping, and inventory availability, pricing, and shipping speed change a lot. You need API feeds, which Shopify offers. There are other players that offer these API feeds now, and it has gotten a lot better since then.

I spent more time than I’d ever care to admit going through purchase flows on ChatGPT when they were doing this, and you were getting a lot of mispriced products and out-of-stock products because it was more of a brute force scraping than API feeds, which leads to a bad experience. To your Amazon point, the market now more than ever has always known FBA is a huge moat. I could offer you billions of dollars and you couldn’t dislodge Amazon because of this physical layer. Now it is critical because what it is going to do if you have a physical fulfillment aspect to your business is give you leverage with merchants.

Even if this world does change and Muse gets to 20% of e-commerce, Amazon has to rethink their strategy and say they don’t control consumer demand as much as they thought. This means fewer visitors to their website and less onsite ads. How do they deal with this 20% that then is going to flow through agents? Well, what they are going to be able to do is look at their merchant base and say if you want to be part of our API feed to these AI platforms, it is a 3% take. I don’t know what the fee will be, obviously, but they can offset the decline in ad spend they need. If you are a merchant on Amazon, how else are you going to get your goods to the person same day or next day? There is no other channel to the consumer in this fashion, and Amazon has all the leverage here.

**ES:** I think when people argue in favor of total consumer adoption of agentic commerce, it is only out of ignorance of Amazon’s scale and business model. There was a tweet going around yesterday that I responded to that this is the time where a startup could disrupt Amazon. That just seems, if you understand Amazon’s business, deluded on its face. I responded to it and said for all these reasons, who does fulfillment? Amazon has no incentive to do that if you take away the profit engine.

Think about that pitch. You are a VC and I am coming to you and pitching you that agentic commerce is the future. I am going to build an Amazon that is agent-first. Here is what it looks like. Amazon was founded in ’94, so 30 years into the future, it is Amazon but I rip out AWS and I rip out ads. You say wait a minute, this is going to take an incredible amount of investment to get to that point in 30 years, and there is essentially no net income. Why would I invest in that? No VC would invest in that if that is the vision in 30 years. It is no AWS and no ads, so take away essentially all the net income. It is a very low-margin retail business with a massive amount of infrastructure investment needed to get there. No one would invest in that; that is not investable.

Then someone would say you can unbundle Amazon, so you’re going to unbundle everything, and you have some company handling logistics. That’s not profitable. You can’t make money doing this. My neighbor runs a business that does delivery for Amazon. Do you know how Amazon gets deliveries done in three hours? They finance the trucks. You go to Amazon and get a loan for the trucks. That’s not an investable business on its own. As an independent business, that would never evolve past some venture stage. It wouldn’t be able to make money. It is only in the bundling where that becomes a valuable business.

**MM:** They are bigger than UPS now. It is truly the largest physical moat I can think of in my investment research career. I saw a great response to the tweet you were referring to where a guy was joking as he’s pitching to Shark Tank. He says he wants a trillion dollars for a quarter of his business so he can buy 100,000 vans and build 1,500 fulfillment centers. This is 20-plus years in the making of solving for last mile in every nook and cranny and neighborhood in America. Now more than ever, consumer demand for time-sensitive deliveries is increasing constantly.

When you talk to all these operators and go to the conferences and listen to the fireside chats, Amazon took everything to two days and killed a lot of retailers along the way, but it is now beyond that. Now it is a race for same day. Walmart gives these sub-multiple hour delivery stats, and these executives talk about conversion going to the moon when they show people they can get them this in six hours. It is now even harder than ever to satisfy the needs of the consumer. Where the incremental growth and penetration is coming in e-commerce is the most time-sensitive stuff; it is these everyday essentials and grocery aspects.

**ES:** It is this fantasy land that the people who are championing this framework the loudest live in. Maybe that fantasy land is just Silicon Valley, but it just doesn’t comport with reality.

**MM:** For investors, we have to be, and our client base has to be, extremely sensitive to shifts like this because everybody likes to knock Wall Street as being short-sighted. I am telling you, there is nothing more long-term focused than tech investing on market share dynamics. The second the market gets a whiff of market share loss for a given operator, it will take a multiple and cut it in half. That is why Wall Street is particularly sensitive to this. When these types of paradigm shifts present themselves and they go back and look at generations of pattern recognition, that is why you see investors get especially concerned.

**ES:** Why do investors give no credit to Rufus, or the artist formerly known as Rufus, now Alexa for Shopping? That exists; it is years old and 12 billion in incremental annualized sales in 2025. This is not some skunkworks tool that is being incubated currently; it is working and consumers adopted it. If you think agentic commerce is the future, I agree with you, it is the future, and Amazon proved it. But it may take place on Amazon. The response to that would be that I need one agent that does everything. But consumers have demonstrated that they are willing to use the Amazon agent.

