How Ferrari, Tesla, and Rolls-Royce
Control the Customer Relationship

Section 1

EXECUTIVE FINDING

The central verdict

Corporate control over the customer relationship is not, in itself, a hostility signal — but it is not neutral either. Control determines the ceiling of a customer's exposure. What determines the outcome is the direction the control points, whether the customer retains any procedural standing inside it, and whether a remedy exists that the customer does not have to fight for individually.

On the reachable record:

• Rolls-Royce Motor Cars holds extraordinary control and points almost all of it at the customer's own specification. It has built few of the mechanisms that would allow it to act against a delivered car, and has exercised almost none of the ones it has drafted. Its architecture is the most customer-aligned of the three — with two qualifications that are load-bearing and are stated everywhere in this piece rather than in a footnote.
• Tesla's control extends furthest in time and is the most procedurally concentrated, and it is the only one of the three documented to have reached into a car it had already sold. Its delivered value is the most extensively and independently verified of the three. It is also the only one of the three in which the customer holds no procedural rights whatsoever inside the mechanism that governs what they own. Those two facts are produced by the same architecture.
• Ferrari holds control that is concentrated almost entirely before the transaction, and the concern it produces is not that Ferrari declines people. It is that the criteria for being chosen include commercial behaviour the customer would not otherwise have undertaken, priced in a currency the customer cannot see.

The strongest supported thesis

The working thesis put to this investigation was: customer hostility is not determined by how much control a company possesses; it is determined by how that control is exercised, disclosed, reciprocated, and remedied.

The evidence supports the thesis in its main claim and complicates it in two specific, evidenced ways.

Supported. Rolls-Royce and Tesla each hold exceptional control over their customer relationships — located differently, and not measured against each other for total quantity — and produce opposite outcomes on nearly every dimension. Rolls-Royce's frozen CHI is 23, "Low hostility"; Tesla's frozen CHI is 68, "Concerning." Quantity of control does not predict the finding. And the record's harshest documented individual remedy — profit disgorgement plus the dealer's legal fees on an 18-month resale window — appears in the Ferrari market, imposed by an independent dealer none of the three manufacturers controls. Control and harm are not proportional, and control and location of harm are not the same question.

Complication one: reciprocity does not discriminate, and cannot cure a control defect. Tesla's delivered value is the most extensively and independently verified of the three — a charging network ranked first by J.D. Power for five consecutive years, the lowest ten-year maintenance cost of any brand in Consumer Reports' member data, crash performance validated by three independent regimes, and a decade of free capability added to cars already sold. Its CVI is 81, "Strong." That reciprocity is real, and it does not reduce the hostility finding by a point, because the two are produced by the same channel and score on separate axes. Reciprocity is necessary to a legitimate control architecture and insufficient to legitimate one. The thesis's fourth term therefore needs demoting: exercise, disclosure and remedy discriminate; reciprocity, on this record, does not.

Complication two: concentration converts into a qualitative change. The thesis's strong form — that how much control a company holds is irrelevant — does not survive the Tesla record. Tesla holds the product, the price, the evidence (telemetry), the forum (arbitration with a class waiver), the repair channel, the insurer, the account and the exit simultaneously. No single exercise of any one of those has to be hostile for the customer to have nowhere to stand. Past a threshold, accumulated control stops being a set of separate powers and becomes the absence of customer recourse — and the absence of recourse is a customer harm that no individual act of control produced. This is the one place where the amount of control does the work, and it should be written into the framework rather than argued away.

The most important qualification

The three records do not see equally well, and they do not see the same things.

Rolls-Royce's favourable verdict rests substantially on documents — booklets, terms, releases, regulator data — and only lightly on observed conduct, because the venues where adverse conduct would be visible (owner forums, complaint databases, non-US regulators, confidential settlements, sealed dealer agreements) were unreachable. The dossier states this as an evidence-insufficiency finding, not an exoneration, and this piece keeps that framing.

Ferrari's record is narrow by design and assesses nothing at all on price stability, service, failure or remedy. Ferrari is not clean on those dimensions; it is unexamined on them.

Tesla's record is the most complete, and its completeness is partly a function of scale, litigation and press attention that the other two do not attract. A comparison that counted incidents would find against Tesla for reasons that are partly artefacts of visibility. This piece therefore compares architectures, terms and mechanisms — which are equally observable across all three — and treats incident counts as the weakest evidence in the record.

