Hermès

A regulator found the ownership terms unfair

The Korea Fair Trade Commission required Hermès to amend contract terms banning resale and penalising suspected resellers, holding that an owner "should be able to freely decide how to dispose of it." Regulatory action Hermès' e-commerce terms still prohibit purchase intended for commercial resale and reserve the right to limit quantities per customer. A product a customer owns but may not freely dispose of delivers less value than one they do, and Trust, Safety & Reliability is marked down accordingly.

Anti-speculation friction falls on the innocent majority

ID matching at Paris appointments, in-person collection requirements, single shipping addresses, household account consolidation — all defensible in purpose, all experienced by ordinary customers as suspicion. Strongly supported And they demonstrably do not stop the professionals: daigou networks, an $18m-a-year US reseller sourcing bags "fresh from the boutique," and organised criminal rings all persist. Friction applied broadly to catch a few, with the few still getting through, is a value deduction rather than a value delivery.

The below-market price only reaches the winners

The surplus is real — roughly $10,000 a bag on the 2024 figures, less as premiums compressed through 2025 — and it goes to whoever is allocated the bag. Every other customer who wants one faces the same product at an average of roughly 1.4× retail as at late 2025 — and materially more at the 2022 peak — from an intermediary. Fact Hermès' pricing restraint is therefore genuine customer value and the funding source for the reseller economy its own chief executive laments — a tension the company has chosen to live with for decades rather than resolve.

The headline repair figure is the company's own

The 96,000-items figure and the unlimited-duration repair commitment come from Hermès' own corporate publications. Company claim They are consistent with a long, independently observed reputation and with the founding doctrine, and CHI treats them as reliable — but they are self-reported, no independent audit of repair turnaround, cost or refusal rates was located, and the score reflects that ceiling. This is the largest single unverified input into the CVI.

The same object costs 44% more than it did in 2016

Value is delivered relative to price. A Birkin 25 rose from roughly $9,400 to $13,500 across the decade, across five consecutive years of increases well above the company's historical norm, including a US tariff surcharge passed through in full — while the product itself did not change. Fact The craft is the same craft. Core Product Value is scored against the current price, not the 2016 one.

What we looked for and did not find

CHI searched specifically for evidence that Hermès degrades products it has already sold, withdraws functionality, charges to restore included value, operates a subscription, personalises prices, or makes ownership conditional on continued payment. No evidence of any of it was found. Negative finding The absence is recorded here because it is the reason a company scoring 48 on CHI can also score 80 on CVI without contradiction: the harm at Hermès is concentrated almost entirely at the point of acquisition and stops the moment the customer owns the object.

Trust & Transparency — 12 / 15

Proportionally the heaviest dimension on this page, and the correct home for the opacity finding. Hermès publishes no allocation criteria, no eligibility threshold, no queue mechanics, no timing guidance, no explanation to a customer who is passed over, and no list of which products are restricted. The only written customer-facing allocation rule anywhere in the world is a numerical limit posted in Japan. Rule changes — quota expansion to new models, account consolidation, single shipping addresses — reach customers as sudden unexplained shifts. Strongly supported

This sits two points above Audemars Piguet's 10 for a specific and evidenced reason: AP is silent, and silence is not deception. Hermès states publicly that "whatever we have, we put on the shelf, and it goes" and that it prohibits conditioning purchases, while arguing in federal court that prioritising highest-paying customers is lawful and operating a client-vetting apparatus its own chief executive describes. The band anchor for 12–15 — opacity that is systemic and materially affects access and customer decision-making — is met, because customers make five- and six-figure spending decisions in reliance on a criterion nobody will state. Rolex also makes a contrary public claim and scores 10: its denial concerns whether scarcity is a strategy, and the allocation process itself is genuinely run by independent dealers. Hermès denies the character of a process it runs itself.

Three points withheld. Prices are fully published, uniform and never discounted; there is no hidden fee, no personalised pricing and no cancellation trap; Special Order terms are clear and customer-favourable; the Japanese disclosure shows the company can publish a rule when it chooses; and the litigation record itself is public. This is opacity about one commercial process — not a company that misrepresents what it sells or what it costs.

