I have spent well over a hundred thousand dollars, across my life, on luxury clothing and goods. I mention it at the start not as a credential but as a disclosure, because it is the thing that makes everything after it legible. Customer experience
There are objects I want and cannot have. Some of them I will probably never have. I am not upset about this. Scarcity is one of the more honest things left in the customer relationship — a manufacturer saying, in effect, we are not going to make more of these simply because you would like us to. When a company declines to dilute something in order to sell more of it, that is a form of respect for the object, and indirectly for the person who ends up owning it.
So this is not an argument that luxury should be democratized. I do not think it should. It is not an argument that everyone deserves a Birkin, a Daytona or a limited Ferrari; nobody deserves any of those things. It is not an argument about price, and it is not an argument that a company owes its inventory to whoever turns up first with the money.
It is an argument about a much smaller thing: that there are two very different kinds of no. In the first, a customer is refused and can still describe the refusal — why it happened, roughly what would change it, roughly when. In the second, a customer is refused and cannot afterwards say what the no was about, or whether they were declined, deferred, ranked, or never registered as existing at all.
Those two kinds of no are not distinguished by how exclusive the company is. That is the proposition the rest of this piece tests against six published CHI assessments — and it survives. The most exclusive company in the group is among the clearest about its structure; a less exclusive one is among the most opaque.
Two questions that keep getting merged
Almost all writing about luxury access collapses two separate questions into one complaint.
The first question is why isn’t there enough? This is a question about supply, and it has legitimate answers: craftsmanship, manufacturing capacity, long production times, small-series production, deliberately preserved rarity, demand simply exceeding what the hands or the machines can produce. Call this scarcity.
The second question is given that there isn’t enough, who gets them — and what is the customer expected to do to improve their chances? This is a question about people, and its answers are queues, lotteries, retailer discretion, purchase history, relationships, invitations, internal rankings, unrelated purchases, retention behaviour, participation in a brand ecosystem, or criteria nobody outside the company has ever seen. Call this scarcity administration.
Scarcity is generally not the finding. The investigation begins when a company uses the power created by scarcity to structure customer behaviour, or when the rules governing access become materially opaque.
CHI looked at six companies through that lens: Lamborghini, Ferrari, Patek Philippe, Audemars Piguet, Rolex and Hermès. Every one of them has genuine scarcity. Not one of them is criticised here for having it. What separates them is what happens next.
The control case: Lamborghini
Start with the company that makes the whole comparison possible, because without a control the rest is just a description of luxury.
Lamborghini sells extremely expensive cars in constrained numbers. Reuters reported in March 2025 that it holds an order book of roughly eighteen months, with the stated aim of achieving controlled growth, and its chief executive has been open about not aiming for peaks in volume. That is deliberate undersupply, and CHI records it as such. Reported
What it is not is manufactured shortage. Lamborghini roughly doubled its production site for the Urus and grew from 3,815 deliveries in 2017 to a record 10,747 in 2025. A company that expands output roughly 2.8-fold in under a decade is not suppressing supply to create a shortage. The accurate description is deliberate undersupply inside an expanding production system — a materially weaker finding than the one usually asserted, and CHI records artificial scarcity as not established. CHI finding
The reason Lamborghini matters, though, is the access architecture rather than the volume. For ordinary production the answer to “can I buy one?” is money, an available allocation, and time. A car in dealer stock takes money. A build slot takes money and a queue measured in months. There is no evidence of a broad system in which ordinary customers must cultivate years of purchase history simply to become eligible for a normal production car.
Selection does appear — sharply — at the Few-Off boundary, where runs are 13, 29, 40, 63 or 112 cars, and where some allocation mechanism is unavoidable by arithmetic. Lamborghini is as capable of an inner collector circle as anyone in this cohort. But that behaviour lives at the edge of the range rather than describing the range.
Lamborghini primarily monetizes what the customer wants to buy, rather than monetizing the customer’s desire to remain eligible to buy.
That sentence is the axis of this entire comparison. It is also why Lamborghini’s own CHI assessment finds its principal problem somewhere else entirely — in how ordinary functionality has been packaged into expensive option groups, which is a revenue-extraction question, not an access one. On the axis measured here, Lamborghini is the control that proves the others are describing choices rather than describing physics.
Ferrari, and the difference between a hierarchy and a demand
Ferrari is the company most people reach for when they want to complain about luxury gatekeeping, and it is the company where the complaint most needs discipline.