Why the comparison matters to CHI

CHI's standing objection is that it is a machine for disliking powerful companies. The three-company comparison is the strongest available refutation, because all three subjects hold exceptional control and the findings still diverge. It is not a controlled experiment: the records do not supply a common measure of corporate power, and Ferrari's is too narrowly scoped to be quantified against the other two. What the comparison holds steady is the presence of exceptional control, not its amount.

All three companies exercise control that would be intolerable in an ordinary consumer market — and each exercises it in a different place. Ferrari's documented control is principally pre-transactional and relational: it decides who may buy its scarcest cars, and standing in its hierarchy survives the sale. Tesla's is principally post-purchase, technical and procedurally concentrated: it can alter a delivered car, and it holds the evidence, the forum and the remedy. Rolls-Royce's is principally access-, production- and specification-based — who is invited, what is built, and on what terms it is supported — with some reserved contractual powers it has not exercised and platform-level technical capability held one level up at BMW Group. One of them produces a frozen CHI of 23 alongside a CVI of 85; one produces a named structural concern with the score deliberately withheld; one produces a frozen CHI of 68 alongside a CVI of 81. Two of the three are now scored, and the scores still do not settle the comparison — they are calibration context, not a league table, and Ferrari's deliberate absence from the numbers is itself a finding. If CHI simply punished power, those three results would not be available from those three inputs.

The comparison also does something CHI has not previously been able to do: it isolates the mechanisms that actually discriminate. Of the eleven candidate legitimacy factors tested in Section 6, only six do real work on this evidence. Two are necessary but weak on their own. One is untestable on any of the three records. And two behave in ways the thesis did not anticipate: reciprocity, which turns out to be a threshold rather than a determinant, and ability to exit, which counts only when the thing lost at exit was sold to the customer as theirs. That result is more useful to the index than any ranking of the three companies.

Why these three companies legitimately belong in the same research piece

They are not competitors, and this piece never treats them as competitors. A Model 3 buyer and a Phantom commissioner are not choosing between two options.

They belong together because they are the three clearest available specimens of the three ways a manufacturer can hold power over a customer, and because each is close to a type specimen of its architecture:

• Ferrari is the cleanest available case of control exercised through scarcity and permission — genuine, published, physically real scarcity, openly declared as strategy, allocated through an unpublished hierarchy.
• Tesla is, by the index's own designation, the type specimen of Digitally Contingent Ownership and Promissory Product Monetization — the first mass-market case in which a manufacturer's leverage over what it sold grows after delivery instead of ending at it.
• Rolls-Royce is the corpus's cleanest example of curated scarcity honestly disclosed, and the only company in this comparison whose control apparatus is aimed predominantly at producing the object the customer specified.

Because each is near-pure, the comparison can isolate what changes when only the direction and timing of control change. Two ultra-luxury manufacturers with almost identical access architecture reach different findings; two companies with almost opposite market positions reach the same conclusion about disclosure. That is only visible when the companies are chosen for their architecture rather than their showroom.

↩ Back to the brief

Section 2

COMPARATIVE OPENING

The paradox

Each of these three companies exercises a degree of control over its customers that would be considered outrageous almost anywhere else in consumer markets.

Ferrari decides who is permitted to buy its most desirable products, on criteria it has never published, and its own disclosed sales mix shows how dominant the existing-owner channel is: in 2025, 84% of Ferrari sales went to existing Ferrari owners and 56% to customers who already owned more than one — a company figure reported by Reuters in June 2026, with the 2023 and 2024 equivalents in those years’ Annual Reports. Those figures do not by themselves prove a ranking system; they establish that the existing-owner base absorbs the overwhelming majority of output. A financially qualified stranger cannot buy an F80. There are 799 of them.

Tesla can alter a car it has already sold, through a channel the owner cannot decline, cannot audit and cannot reverse — the company's own owner's manual states that "reverting to a previous software version is not possible," and warranty coverage is conditioned on accepting updates. Tesla has used that channel to add a dashcam, Sentry Mode, Dog Mode and a five per cent power increase to cars already in customers' hands, free. It has also used it to cap the voltage of aging Model S battery packs, and to inform Model 3 Standard Range owners that range, acceleration and Autopilot delivered with their cars would be removed unless they paid $4,500.