Behavioral Manipulation — 15 / 25

Three distinct mechanisms are scored here, together, once. One: customers are induced into open-ended spending in categories they did not come for, toward a threshold nobody will state — documented at high confidence in the US, China, Australia and Europe, endorsed on the record by a former employee, and structurally reinforced by the Podium's mandatory cross-category buying. Two: offers arrive unannounced with minutes to decide and frequently do not match the customer's stated request. Three: a deliberate long-run restraint on production growth is maintained through a decade of excess demand, for reasons that include desirability. The public denial that accompanies that restraint is scored under Trust & Transparency, not here — it is one act, and it is charged once.

The 15–19 band — behavioural mechanisms central to conversion, creating recurring pressure, uncertainty or dependence — is met at its floor, and the floor is deliberate. The commission asymmetry, which would justify a materially higher score if proven, is an unadjudicated allegation and is not scored as though it were established. Nor is any spend ratio scored: no ratio has ever been documented. Against Rolex's 11, the four-point premium reflects that the currency here is unrelated categories rather than within-category bundling, and that Hermès runs the offer system itself rather than leaving it to independent dealers. That second point is a contrast with Rolex specifically, not a cluster-wide one — AP has internalised more of the relationship than Hermès and scores lower.

Ten points withheld. There is no engineered countdown, no gamified tier ladder, no manufactured deadline, no variable-reward loop and no notification machinery. Prices are not personalised, no purchase is promoted on an invented deadline, Hermès is reported not to gift bags to celebrities as peers do, and the quota cap actively restrains the customers most able to spend. Recent first-person accounts report no penalty for declining an offer.

Customer Restriction — 12 / 20

A customer willing and able to pay the published price cannot buy the product. There is no display, no online listing, no order book, no waiting list, no request path, and a hard cap of two quota bags a year that is enforced globally and published nowhere but Japan. On top of that sits the lock-in: purchase history is visible across the network but credited at the home store, so changing sales associate, boutique or country forfeits years of accumulated standing. Strongly supported Purchase limits tightened measurably between 2024 and 2026 — more models absorbed into quota status, account consolidation, one shipping address, in-person collection.

The 12–15 band — historical value materially restricted, switching carrying significant friction — is met by the non-portability finding alone. Four points above AP's 8: AP has no comparable accumulated standing to forfeit, no quota cap, and no contractual restraint on what an owner may later do with the object.

Eight points withheld, and they are real. The restriction covers a minority of the catalogue: roughly 85–90% of what Hermès makes can be bought immediately, in store or online, by anyone. There is no contract, no subscription, no cancellation penalty, no exit fee, and nothing is withdrawn after purchase — the repair service in particular follows the object, not the buyer, and works for second-hand owners. Genuine production constraint explains a large part of why a given bag is unavailable, and genuine constraint is not scored.

Revenue Extraction — 7 / 25

Five consecutive years of increases far above the company's own ~1.5% historical norm, and above inflation in most of them — roughly 3.5% in 2022, 7% in 2023, 8–9% in 2024, 6–7% in 2025 plus a US tariff surcharge of about 4–5% passed through in full, and 5–6% in 2026 — against a recurring operating margin of 41% and a stated justification of covering production costs. Fact That is what is scored, and it is scored once. Band placement is the largest single judgement call on this page and is recorded rather than buried. The v2.0 10–14 anchor opens with "repeated increases," and on that disjunct alone this record could sit at that band's floor. It is scored at 7 because every other mechanism the anchor lists is absent, because the increases fall on a wholly discretionary purchase at a published price, and because peer consistency requires it: AP earned 7 on a comparable pricing record and Lamborghini 7 on paid equipment packs. A stricter reading would put this at 10 and the total at 51 — and would require re-scoring both of those pages the same way.

The pre-spend money is deliberately not scored here. It is the same underlying conduct already scored under Behavioral Manipulation, and v2.0's anti-double-counting rule prohibits charging it twice merely because it maps to two named patterns. Scoring it in both places would have produced a CHI in the 60s on a single body of evidence.

Eighteen points withheld. No subscription, no fees, no paywalls, no advertising, no tiers, no convenience charges, no personalised pricing, no discount theatre, no charge to restore value, and no customisation premium on Special Orders. Quota bags were sold roughly $10,000 below the immediate dealer bid on the 2024 figures. Of the mechanisms this dimension measures, Hermès triggers exactly one.