**MM:** This is a really hard question. There are some things I have firm beliefs on that I am willing to change on a weekly basis, and then there is stuff I don’t even have a firm belief on and that I am constantly debating in my head. We have published some notes talking about it, and it is what consumers are going to look for. Do they want one front door to the internet or do they want to go to Amazon for this portion of their shopping?

I have tried to look at current consumer behavior and how consumers move about the digital economy today. You do see verticalization; you go to DoorDash for your food aspects. When you look at these top 15 e-commerce players representing 70% of the US GMV market, they all generate a majority of their traffic direct. Google is a third of shopping experiences getting started, and some have way less Google traffic and get more through social. In general, 65% of their web traffic is direct. Someone opens up their browser and they type amazon.com, ebay.com, or wayfair.com.

I’m doing my best to try to study how consumers have behaved for almost the entirety of my life and then try to think about how they are going to behave in the future. The pushback that the agentic bulls would give me is that traditional Google search for product discovery wasn’t that great; brands showed up in the carousel at the top for the PLAs that weren’t actually the best answers. Now they are going to have this front door that gives them the best answer and this product experience is going to get so much better. I think it is very fair feedback, but when I look at how consumers have behaved for almost the entirety of my life, a majority of it goes right to where it wants to go.

For the stuff where you really don’t know, for high-consideration purchases, you’re starting at Google. To be fair, I am skeptical of almost every disclosed data point from all of my covered companies. What I found with Alexa, and at the time it was Rufus, they were auto-defaulting me into it. So I was starting a search and all of a sudden I realized the feed didn’t look the way it traditionally does within the Amazon experience. I was like, oh, I’m in Rufus right now. I didn’t even click to enter this; I am clearly part of some AB testing. When I see the stats they disclose around that, I take that with a heavy grain of salt.

**ES:** I want to bring Muse into the discussion. I want to quote from your recent report on Muse. The top 16 e-commerce operators in the US, who represent 70% of US e-commerce GMV, get 65% of their total web traffic as direct traffic. Dominant e-commerce operators spent decades building brands, trust, and reducing friction with the consumer. Marketplaces have moats. The rumors of their death are greatly exaggerated. Talk to me about what exactly that moat is, because I think the bull argument back would be that sounds like a brand to me, and brands can erode over time if you give the consumer a better experience. They’ll just embrace it and the brand can only delay the inevitable for so long. Does Amazon have enough share to just foreclose this path, and what is the moat?

**MM:** Let’s talk about that a bit. I think it’s important for people who live outside of my ecosystem to understand the genesis of a lot of these reports we write. For my job, I research, I write reports, and then I spend a big portion of time talking to investors. Yesterday I was in Chicago in meetings all day. Today I have already done six calls with investors, and we are just talking through the risk and rewards in consumer internet right now. The way these risks show up in the market is valuation multiple compression.

There is actually not a lot of movement right now with fundamental numbers and forward-looking estimates. There is always some up and down, but really the big volatility swings you are seeing right now are perceived existential threats that are taking stuff up and down. When I write these reports and I’m arguing with an invisible person in the sky, it is addressing what are some of the really topical debates showing up in the markets.

Especially magnified by the release of Instacart and Muse has been this view that marketplaces and apps and aggregators are dead. What we wanted to do with that report to try to explain that marketplaces have moats is to show that what they do is extremely challenging and also very expensive. The way I thought we could do it was to bridge through the average take rate that a marketplace has and what portion of that take rate is expenses and really messy work that people don’t want to do.

When you listen and debate with people about this, everything is under threat right now. The AI platforms are going to build payments and checkouts; they are going to take everybody’s payments revenue. So I thought let’s walk through this. The average marketplace take rate excluding fulfillment, because Amazon’s FBA will take it much higher, is call it 17%. The point that we make is 10% of this 17 is costs, and it is unavoidable costs if you are going to operate a scaled marketplace.

Call it 3% of that 10% is your payments processing revenue. That is a sub-30% gross margin business, but what that does is it covers the costs of what it means to be a merchant of record. It is something everyone forgets in this whole agentic e-commerce debate. You want to be a merchant of record? Well, in the US you are then legally responsible for refunds, fraud, consumer protection laws. This is a really messy business. Being merchant of record carries a bunch of costs around it that basically gets covered by these companies’ payments revenue. That is three.