Rolls-Royce decides which clients are invited into Coachbuild at all — "The invitation-only service offers individuals of extraordinary achievement, culture, and vision to craft an entirely original motor car" — publishes no price for any car, option, service plan or umbrella anywhere in the world, and reserves, in live contract terms, the right to charge its owners' club members a joining fee, an annual subscription and a termination fee, to refuse membership "for any reason or no reason," and to hold members "accountable" if they make negative statements about other members, partners or events in public.

Three companies exercising exceptional control over their customers — in three different places in the relationship. Three materially different customer outcomes. That is the paradox this piece exists to resolve, and it cannot be resolved by measuring how much power each one has. The records supply no common quantitative measure of corporate power, and Ferrari's is scoped too narrowly to be weighed against the other two. What this comparison tests is the direction, duration, location and contestability of control — not its total quantity.

What this comparison is not

It is not a comparison of cars. Performance, acceleration, specifications, price tiers, market share and product quality appear in this document only where a fact about them bears directly on the customer relationship — Tesla's charging network appears because it is a component of the product the customer bought and a demonstration of what vertical integration delivers; Rolls-Royce's instrumented range and noise results appear because they test whether the company's representations are true, not because they are good numbers.

It is not a ranking of companies by desirability, and it does not conclude that any customer should buy or avoid any of them.

It is not a claim that the three occupy one market. Ferrari shipped 13,640 cars in 2025. Rolls-Royce delivered 5,664. Tesla operates at a scale neither approaches, in a mass market with a mass market's regulatory surface and litigation volume. They are compared on architecture, not on category.

The working formulation, and its correction

The investigation opened with a formulation that is rhetorically clean:

Ferrari controls whether you may buy the car. Tesla controls what the car you bought may become. Rolls-Royce uses its control to help determine what your car will be.

Tested against the record, the middle clause survives intact, the first clause is incomplete, and the third clause is accurate about what Rolls-Royce has done and silent about what it has reserved the right to do.

Ferrari. "Whether you may buy the car" understates the finding. Ferrari's control over access does not stop at a yes or no on one car; it reaches backwards into the customer's other purchasing decisions, because the criteria for standing include spending, retention and engagement that are separable from the scarce car itself. Ferrari's own clients found "buy this to protect your standing" a plausible description of their relationship — which is why a June 2026 report required a corporate response at all. Ferrari's control is over the terms on which you become eligible, and eligibility is fed by behaviour.

Tesla. "What the car you bought may become" is exact, and if anything conservative, because the same clause should also read what it may cease to be. Battery capacity, acceleration, lane centering, network access and — through non-transferability rules — resale value are software states rather than fixed properties of the object.

Rolls-Royce. "Uses its control to help determine what your car will be" is supported: Bespoke content now runs at roughly 40% of car value, uniqueness undertakings have been externally observably honoured across five to nine years, and refusals are documented as safety- or legality-based only. But the clause is silent on two things the record contains — a fully drafted legal architecture that has never been used, and one categorical corporate promise that was reversed.

Corrected formulation, carried for the rest of this piece:

Ferrari's control decides whether you are permitted to buy — and quietly prices what that permission costs in behaviour.

Tesla's control decides what the thing you already bought will be allowed to remain.

Rolls-Royce's control is aimed, so far, at building what you asked for — and holds in reserve powers it has drafted and not used.

The phrase so far is doing deliberate work. It is the honest form of the Rolls-Royce finding, and it is carried in the framework as the middle state of Factor 1 — capability present but unused, which is a watchlist condition rather than either a hostility finding or a clearance (Section 6.2).

↩ Back to the brief

Section 3

THREE ARCHITECTURE PROFILES

Each profile answers the same six questions: where the power comes from, what the customer gives up, what the customer gets, the strongest alignment evidence, the strongest hostility or asymmetry evidence, and the shape of the resulting relationship.

3.1 Ferrari — Scarcity and Permission

The source of corporate power. Physical scarcity that is genuine, published and admitted. Ferrari shipped 13,640 cars in 2025; the F80 is a 799-unit run and the Daytona SP3 a 599-unit run, both confirmed at unit level; the order book extends towards the end of 2027. Ferrari's own FY2025 results describe demand as "managed with discipline in every market reflecting our exclusivity model" and 2025 shipments as "deliberately designed to be substantially flat." The scarcity is real, the strategy is disclosed, and CHI rejects Artificial Scarcity here rather than merely failing to prove it.