Ferrari’s scarcity is not merely unproven as artificial; CHI rejects the claim outright. At runs of 799 and 599 cars the constraint is not an inference drawn from a waiting list — it is a published number. Ferrari shipped 13,640 cars in 2025 against an order book publicly acknowledged as full to the end of 2027. There is nothing to expose here. Ferrari says what it is doing. Fact
It also says, in its own disclosures, who gets priority. Ferrari disclosed that 84% of its 2025 sales went to existing owners, and 56% to customers who already owned more than one Ferrari. The hierarchy is not an allegation extracted from anybody. It is visible in the company’s own sales mix, published in its own materials. Fact
CHI treats that disclosure as a substantial point in Ferrari’s favour, and it should be said plainly: rewarding loyalty is defensible. A company that has sold you four cars knowing you will keep them, drive them and turn up to its events has learned something real about you, and preferring you for a 799-unit car is not a moral failure. Somebody has to be preferred.
The concern is narrower, and it has a name: Scarcity Leverage. Genuine scarcity is a fact about supply. Scarcity Leverage concerns what a company can induce customers to do because future access is valuable. Once a customer correctly believes that standing in the ecosystem improves the odds on the next limited car, the customer will act on that belief — and nobody at the company ever has to say a word for the belief to shape spending. That some customers buy Ferraris they wanted less in order to protect their standing is reported, and the assessment treats it as structurally supported. It is reported behaviour, not a company requirement, and that distinction carries the whole finding. Reported
Which is exactly why the 2026 Luce episode belongs here rather than in a list of grievances. After reporting characterised the roughly €550,000 Luce as a stepping stone toward limited editions, Ferrari’s chief marketing and commercial officer denied it publicly, called a mandatory-purchase arrangement a “huge mistake,” and warned that it would risk creating negative ambassadors. Fact
CHI records a company’s denial as evidence of its position, never as proof of the underlying fact. But read as a statement of intent, that is not a defence of leverage — it is a rejection of it, and in stronger terms than a bare denial required. So the distinction has to be held: rewarding loyalty is one thing; requiring unwanted purchases in order to remain eligible is materially different, and it is not established as Ferrari’s general practice. What survives is a smaller, more precise observation.
Ferrari explains the system. It does not explain your place in it.
Patek Philippe: candid about the rarity, silent about the rubric
Patek is the most interesting company in the group, because it is simultaneously the most forthcoming and among the least transparent, depending on which question you ask.
Ask why the watch is rare and Patek will tell you, at length, on the record. Its leadership has said that rarity is one of the keys to value, and has explained — specifically, about the reference the market wanted most — that it did not want one model to dominate the collection. That is not a company hiding a strategy. That is a company publishing one. Deliberate suppression of total output, meanwhile, is not established: the production ceiling reported in trade press has risen rather than fallen. Rarity is stated policy for particular models. It is not the explanation for the catalogue. Company claim
Now ask the other question — who gets one — and the candour stops at a very specific point. Patek allocates watches to the retailer. The retailer chooses the client. This is not an inference either; the CEO has described it publicly, and has been blunt about what it means for a disappointed customer. Purchase history and retailer relationships can matter, and at the very top of the range there is evidence of manufacturer involvement in approving some buyers — a comparison-level finding; the Patek assessment itself attributes no specific client decision to Patek corporate.
What does not exist anywhere is a rubric. No published criteria, no queue, no standing, no way for a customer to establish whether they are being considered or merely tolerated. And it is worth being careful here too: a universal Patek corporate spend requirement is not established and is not asserted. The one pleaded case on the subject named a retailer, was dismissed with prejudice with no findings of fact, and did not name Patek as a defendant. Allegation · unadjudicated
Patek doesn’t hide that it values rarity. It hides the rubric for deciding who gets the rare thing.
The other half of Patek is the part that makes a single score impossible to publish honestly. Patek commits to service, repair or restore any timepiece it has made since 1839. It publishes maximum prices for most service categories rather than quoting privately. It keeps a restoration atelier capable of producing obsolete parts. Hard to become a customer; remarkably good once you are.
That ownership proposition is real, and it materially offsets the acquisition-side opacity — but it cannot be used to erase it, and Patek’s own CHI assessment declines to publish a net figure for precisely that reason. The two findings describe different people. The person who never got the watch is not compensated by the service commitment enjoyed by the person who did.
Audemars Piguet: the discretion moved indoors
Audemars Piguet did something structurally distinct, and it is often misread as being more restrictive than its peers when it is really something else.
AP’s scarcity is genuine and not established as artificial: output moved from roughly 32,000 pieces in 2012 to about 52,000 in 2025 on the company’s own figure, a large new manufacture was inaugurated in January 2026, and a materially higher long-term target has been stated publicly. A firm engineering a fake shortage does not spend a decade building the capacity to end it. Equally, the constraint is not purely physical — the 2015 decision to hold output flat for five years was taken with the demand data in hand. Genuine scarcity inside a deliberately controlled growth strategy: not a fake shortage, and not an accident either. Company claim CHI finding
The distinctive move is on the retail side. AP’s points of sale fell from more than 500 in 2012 to 73 by August 2024 on AP’s own website, with wholesale estimated down to roughly 28% of distribution by 2021. CHI does not claim AP eliminated third parties; that would not be accurate. What changed is where the discretion sits. When a customer is told that a steel Royal Oak is not possible as a first watch, that sentence is now being said under AP’s roof, by AP’s staff, inside AP’s system. And AP’s own customer FAQ has no purchasing category at all.