Then you move to the next 3% of the cost. These are huge marketplaces with hundreds of millions of listings and enormous infrastructure and cloud costs, customer support, merchant onboarding. A lot of the public companies break out these costs in their cost of revenue, so you don’t even need to guess. Then you can look at it in how they charge some of their fee structures. Uber and DoorDash, if you just want to do a takeout order, the fee to the merchant is 6%. I just told you there is kind of 6% merchant of record, payment revenue aspect, and 3% of running the infrastructure. So three plus three is six. Uber and DoorDash are basically trying just to cover their costs so they’re not losing money on these takeout orders.

You are now at 6-7% of this 10 that is all cost. The rest of that is demand generation. Even Amazon buys ads. Muse is trying to layer in between this right now. I cannot remember such a full-court marketing press in consumer internet in a really long time as what I am seeing with Muse right now. I was in Chicago yesterday and there were Muse billboards and commercials. You want to run one of these e-commerce businesses, you have to have demand generation, and that’s expensive.

Ten of the 17 is costs. That is not a fun business and that’s for really large-scale marketplaces. So if you are someone who is trying to layer in and displace these players, imagine how much of a portion of your cost that’s going to be when you’re not the best at it and you’re not scaled doing it. Then what’s left in these businesses? Six to seven percent is the profit pool. Four to five percent of that, depending on the platform, is the onsite ads. I have talked and written a lot about how I’m worried about that. There is definitely some risk there.

Then the last portion of it is what I call trust, and that’s where they can get this surplus and earn themselves a profit. It is trust and reviews and it is their content. I worry right now that these marketplaces are undervaluing the content that they have spent decades building. If you look back through the history of content distribution, distributors pay the content providers. In the report on personal agents that we published, we tried to make it super clear: if you are an investor and you’re waiting for the sell-off in marketplaces to end so you can get long one of these positions, I would be terrified of the Muse numbers for the near to medium term.

Meta can see how sticky the user base has been for ChatGPT. Consumer-facing AI rewards first movers tremendously. We have this chart showing the MAUs up and to the right for ChatGPT to up over a billion from the beginning to then. Then you do the engagement stats, like DAU over MAU, and it is incredible. They have been able to grow this base, and we plot along the line of all the competing products as they launch: Grok, Meta AI, Perplexity’s new additions, and Gemini, and it just is really sticky and difficult to displace.

Mark Zuckerberg and Meta have obviously seen that. They know they have the first-to-market personal agent that can scale. Honestly, if I was on the team at Meta, I’d say let’s spend on every possible channel we can to the most we can do it because the land that we can grab now we will be able to keep. So if you’re trying to put some bet of this slowing down, it is going to be a while. The numbers are going to keep looking really good, and it is not going to be for a while until people realize that the path to monetization with this is not a commission rate on commerce, it’s ads, or the fact that it is really good helping a two-job family with two kids manage their life.

**ES:** I think it’s going to be ads. I want to talk about Walmart because you brought up Walmart, and that is a really interesting case. Is that a bellwether or a red herring when we talk about agentic commerce? We know their history with OpenAI. Then after Walmart, talk to me about Stripe and Shopify because they all three of these companies have different incentive structures.

**MM:** I love watching Walmart here because in traditional retail they are king. The largest grocer in the US, everybody lives within a few miles of a Walmart; they control the world. In e-commerce, I won’t call them an ankle biter, they have eight or nine percent market share, but we all get it. They are coming from behind trying to displace Amazon. It makes all the sense in the world for them strategically to embrace every one of these channels that pops up that Amazon is not in. You should do that because ads drive the profits of Amazon’s retail business.

Walmart has their own ads business and they are following the Amazon blueprint to a tee, like buying Vizio. They have almost nothing to lose and everything to gain by one of these new platforms gaining share and riding it to take share from Amazon. It makes perfect sense for them.

Stripe, I think the Collison brothers are brilliant. This is a great opportunity for them; Stripe Link, having an agentic wallet, the ball is theoretically coming their way. Where we care more about this overlap and you get a lot of misunderstandings in the equity market over the last month is they are trying to disintermediate Shopify. We put this picture in a recent report of John Collison, Tobi Lütke, and Mark Zuckerberg after they spent a day together riding go-karts and they are covered in dirt. Zuck put it up on his Instagram.

That is a very important separate ecosystem to the Amazon one. It is the Meta, Shop, Stripe e-commerce ecosystem that has an incredible opportunity to thrive right now. I do believe AI product discovery is changing the consumer’s product discovery funnel. That is a hill I will die on. Historically, and we talked about this in a report on Amazon or Shopify, can they both thrive in this world?

