This was a good read, thoughtful and comprehensive. It brought a few thoughts to mind:
1. Might there be a regulatory component to this at some point? Like how car dealers have leveraged franchise laws in the US to prevent direct-to-consumer sales, but in reverse?
2. AI shopping will make accurately measuring net-new customers (via their agents) crucial (i.e., to capture the operating income-ad revenue tradeoff you astutely observed.
3. Wouldn't advertisers pay less to advertise on sites like Amazon if more traffic is coming via agents? So there might be a one-two punch to decreasing ad revenue from intercepted orders.
Really well reasoned and explored. Thank you for sharing these thoughts.
I don’t understand how facebook’s muse maintains market share in the face of iOS and Android. Sure Facebook has data on me but nothing like Apple and Google Android. What muse does seems an obvious concept to embed in Siri and Gemini. Why is muse relevant except as a first mover?
Apple and Google have monetization methods beyond Facebook. The question is whether social media and messaging remains the same in the AI age. That’s open. No one has created AI that would displace social media at this point.
Amazon, DoorDash and Instacarts value is less in being an aggregator and more about them implementing delivery. Yelp and Google biz is dominant in local food and is used by AI to train itself. DoorDash et al are gotos if I’m looking for delivery. Amazon the same.
I tend to look at who has a sustained differential advantage (aka a moat).
How ads and memberships (ala doordash and amazon) play out is also interesting.
I own cookie shops. All but Amazon are relevant for buying. Should I support muse? See no reason not too. I’m comfortable in letting the elephants battle it out because none of them changes my sustained competitive advantage.
I mostly think about how AI is trained to displace Yelp and Google business in recommendations locally. Also how AI is trained to find me for national deliveries.
But a very interesting analysis of the elephants playing for dominance.
An agent can become the intent layer surprisingly quickly, but transaction trust is much harder to move.
Travel is a particularly good example: Muse may decide what I should book, while Booking , Expedia Airbnb can still capture the transaction because that’s where I’m more comfortable completing the purchase.
I trust Brian Chesky is cooking something here too…
The piece frames this as agents vs. marketplaces fighting over ad revenue. There's a quieter shift underneath: when there's no sponsored slot to buy, the product data is the ad. Brands that structure their catalog for agents will win orders that ad spend can no longer buy.
Interesting perspective! As AI takes over more of the shopping journey, the real opportunity lies in how payments, merchants, and platforms adapt to this new ecosystem.
Building an AI shopping agent made this feel very real. The hard part wasn't recommendations, it was that most stores aren't set up to be read by an agent at all, product data, stock and pricing are built for a human scrolling a page. Whoever makes their catalog easiest for agents to trust probably wins before the profit pool debate even starts. Do you think marketplaces will open structured feeds to assistants, or keep blocking them the way Amazon did?
Even the cover image illustrates the gap in its framing. It shows a shopping cart on a circuit board, as though AI shopping begins at the cart! The cart is the end of the journey. Where did the desire for the product begin? Who helped the consumer discover and evaluate it? Who should be credited and paid when that discovery leads to a purchase?
The article asks who will own the customer and keep the ad revenue. It cites how much platforms earn from advertising, but does not show how much of that spending produces incremental sales rather than claiming credit for purchases that would have happened anyway. How can we measure it?! Some advertising creates demand; some merely captures credit for it! We want to make that difference measurable and pay people for the commerce value they actually create.
That is the question MarkitPlace https://markit.place/ is built to address. The reason this platform exists is right there in the article’s first sentence: “No one wants to be disintermediated.”
MarkitPlace is a network orchestrator, not another shopping destination. Intent can begin with a film, a creator, a publisher, a physical experience, or an AI conversation. We connect that discovery to relevant products, retailers, marketplaces and transactions while preserving attribution to the parties that helped create demand. We can follow the path from discovery to purchase and, where the data allows, compare exposed and unexposed groups to test whether an experience generated additional sales. That gives participants a way to share in a larger market without forcing consumers to begin at any one destination.
Consumers have to remain in control! We should give people the product information and analysis they need to decide what to buy, where to buy it, and whether to buy at all. They should be able to remain anonymous, and we SHOULD STOP SELLING their personal data! WE NEED A FUNDAMENTAL CHANGE, not another round of lipstick on this pig, again and again and again!