Information & Privacy — 2 / 15

Re-derived from zero in a dedicated audit after this page was built, and cut from 5 to 2. The question was never whether Hermès holds customer data. It does: a networked record of what you bought, where and when, visible across the group and linked internationally — Asian profiles to Paris around 2018, US–Paris visibility reported as completed in 2022 Plausible, related accounts consolidated in 2025. Strongly supported The question is what informational harm remains once that record's consequences are returned to the dimensions already scoring them.

Almost nothing does, and the anchor is unforgiving. The 0–2 band is "minimal data necessary to provide the service" — and a purchase record is close to exactly that for a retailer that repairs its products without a time limit and enforces an annual purchase cap. Wider visibility of the same record, inside the same company, for the same purpose, is operational CRM at international scale; it is not cross-service matching, behavioural tracking, inference at scale, advertising, retention expansion or third-party sharing, none of which this file establishes and none of which is invented here.

Two practices survive, and they are the whole of the 2. First, government-ID collection for the Paris appointment lottery — reportedly around 10,000 applications a day against 30–60 slots, meaning passport or ID numbers taken from a very large number of people who will never transact, with no published retention period, purpose statement or deletion commitment. Mechanics strongly supported Second, the 2025 consolidation of related customer accounts, which requires inferring a household relationship between people who registered separately. Strongly supported Both are data practices in their own right, neither is charged anywhere else on this page, and both carry a documented anti-resale purpose that limits how far they can be scored.

What was struck, and why. The earlier 5 rested partly on customers being unable to inspect, correct or port the record behind their standing. The evidence establishes no such refusal — and Hermès trades in jurisdictions where those rights are statutory. An assumption is not a finding. Struck. The 9–11 anchor also lists "personalized commercial treatment," which Hermès unambiguously practises: that is the allocation decision, it is charged once under Customer Restriction, and reaching for it here would be precisely the double count this dimension is most prone to. The secrecy of the criteria is charged under Trust & Transparency; the spending the record induces is charged under Behavioral Manipulation. Three consequences, three homes, none of them this one.

Against AP's 0 and Rolex's 0: AP recorded zero because no customer consequence of its data centralisation was established. Hermès' record does have consequences — but they are scored elsewhere, so they cannot buy points here either. The two-point gap therefore rests only on the ID collection and the household inference, which are data practices AP has no documented equivalent of. If those two items were removed from the file tomorrow, this dimension would read 0.

How this compares — and why it is higher than the cluster

CHI 48 makes Hermès the most hostile company in the luxury cluster by a wide margin: Rolex 28, Audemars Piguet 28, Lamborghini 18. The gap is not a judgement about which brand is more admirable. It is four specific evidentiary differences: (1) the currency of access is unrelated products rather than more of the same product; (2) accumulated standing is non-portable, which no comparator page in this cluster has established; (3) quota limits are real, tightening, tracked globally and published nowhere but Japan; and (4) Hermès pairs the cluster's lowest access transparency with explicit public denials. Internalised control is not among them: Audemars Piguet has internalised more of the customer relationship than Hermès and scores 28.

It is also worth stating plainly what 48 means on this site's scale: it sits just below Netflix at 53 and Spotify at 49, and well below Ticketmaster at 63. Hermès is not among the most hostile companies CHI has assessed. It is the most hostile luxury company CHI has assessed, and at CVI 80 it is also the second most valuable to its customers of the three luxury companies scored for value so far.

Is this conduct already captured elsewhere?

It was, in three places. Pre-spend was scored in both Revenue Extraction and Behavioral Manipulation: six points. The shortage itself was scored in Customer Restriction rather than the discretion layered on top of it: three more. And every consequence of the customer record had already been charged in another dimension before Information & Privacy charged it again: six more. Fifteen of the nineteen removed points are double-counts.

Are we scoring harm, or merely exclusivity?

Exclusivity scores nothing. Hermès is not charged for being expensive, desirable, selective about its retail environment, or for having more customers than products. The harm scored is specific: money spent under an unstated condition, standing that evaporates if you change stores, and a criterion the customer is never told.

Are we penalising genuine production limits?

No — and three points were removed to make sure of it. The craft constraint is verified from primary sources and is explicitly excluded. What remains is the trend-level restraint, scored lightly under Behavioral Manipulation; the denial that any strategy exists is scored under Trust & Transparency, not twice.

Are we treating allegations as facts?

The commission rates, every spend ratio, blacklisting, address-vetting and the "30-minute rule" are all labelled and none is scored. The dismissal is not treated as exoneration and the complaint is not treated as proof. Three points were removed here.

