> ## Content Index
> Fetch the complete content index at: https://thepaymentsoperator.com/llms.txt
> Use this file to discover other available public pages before exploring further.

# Zalando Will Buy From Other Shops for You. Your Statement Will Name Someone Else.
- URL: https://thepaymentsoperator.com/zalando-will-buy-from-other-shops-for-you-your-statement-will-name-someone-else/
- Published: 2026-11-10T12:05:00.000Z
- Updated: 2026-11-18T11:13:42.000Z
- Description: A marketplace that holds the checkout but not the inventory has an acquirer's power and none of its obligations.
- Author: Armando

*A marketplace that holds the checkout but not the inventory has all the commercial power of an acquirer and none of its obligations.*

Zalando announced something on 3 November that is more interesting to a payments person than to a retail analyst, and the retail analysts got most of the coverage.

Alongside its third-quarter results, the company unveiled Zalando Extended: products from other online retailers inside the Zalando app, and from 2027, an AI-assisted checkout that completes those purchases without the customer leaving. The first pilot partner is Westwing, the home furnishings retailer.

The retail reading is straightforward and correct. Zalando is building a defence. JPMorgan said so in plain language, warning of *ongoing structural risk from agentic commerce development* — which is analyst for: if agents are going to shop across every store regardless, the only safe place to stand is being the agent.

Walmart reached the same conclusion from a different direction when it pulled out of somebody else's chat and put its own assistant inside two of them. The lesson is converging across very different retailers: discovery can be rented out, but the customer relationship cannot.

That part has been written about. What has not been written about is what this does to a card transaction, and it does something genuinely new.

## Three different answers to one question

When Zalando's agent buys a sofa from Westwing on a customer's behalf, ask a simple question: who did the customer buy from?

There are three answers, and they do not match.

**The customer thinks they bought from Zalando.** That is where they were, that is the app they opened, that is the brand that recommended the thing, and that is who they will blame if it is wrong.

**The goods and the terms belong to Westwing.** The returns window, the delivery promise, the warranty, the condition on arrival — all of that is a contract between the customer and a company they may not have consciously registered as a counterparty.

**And the statement will say whatever the merchant of record is.** If Westwing processes the payment, the descriptor says Westwing, and a customer who remembers buying from Zalando sees a charge from a company they have no memory of dealing with.

I have written before that the merchant descriptor is where a transaction's identity attaches, and that when the buyer's memory and the statement string drift apart you get a dispute that was never about fraud. This is that problem, made structural and then scaled.

## Neither choice is free

Zalando has two options and both cost something, which is why this is worth watching rather than dismissing.

**Option one: Westwing stays merchant of record.** Clean from a liability standpoint — the seller sells, the seller is paid, the seller handles returns. But Zalando has just built an experience, recommended a product and closed a sale, and then handed the brand credit to somebody else at the exact moment the customer's bank records it. It also inherits the dispute volume without the dispute rights: the customer calls Zalando, and Zalando can do nothing except pass them to Westwing or to their bank.

**Option two: Zalando becomes merchant of record.** The statement says Zalando, the customer recognises it, the experience is coherent. And now Zalando is contractually the seller of furniture it never saw, never shipped and cannot inspect, carrying chargeback liability for delivery disputes and condition claims on goods that physically moved between two other parties.

There is no third option where you keep the brand and avoid the liability. That trade is the whole business model question, and it is the same one every marketplace has faced — except that this time an agent, not a person, is deciding what gets bought.

## The orchestrator nobody licenses

Here is the part that I think is genuinely novel, and I have not seen it stated anywhere.

Walk through what Zalando controls in this flow. It decides which merchant gets the sale. It decides which product is recommended. It holds the customer's stored payment credential. It triggers the purchase. From 2027, through an agent, it decides when the purchase happens at all.

Now walk through what Zalando is, in regulatory terms. Not the merchant of record, if Westwing keeps that. Not an acquirer. Not a payment service provider. Not a payment institution. It may never touch the funds, and if it never touches the funds, most of the frameworks that govern payment intermediaries do not obviously reach it.

So you get an entity with decisive influence over who gets paid, by whom, with which instrument — and essentially none of the obligations that come with being any of the licensed roles in that sentence.

This is not an accusation of bad faith. Zalando is building a product, and the regulatory perimeter simply was not drawn with this shape in mind. Our definitions of who must be licensed to intermediate a payment were written for a world where intermediation was visible: you held the money, or you routed the message, or you acquired the merchant. A party that selects the merchant and the card but never handles either is not obviously captured by any of it.

And this shape is about to become common. Every retailer that reads JPMorgan's warning reaches the same conclusion Zalando reached, which is to become the agent before somebody else's agent takes the customer. Each one that does creates another orchestrator with commercial control and no defined obligation.

I do not think the answer is to rush licensing at it. Premature regulation of a shape nobody understands produces worse outcomes than waiting. But somebody should at least be asking the question, and in most of the markets I watch, nobody is — because the entity doing it looks like a shop.

## The timing is the tell

One detail from the results that I would not skip, because it says something about when decisions like this get made.