This is where an earlier version of CHI’s own analysis went wrong, and the correction matters more than the finding. An earlier draft effectively imported Rolex’s leverage finding and applied it to AP without equivalent evidence. An audit removed it. Access Leverage was tested at AP and not established — it scores nothing, and it is not quietly reintroduced under another heading. The counterexamples are real: a documented buyer with no prior AP purchase history received a steel Royal Oak, and the two collections customers report being steered toward were trading well below retail in mid-2025, which is not what a functioning leverage machine looks like. CHI finding
What survives is a single, clean finding, and it is the smallest one in this piece: Access Opacity. Existing purchases and relationship development may improve prospects for scarce references, and no customer can determine by how much, because AP has never said. That second clause is the entire charge.
Rolex: the most careful paragraph in the file
Rolex requires more precision than any other company here, because the popular version of the Rolex story is almost entirely unsupported.
CHI does not claim that Rolex manufactures scarcity overall. It does not claim Rolex could simply produce unlimited Daytonas tomorrow. It does not treat aggregate production as proof that any particular reference could be made without constraint, and it does not claim Rolex suppresses production of specific references in order to create shortages. Those propositions are not established, and CHI does not assert them. Not established
The evidence separates into three levels, and they get three different answers. At the level of total production, deliberate suppression is not established: capacity constraints, quality criteria and continuing expansion all cut against it. At the level of the individual reference, the question is unresolved. Rolex chooses its production mix and does not disclose model-level output, so annual production above a million watches settles nothing in either direction — a very large total is compatible with a deliberately thin allocation of one reference, and equally compatible with a genuine model-specific constraint. Neither mechanism is asserted and neither is falsified. An open question is recorded as open, because filing it under “not established” is how a clearance gets manufactured by accident.
At the level of access, the finding is strongly supported, and it is the one carrying the score. A qualified customer can have the money, know the exact reference, walk into the authorized channel, and be refused — with no published queue, no objective priority number, no published qualification criteria, and no reliable explanation of why somebody else received the watch first.
Rolex’s answer is that its Official Rolex Jewelers independently manage allocation and sales, and that is what its own terms say. CHI accepts the delegation as real. It does not accept it as the end of the question, because a company able to govern how its retailers sell is able to govern how transparently they allocate. It is also why reports of bundling at dealer level stay exactly where the evidence puts them — as reports about dealers, not as Rolex corporate policy. Neither a corporate pay-to-play requirement nor a formal ancillary-spending threshold is established. Reported
The evidence does not establish that Rolex manufactures scarcity. It strongly establishes that Rolex delegates the power created by scarcity without giving customers meaningful visibility into the rules.
The assessment records a second supported finding alongside the opacity: Soft Scarcity Leverage. Relationship-mediated access creates a real incentive for a customer to become, and remain, commercially valuable to a retailer — scored as a structural incentive rather than as coercion, with the harder version of that finding expressly reserved. It is precisely the finding AP was wrongly charged with and does not have.
And the observation that costs Rolex nothing to act on: publishing the rules would not give a watch to a single person who cannot have one now.
Hermès: the strongest defence in the cohort, and the narrowest failure
Before anything else, scope. This concerns Birkin and Kelly quota-bag access, and increasingly the Constance and Kelly variants. Roughly 85–90% of what Hermès makes — scarves, ties, perfume, jewellery, ready-to-wear, shoes, homeware, watches, most leather goods — can be bought by anyone, immediately, at a published price that is never discounted. That business is one of the most customer-favourable operations in luxury and nothing here describes it.
Hermès also has the best scarcity defence of the six, and it is worth stating at full strength because it is largely correct. One artisan per bag. Roughly fifteen to sixteen hours per Birkin, by Hermès’ own figure. Eighteen to twenty-four months to train an artisan. Roughly one new workshop a year, with more scheduled. A company engineering a fake shortage does not spend a decade building capacity to relieve it — and it certainly does not leave the money on the table, which Hermès demonstrably does: these bags are priced far below what the secondary market pays within the hour, and Hermès declines to capture the difference. Fact
CHI does not criticise Hermès for refusing to industrialize. It would not want that refusal reversed. The qualification is only that the rate of expansion is a choice rather than a ceiling — leather growth held to roughly 6–7% a year through a decade of excess demand. In any given quarter the constraint binds; over any five-year horizon the growth rate is a strategy. The accurate word is curated, not artificial.