The power does not come from the shortage. It comes from what sits on top of it: a five-tier ladder running from regular production (order and wait; genuinely open to newcomers) through dealer-discretionary high-demand cars, invitation-only special series, curated Icona client lists, and finally the hypercar tier, where "selection criteria are not published." At the top of that ladder Ferrari holds the most valuable thing it can offer a customer, and the criteria for receiving it are known to be relationship-weighted and are nowhere stated in terms a customer can check.

The autonomy the customer surrenders. The ability to buy the product they want by being willing to pay for it. Beyond that — and this is the finding — the ability to make purchasing decisions about other Ferraris on the merits of those cars alone. Reported factors in allocation standing include prior ownership, the number of cars owned, retention rather than rapid resale, service history with the dealer, participation in Ferrari's client activities, and general brand engagement. Each of those is a commercially valuable behaviour, and none of them is the scarce car.

The value received in return. Access to a genuinely rare object; membership of a community whose exclusivity is part of what was purchased; a strategy the customer can read in the company's own filings before spending a euro; and the protection that anti-flipping controls give to buyers who actually want to keep the car. The disclosure point is substantial and CHI treats it as a mitigating factor: Ferrari says out loud what most companies accused of manufacturing scarcity deny.

The most important alignment evidence. In June 2026, after Bloomberg reported that Ferrari had signalled to top clients that buying the Luce electric model could function as a "stepping stone" toward more desirable limited editions, Ferrari's Chief Marketing and Commercial Officer Enrico Galliera went on the record within five days denying that Luce purchase conditions access to limited editions, warning that Ferrari would "run the risk of creating negative ambassadors," and calling a mandatory-purchase policy a "huge mistake." A company that wanted to operate a spend-to-qualify system had an easy alternative: say nothing. Read as an internal control, that response places Ferrari's own line in almost exactly the place CHI draws it. Nine separate counterevidence findings support Ferrari on this record, against one central concern.

The most important hostility or asymmetry evidence. The concern survives the denial because the denial addresses policy and the concern is structural. For the Bloomberg report to have been credible enough to require a corporate response, Ferrari's own top clients had to find "buy this to protect your standing" a plausible description of how their relationship works. A hierarchy fed by commercial behaviour, sitting on top of a product customers badly want, produces the incentive whether or not anyone states a requirement. Ferrari explains the system. It does not explain your place in it, what would improve it, or whether any given purchase counted.

The resulting customer relationship. Voluntary, asymmetric, and priced in a currency the customer cannot read. Ferrari's qualitative verdict is Genuine Scarcity — Scarcity Leverage Concern, with CHI, CVI and CFS deliberately reserved rather than manufactured. The relationship is not one of dependency — an owner who leaves loses standing, not function — and it is not one of deception. It is one in which the price of proximity to the thing you want is denominated in behaviour, and the exchange rate is never published.

3.2 Tesla — Software and the Promised Future

The source of corporate power. Two architectural choices, both dual-use. Over-the-air control makes the car's material properties — battery capacity, acceleration, lane centering, network eligibility — software states held on Tesla's servers rather than fixed properties of the object in the customer's title. Vertical integration removes the dealership and concentrates pricing, evidence, adjudication, parts, repair, software, insurance and remedy inside one firm. Tesla operates at least seven distinguishable customer relationships — new vehicles, vehicle software, charging, service, insurance, energy, and the account layer — and in every one of them controls more of the transaction, the product, the data, the remedy and the exit than any comparable company in its sector.

The account layer is the structurally novel part: nothing is purchased there, and everything is mediated through it — app and key access, camera and telemetry data, data-release decisions, and the account linkage of purchased features. Its exit route is "none. The account is the ownership interface."

The autonomy the customer surrenders. The right to refuse a change to a product they own. Tesla's owner's manual states that "reverting to a previous software version is not possible"; warranty coverage is conditioned on accepting updates; harm resulting from failure to install an update is not covered. The customer therefore cannot decline the channel, cannot audit what an update changed, and cannot return to the prior state. There is no procedural right of any kind inside the mechanism. Beyond that: the forum, through arbitration with a class waiver whose documented effect has been class extinguishment; the evidence, because Tesla holds the telemetry that decides most disputes about its own products and releases subsets of it selectively; and portions of the exit, because free unlimited Supercharging was made non-transferable, FSD is non-transferable by default, and salvage vehicles are permanently locked out of fast charging.