Zalando's third quarter: revenue up 5.1% to €3.17 billion, slightly ahead of expectations. Gross merchandise value up 7.5% to €4.54 billion. And gross margin down 0.9 percentage points to 38.7%, which missed expectations, attributed mainly to investment in attractive offers — discounting, particularly in sneakers. The finance chief called the pressure targeted and temporary, and guided to a broadly flat gross margin for the year.

I have no view on whether that is right. What I notice is the sequence: a company under visible margin pressure announcing a channel in which it will, under either structure, take on payment risk or cede brand attribution for goods it does not own.

That is not a criticism. It is how these decisions actually get made — under pressure, with the strategic logic clear and the operational cost not yet quantified, because nobody can quantify it until volume exists. I have sat in versions of that meeting. The payments consequences of a commerce decision almost never arrive in the room where the commerce decision is taken.

## From a Panamá desk

The descriptor problem is not coming to this region. It is already here, and we have been absorbing it quietly for years.

Any issuer with a dispute queue in Latin America already deals with marketplace charges the cardholder does not recognise — a third-party seller's name on a statement for something bought on a platform, a payment facilitator's descriptor standing in for a merchant the customer has never heard of, a sub-merchant aggregated under a name that means nothing to anybody. We handle it. It costs us representment cycles and call-centre minutes, and most institutions do not measure it as a category because it does not have a reason code of its own.

Which means we have a baseline problem and no baseline measurement. If anybody reading this runs an issuing portfolio here, the useful exercise is not speculating about Zalando. It is going into the last twelve months of *I do not recognise this charge* disputes and asking how many trace back to a marketplace or a facilitator descriptor rather than to actual fraud. That number exists. It is just not in any report, because nobody has asked for it.

Get that number now, because the agentic version multiplies it. Today the customer at least clicked buy themselves, at a moment they remember. In the 2027 version an agent buys, at a moment the customer was not present for, from a merchant they did not choose, and the descriptor still says somebody they never heard of.

The second regional point is about the perimeter. In most markets here, the rules defining who needs authorisation to intermediate payments are newer than Europe's, drafted more recently, and in several cases drafted precisely to capture fintech intermediaries. That is an advantage, not a disadvantage — a recently written perimeter is easier to extend than one drafted in the nineties. But it only helps if somebody notices the new shape in time to ask whether it falls inside.

## My judgment

Marketplaces without inventory are forty years old. That is not what is new here.

What is new is a marketplace that holds the *checkout* without holding the inventory, the licence, or — from 2027 — the human who decides. Strip it down and you have a party with maximum influence over a transaction and minimum defined obligation for it, which is a structure that has never been stable anywhere in payments and will not be stable here.

It resolves in one of two directions. Either these orchestrators take on merchant-of-record status and the obligations that come with it, which is expensive and which most of them will resist until something forces it. Or the perimeter moves to reach them, which takes years and arrives badly. The industry's preference should obviously be the first. The industry's track record this year suggests the second.

And the thing I keep coming back to is how familiar this is. It is the same shape as the agentic liability question I have been writing about since April: a new party acquires real influence over a transaction, the existing rules do not name it, and the loss keeps landing on whoever the old rules did name. Zalando is not an AI story. It is the same story, arriving through retail instead of through a protocol.

## Three questions for the people who own this

**1.** In your last twelve months of disputes, how many *I do not recognise this charge* claims trace to a marketplace or facilitator descriptor rather than to fraud? If that number is not in a report today, it will not be in one when agents multiply it.

**2.** When a platform's agent buys at a third-party merchant, what does your institution want to see in the authorization — the platform, the merchant, or both, tied together? Has anyone put that requirement to a network in writing?

**3.** Under your market's rules, does a company that selects the merchant, holds the credential and triggers the purchase, but never touches the funds, need any authorisation at all? If nobody at your institution knows, that is the answer.

Reply to this email with a line or two. The sharpest answers will shape the next issue.

## Sources

**Reuters, 3 November 2026** (read via RTÉ) — Zalando unveils Zalando Extended alongside third-quarter results: products from other online retailers inside the Zalando app, AI-assisted checkout planned for 2027, Westwing as pilot partner; revenue up 5.1% to €3.17 billion against €3.15 billion expected by analysts polled by LSEG; gross merchandise value up 7.5% to €4.54 billion; gross margin down 0.9 percentage points to 38.7%, attributed mainly to investment in attractive offers, especially sneakers; JPMorgan's comment on ongoing structural risk from agentic commerce development; and finance chief Anna Dimitrova's description of the margin pressure as targeted and temporary, with a broadly flat gross margin forecast for 2026\. [Link](https://www.rte.ie/news/business/2026/1103/1597792-zalando-launches-ai-shopping-agent-as-margins-squeezed/?ref=thepaymentsoperator.com)

Every figure above is from that report. The analysis of merchant-of-record options, the descriptor consequences and the regulatory perimeter question is mine; Zalando has said nothing publicly about which payment structure Zalando Extended will use, and nothing above should be read as describing a choice the company has announced.