Are positives reflected in CVI?

CVI 80 sits above AP's 76 and just below Rolex's 82. Repairs, below-market pricing, no-premium Special Orders, fixed honest prices and an unrestricted majority catalogue are all scored as value. They do not reduce CHI, because v2.0 forbids netting — they reconcile through CFS, at +32.

Would it survive the cluster comparison?

Tested against all five comparators under the same evidentiary standard. Each of the four gaps that produce the 48 is a difference the comparison matrix independently confirms. The number was derived dimension by dimension, then checked against the cluster — not chosen because Hermès feels more hostile than Rolex.

NOT ESTABLISHED — tested, and unsupported

That Hermès fabricates the shortage. That Hermès requires a fixed spend ratio. That Hermès has a written pre-spend policy. That Hermès profits from the resale premium — it declines to. That Hermès was cleared by a court of requiring pre-spend. That Hermès charges a premium for customisation. That Hermès discounts, surge-prices or personalises prices. That Hermès degrades or restricts products after sale. That customers are penalised for declining an offer. That customers are blacklisted for reselling. None scored

SUPPORTED — and scored

That there is no path to purchase a quota bag, only a path to being offered one. That purchase history and relationship materially influence who is offered one. That the criteria are published nowhere on earth except a numerical limit in Japan. That accumulated standing is forfeited by changing store. That offers are time-pressured and often mismatched. That quota limits are real, tightening, tracked globally and unpublished. That Hermès denies in public what it defends as lawful in court. That list prices have risen far above the company's own historical norm for five consecutive years, and above inflation in most of them, against a 41% margin. Strongly supported or proven

WOULD REDUCE THE FINDINGS

None of these requires Hermès to publish a formula, name a threshold, or abandon discretion.

— Publish, in any market, what factors influence quota-bag allocation, in the way Japan already publishes the numerical limit.
— State the two-bag annual limit and the list of restricted models in writing, worldwide.
— Let a customer see their own client record and their own standing, as they can with any bank or airline.
— Give a passed-over customer a reason, or an escalation route.
— Make purchase history portable between boutiques, which would end the lock-in finding outright.
— Publish the commission structure, or confirm that quota bags carry commission. This single disclosure would resolve the largest open question on the page in either direction.
— Stop saying "whatever we have, we put on the shelf, and it goes" while the offer system operates. Candour costs nothing and would move Trust & Transparency immediately.

WOULD STRENGTHEN THEM

Any of these would move the score materially upward.

— A Ninth Circuit reversal. Briefing is complete as at August 2026. A reversal reopens the only vehicle likely to produce discovery into commission tables and allocation directives, and could move the central finding from strongly supported to proven — or falsify it.
— Production of an actual internal directive on purchase-history gating. Everything observed is equally consistent with incentive-emergent behaviour; only discovery or a whistleblower resolves it.
— Documentary confirmation of the commission rates.
— Verification that the Paris lottery uses "client segmentation," which would mean even the status-blind channel is profiled.
— A named, documented case of a customer terminated for reselling.
— Evidence that access decays without maintenance spending — the Loyalty Penalty finding is currently held at partial for want of it.
— Any French-language regulatory action; the negative finding there rests on English-language sources.

Choice Restriction · Supported, dominant

The purest instance in the CHI corpus to date: the restriction is the retail model, not an edge case of it. A customer willing and able to pay the published price cannot buy the product — no display, no listing, no order book, no waiting list, no request path. Allocation, product matching, timing and eligibility are all controlled through a discretionary offer, capped at two bags a year.

Scope caveat, stated because it matters: this applies to a small part of the catalogue. Roughly 85–90% of Hermès products can be bought by anyone, immediately. The restriction itself is scored under Customer Restriction, 12 of 20; the cross-category spending it induces beforehand is scored under Behavioral Manipulation, 15 of 25 — two harms, two dimensions, neither charged twice.

Deceptive Simplicity · Supported

The publicly legible proposition — walk into a boutique, whatever we have is on the shelf, it goes — is materially simpler than the documented process, which is a multi-year, multi-category, relationship-mediated qualification whose existence the company denies in public and whose legality it defends in court. The evidentiary quality here is unusual: both sides of the gap are Hermès speaking, on the record.

This is where the luxury cluster's provisional Access Opacity concept lives on this page. Scored under Trust & Transparency, 12 of 15.