So the scarcity survives scrutiny. The administration is where the finding lives. There is no display, no order book and no direct path to purchase. A customer builds a relationship with one sales associate at one home store, accumulates a purchase history tracked across categories, lodges a wishlist that is not an order, and waits for an offer that may not be the bag requested and typically has to be accepted or declined within minutes. Stock is set twice a year in Paris. The reported cap is two bags per customer per year. Because supply is rationed twice — first to the store, then to the client — an associate can say truthfully that they cannot order you a bag. The system distributes the refusal so that nobody in the building owns it.
Purchase history appears to influence access. Cross-category buying appears to improve it. Both are strongly supported as practice. Neither is proven as written corporate policy — and every specific ratio circulating online, 1:1, 1.5:1, 2:1, 3:1, is folklore. Hermès has never stated one, none has ever been documented, and the customer community’s own reference site says exactly that. Printing a folklore ratio in a serious typeface is how rumour gets promoted to evidence, and CHI does not do it. Reported
Which leaves the actual problem, and it is a narrow one. Customers can spend very substantial sums, believing correctly that purchase history is relevant, while having no usable account of how that history affects allocation. And unlike its peers, Hermès pairs undisclosed criteria with a public position that there is nothing to disclose.
Hermès’ strongest scarcity defense explains why everyone cannot receive a Birkin. It does not explain why customers cannot know the rules governing who does.
None of which cancels the rest. More than 96,000 items repaired in 2025 on the company’s own figure, with no time limit, whatever the age, whoever owns it now, wherever it was bought. Fixed prices, never discounted, never personalised. Special Orders at no customisation premium. A two-bag cap that binds the wealthiest customers hardest. Among the three luxury assessments that carry a customer-value score, Hermès holds the second-highest; of the four that carry a hostility score, it holds the highest. Both figures are load-bearing. Ferrari and Patek Philippe carry no numbers at all, deliberately, so nothing here places them above or below anyone.
Why the more exclusive company isn’t the more hostile one
Put the six side by side and the ranking people expect does not appear.
Ferrari operates the most explicit hierarchy in the group and discloses it in its own investor materials. Patek preserves rarity deliberately and says so on the record. AP caps production strategically and admits it. Hermès refuses to industrialize a hand craft and is right to. Every one of those is a legitimate exercise of a company’s right to decide what it makes and how much of it exists.
None of them, on their own, constitutes hostility.
What does the damage is something orthogonal to exclusivity: an access architecture that causes customers to guess, to perform, to spend, or to cultivate relationships according to rules that are never disclosed. That cost is paid in behaviour, by people who cannot tell whether the behaviour is working. A company can be less exclusive than all of its peers and still impose more of it.
Exclusivity and hostility are different variables.
This is also why the fix, in every case, is small. None of these companies is being asked to make more of anything. None is being asked to abandon selection, treat a first-time buyer identically to a fifteen-year client, or hand its allocation to whoever refreshes a page fastest.
What CHI actually asks
CHI does not demand unlimited Birkins, Kellys, Daytonas, Royal Oaks, Nautiluses or limited Ferraris. It does not demand first-come-first-served allocation, equal treatment of new and loyal customers, the abandonment of exclusivity, or industrialized production.
It asks four things, and all four are within the gift of every company in this piece.
If customer behavior affects access, identify the behavior. Not the weighting, not the algorithm — just the fact. A customer who knows that owning a car counts can decide whether to buy one. A customer who suspects it counts will buy one anyway, and resent it.
If a relationship matters, acknowledge the relationship. Ferrari does this. Patek does this. It has cost neither of them anything, and it is the reason both read as candid rather than evasive even where their criteria stay private.
If production is constrained, explain what can responsibly be explained. Hermès explains its constraint superbly. Lamborghini publishes delivery horizons in months. Neither disclosure created an obligation to supply anyone.
If allocation is discretionary, own the discretion. The discretion is defensible. The denial is what turns a defensible practice into a customer-hostile one, because it tells the person spending toward an outcome that the outcome they are spending toward does not exist.
Scarcity creates value. Opacity creates power. They are not the same thing, and companies that are excellent at the first are not automatically entitled to the second.
I still think being told no is fine. I accept that there are things in this piece I may never be able to buy, and I would not want any of these companies to make more of them so that I could. The question was never whether luxury is allowed to refuse.
Luxury is allowed to say no. CHI asks what happens before, during and after the no.
Customer experience The opening passage is a first-person disclosure by CHI’s founder, offered to establish analytical perspective rather than as evidence. It is not a universal claim about how these companies treat any other customer, and no finding on this page or in the underlying assessments rests on it. Scores referenced are those published by each company’s own CHI assessment as at 15 August 2026; Ferrari and Patek Philippe carry no numeric value by deliberate decision, and Lamborghini’s CHI 18 remains provisional pending cross-company normalization of the luxury cohort.