This was a good read, thoughtful and comprehensive. It brought a few thoughts to mind:
1. Might there be a regulatory component to this at some point? Like how car dealers have leveraged franchise laws in the US to prevent direct-to-consumer sales, but in reverse?
2. AI shopping will make accurately measuring net-new customers (via their agents) crucial (i.e., to capture the operating income-ad revenue tradeoff you astutely observed.
3. Wouldn't advertisers pay less to advertise on sites like Amazon if more traffic is coming via agents? So there might be a one-two punch to decreasing ad revenue from intercepted orders.
Really well reasoned and explored. Thank you for sharing these thoughts.
Great article. Number of thoughts,
I don’t understand how facebook’s muse maintains market share in the face of iOS and Android. Sure Facebook has data on me but nothing like Apple and Google Android. What muse does seems an obvious concept to embed in Siri and Gemini. Why is muse relevant except as a first mover?
Apple and Google have monetization methods beyond Facebook. The question is whether social media and messaging remains the same in the AI age. That’s open. No one has created AI that would displace social media at this point.
Amazon, DoorDash and Instacarts value is less in being an aggregator and more about them implementing delivery. Yelp and Google biz is dominant in local food and is used by AI to train itself. DoorDash et al are gotos if I’m looking for delivery. Amazon the same.
I tend to look at who has a sustained differential advantage (aka a moat).
How ads and memberships (ala doordash and amazon) play out is also interesting.
I own cookie shops. All but Amazon are relevant for buying. Should I support muse? See no reason not too. I’m comfortable in letting the elephants battle it out because none of them changes my sustained competitive advantage.
I mostly think about how AI is trained to displace Yelp and Google business in recommendations locally. Also how AI is trained to find me for national deliveries.
But a very interesting analysis of the elephants playing for dominance.
An agent can become the intent layer surprisingly quickly, but transaction trust is much harder to move.
Travel is a particularly good example: Muse may decide what I should book, while Booking , Expedia Airbnb can still capture the transaction because that’s where I’m more comfortable completing the purchase.
I trust Brian Chesky is cooking something here too…
The piece frames this as agents vs. marketplaces fighting over ad revenue. There's a quieter shift underneath: when there's no sponsored slot to buy, the product data is the ad. Brands that structure their catalog for agents will win orders that ad spend can no longer buy.
Interesting perspective! As AI takes over more of the shopping journey, the real opportunity lies in how payments, merchants, and platforms adapt to this new ecosystem.
Building an AI shopping agent made this feel very real. The hard part wasn't recommendations, it was that most stores aren't set up to be read by an agent at all, product data, stock and pricing are built for a human scrolling a page. Whoever makes their catalog easiest for agents to trust probably wins before the profit pool debate even starts. Do you think marketplaces will open structured feeds to assistants, or keep blocking them the way Amazon did?
Even the cover image illustrates the gap in its framing. It shows a shopping cart on a circuit board, as though AI shopping begins at the cart! The cart is the end of the journey. Where did the desire for the product begin? Who helped the consumer discover and evaluate it? Who should be credited and paid when that discovery leads to a purchase?
The article asks who will own the customer and keep the ad revenue. It cites how much platforms earn from advertising, but does not show how much of that spending produces incremental sales rather than claiming credit for purchases that would have happened anyway. How can we measure it?! Some advertising creates demand; some merely captures credit for it! We want to make that difference measurable and pay people for the commerce value they actually create.
That is the question MarkitPlace https://markit.place/ is built to address. The reason this platform exists is right there in the article’s first sentence: “No one wants to be disintermediated.”
MarkitPlace is a network orchestrator, not another shopping destination. Intent can begin with a film, a creator, a publisher, a physical experience, or an AI conversation. We connect that discovery to relevant products, retailers, marketplaces and transactions while preserving attribution to the parties that helped create demand. We can follow the path from discovery to purchase and, where the data allows, compare exposed and unexposed groups to test whether an experience generated additional sales. That gives participants a way to share in a larger market without forcing consumers to begin at any one destination.
Consumers have to remain in control! We should give people the product information and analysis they need to decide what to buy, where to buy it, and whether to buy at all. They should be able to remain anonymous, and we SHOULD STOP SELLING their personal data! WE NEED A FUNDAMENTAL CHANGE, not another round of lipstick on this pig, again and again and again!