DTC was never a business model; it was a channel. Customer acquisition never got leverage if you were brand.com; you always ended up running into the arms of Amazon over a five-year window. Every DTC brand dies other than the few occasional ones that explode because it is so hard to win in Google and Meta and you have to pay the tax every time. Now, we have shown empirically that the search traffic coming from AI search platforms is outgrowing brand.com search traffic. So traffic being sent from all the AI search added up being sent to brand.com, which is the Shopify merchant base, compared to all of the largest marketplaces.

The brand.com traffic is outgrowing it by, depending on the window you want to cut it, five times. This is an incredible opportunity for Shopify. We are really bullish on Shopify; we upgraded at the beginning of the year when we published this web traffic that we could empirically measure and show you. This is potentially reworking the infrastructure of product discovery in Shopify’s favor. That is good for Stripe because Shopify is 13% of Stripe’s payment volumes. This is good for Meta as well because the R-squared between Shopify’s GMV and Meta’s revenues is basically perfect. Retail and e-commerce is the largest ad buyer on Meta. This whole ecosystem outside of Amazon getting healthier together is a really great thing for those three players.

**ES:** The incentive structures for Shopify, it makes sense for them to embrace this because they are enforcing the use of Shopify Checkout. You have to; they prohibit autonomous transacting. The human has to approve it, and you have to use Shopify Checkout. So why not? Let’s just align with everybody. Let’s be part of every single commerce protocol because you’re going to work through our payments infrastructure and that benefits us. All of it is good; just more transaction volume through their pipes.

**MM:** It is all upside for Shopify. Anything that you think about this giant product discovery of billions of people on the planet, and you just have Tobi at the bottom and AI-led search just leaning and pulling it just a little bit more in the Shopify merchant base favor has profoundly important impacts to the business, to the health of the ecosystem that they run. I think Tobi is a generational CEO, also Tony at DoorDash and one or two other operators. He is so fast-moving on these changes. It is also driving more merchant adoption for them.

When I talk to brands, they say they’ve been running their custom e-commerce stack for 20 years with a team of engineers and devs doing the best they can to keep up with things. It is hard, but it is fine. It seems like every day there’s a new API platform we’re plugging into because we’re cross-selling on TikTok and all over the place. Now with AI, this is just too much for us to keep up with, and you look over at Shopify, you have a singularly focused platform attacking this issue. They picked up the phone and are working with a system integrator to replace their tech stack and go with them. Product discovery is changing and it is going to drive more merchants into their arms as well.

**ES:** Let’s say I am wrong and this overwhelms Amazon and they capitulate. Talk to me about three companies in particular: Amazon, Booking, and DoorDash. What does each gain and give up by capitulating here and letting an outside agent handle the customer interaction? DoorDash has already made some moves there, so we get a sense of where they are headed. What does each of these three companies gain and give up if they let an outside agent handle the customer interaction?

**MM:** I don’t think they really gain anything. You have to go back to the view again where I love these young entrepreneurs taking risks and growing these businesses. They are really not giving these platforms any credit for the fact that they have spent decades and untold billions of dollars reducing friction. Amazon had the patent on one-click checkout. They are actually not trying to make this hard, because every layer of friction and click you add, you reduce conversion. They are in the business of making this easier.

They have developed these relationships with the customers over decades that I think are critical for them. DoorDash, there is a lot of ways you can analytically prove that this is not how consumers eat. You can look at the data where 50% of people spend somewhere between five to ten minutes looking at a menu. People who are members of restaurant loyalty programs still eat at an equivalent amount of restaurant variety as those who don’t. The reason that US restaurants churn at 20% annually in the US is because we all get sick of eating the same stuff and we want to go eat other places. We are not dogs; we’re not going to eat the same dog food all the time. You’re not going to tell your agent, order me the same thing every week. It is just not going to happen. I just don’t think that these AI platforms are going to really bring incremental demand to DoorDash.

I don’t see what they gain there. Amazon, again, I don’t see what they gain there. I think they have to go if we get this wrong. If Muse is 30% of e-commerce in three years, Amazon has got to go, and I got it wrong. I don’t cover the OTAs. Where I struggle, again, is on human behavior. Maybe I am actually more bearish on this than you are. The average American takes three vacations a year and three round-trip flights. Again, Silicon Valley and Wall Street folks, I take sometimes three round-trip flights in a ten-day window, and I fly a lot, and we’re again extrapolating this onto the population at large.

I have a very hard time thinking that the average American consumer on their either once or twice annual vacation with their family isn’t going to want to sit—and there’s data behind this—they look at eight or nine pictures and look at a couple different hotels, see what the pool and beach situation looks like. I want to go look at what the hotel gym looks like. I want to go see what the food in the restaurant that’s inside the hotel looks like. I just have a hard time believing that the average American family is going to outsource their largest discretionary spending single purchase moment of their year for it to get messed up to some extent or just give up control. I don’t see the demand for that.

**ES:** Someone on X said they’re increasing the use cases served with agents, and one of them is your green card paperwork. What is the market for that? My wife got her green card; you’re going to do that once in your life if you do it. The number of people that ever do it is small and they do it once, and it is high stakes. If you get something wrong, it sets you back three months. This is just not something you’re going to surrender to an agent.

You are going to use these tools for discovery, of course, and you are going to use them to get options. That’s what they’re really good at: help me narrow down the list, help me find these specific properties that meet these specific requirements. But if the agent can only do that, and for high-consideration stuff that is the limit of the authority you give it—find me options, describe them, be a discovery tool, a really good one, a comprehensive one—and then when it comes time to pull the trigger, I’m going to go directly to the website. If the agent can only do that, what’s the opportunity there?

**MM:** What’s it mean for the digital ads ecosystem? There is nothing I would love more from a career perspective than to think that this was really truly going to disrupt the e-commerce ecosystem because I would love to downgrade everything to sell and make this counter-consensus stock call. I just don’t see it.

Studying how consumers behave and what I think people want leads then into this product discovery aspect. I’m kind of bummed out that this agentic debate has run over this discovery conversation, because this is so cool. I spent a week researching a couple of products, like flat-footed running shoes, filtered coffee makers, and natural deodorant. I read 30 different website journals about the best flat-footed running shoes, counting the references to see what the actual best one is. I have done this for a lot of different products. The LLMs nail it. They do what took me 40 hours to do in 30 seconds. This is awesome for the consumer. You’re going to take out so much inefficiency and wasted time from our ecosystem.

Then the question is what it means for the ads ecosystem. This should be monetized through ads, and I think it’s the best clearing mechanism for this type of product discovery. I think it will lead to a higher take rate than a three or four percent commission if you just look at what all these brands budget on ad spend. The company that has been most transparent about it is Mercado Libre. They have said they haven’t commented on their click count, but it is similar to your Amazon number of 14 products clicked per purchase.

What they said at our conference this spring was that they have rolled out LLMs in product discovery on their website, and yes, it leads to a compression in this discovery and fewer clicks, but the higher conversion rate is making the clicks way more valuable, and cost-per-click is going up more than enough to offset the decline in clicks. Michael Nathanson brings me to some of these dinners, and we were at dinner with an ad executive of a billion-dollar-plus ad business. This guy’s more in the open web, and even what he was talking about is he’s seeing the value of clicks go up.

You’re taking things that used to be four or five queries that weren’t particularly high-conviction traffic, and now that is getting honed down into one, and you’re getting a lot of commercial intent. So I’m perfectly comfortable in this world where aggregate click growth rate is slowing, but the value of the traffic, especially as consumers learn to trust the models, is going to be probably a good thing for digital ads at large.

There is waste in the ad ecosystem. We all know that, and I think this future will squeeze out some of that waste. So not everybody is going to get out of this cleanly, but that’s how I think this product discovery shift is going to be probably a good thing for digital ads at large. You’re getting better commercial intent. The better brands are going to win. No longer are these super high-quality brands just getting pushed out of product discovery because the Nikes of the world have this enormous marketing budget and a perfectly converting website and they just do better in Google scores. It’s blending better discovery, and they’re going to be able to then convert higher and bid on it, and it’s going to be more merit-based. I think that’s really exciting for the long-tail brands.

**ES:** Michael Morton, this was fantastic. I really appreciate you coming on the podcast today to share your wisdom with me and with the audience. How can people follow you on social media and how can they learn from you elsewhere?

**MM:** Due to financial regulations, I am a ghost on social media. So when I do occasionally see people make fun of me, I just have to read it and keep scrolling. My email is morton@moffettnathanson.com. I’m on LinkedIn. All these podcasts that I’ve gone on have started some really fruitful relationships and conversations. People in the space who love to talk about this stuff, please reach out. I love talking about this stuff; I am just a lurker in the social media world.

**ES:** All right, then this is it, this is all you get. But hopefully it was worth your while. It certainly was very informative and enlightening for me. Michael, thank you so much.

Source: [https://mdm.co/podcast-will-consumers-adopt-agentic-commerce-with-michael-morton/](https://mdm.co/podcast-will-consumers-adopt-agentic-commerce-with-michael-morton/)