The value received in return. The most extensively and independently verified value in this comparison. The highest-rated DC fast-charging network in the industry, ranked first by J.D. Power for five consecutive years in a market where roughly one in five public charging visits ends with no charge delivered. The lowest ten-year maintenance and repair cost of any brand in Consumer Reports' member data — $4,035 against $4,900 for the next-cheapest — in a category where the dealer service department is the profit centre. Crash performance validated by NHTSA, IIHS and Euro NCAP. A decade of free over-the-air additions to cars already sold. Purchase without a dealership: posted national pricing, no finance-and-insurance office since 2012, and no markup over list even through the 2021–22 shortage. Recall remedies that install fleet-wide within days at zero owner effort. CVI 81, "Strong."

The most important alignment evidence. The dual-use channel at its best, and the honesty of the record about it. Sentry Mode, Dog Mode, dashcam, a blind-spot camera and a 5% power increase all arrived free on cars already sold. Software-limited packs were unlocked remotely for Florida evacuations in 2017 and free evacuation charging became standing practice. Retroactive generosity is documented: the 2014 infinite-mile drive-unit warranty applied to cars already sold; lifetime connectivity promised in 2018 is still honoured in 2026; free HW2/HW2.5-to-HW3 computer retrofits were promised and delivered. No federal civil penalty or consent order against Tesla was located across 2012–2026, against comparators who paid tens and hundreds of millions. And owners who experienced the flaws bought again — first in Consumer Reports owner satisfaction 2016–2020, highest brand loyalty in S&P Global Mobility's rankings 2021–2023 — before loyalty measurably declined in the wake of the resale-value damage.

The most important hostility or asymmetry evidence. Two exhibits, both structural rather than anecdotal.

Promissory Product Monetization. Full Self-Driving was sold from October 2016 at prices laddered from $3,000 to $15,000, under messaging that the price would rise as delivery approached. The launch demonstration video was staged — established by the sworn deposition testimony of Tesla's own Autopilot software director. Hardware sufficiency was represented across three generations; in January 2025 the chief executive conceded that HW3 cars, roughly four million of them, cannot deliver the product, and as of the July 2026 earnings call the free upgrade promised in its place had no programme, timeline or cost. No refund policy has ever existed. The refunds that exist were won one at a time in arbitration and small-claims actions — each win showing the claim had merit and simultaneously showing that no general remedy was available. In December 2025 the California DMV adopted a finding that the marketing was "actually, unambiguously false"; Tesla complied in February 2026 and sued to erase the finding, which is therefore contested rather than settled, and is a state administrative adjudication, not a fraud judgment.

Digitally Contingent Ownership.* Voltage caps applied over the air to delivered Model S packs produced a judgment against Tesla in Norway's Supreme Court in April 2026 — NOK 50,000 to each of 115 owners after a five-year, four-instance fight — and, on the same conduct in the United States, a $1.5 million class settlement without admission, or $625 per owner. In June 2019 Model 3 Standard Range owners were emailed that an update would remove range, acceleration and Autopilot delivered with their cars unless they paid $4,500. In February 2020 roughly $8,000 of purchased Autopilot was remotely removed from a used Model S that had passed through Tesla's own auction, and restored only after press coverage. In July 2022 a used Model S was re-locked by roughly 80 miles and $4,500 demanded to restore capacity the owner had bought; that too was reversed after the case went viral. **The recurring shape is the finding: the remedy arrived because the case became public, not because a process existed.

The resulting customer relationship. High-value, high-dependency, and procedurally hollow. The Tesla page's own summary is the accurate one: the car improves after you buy it; the deal degrades after you sign it. CHI 68 / CVI 81 / CFS +13 — "Finely Balanced," which is not a description of a company of middling consequence but of a very large positive and a very large negative netting out.

Entity controlRolls-Royce here means Rolls-Royce Motor Cars Limited, the BMW-owned automobile manufacturer — not Rolls-Royce plc / Rolls-Royce Holdings plc, the separate aerospace, defence and power-systems group, which is out of scope and is described nowhere on this page.

3.3 Rolls-Royce Motor Cars — Bespoke and Service

The source of corporate power. Capacity that genuinely binds — 25 to 26 cars a day on a single 190-metre line, 5,664 delivered in 2025 — combined with a positioning the company states without euphemism: "Rolls-Royce is not and never will be a volume manufacturer," with the business "steered by profit contribution, not by volume, with scarcity and rarity essential." Above the series range sit pre-allocated Private Collections and an invitation-only Coachbuild and Private Office channel. No price for any car, option, service plan or accessory is published anywhere. Since 2022 the company has operated five Private Offices directly, sitting over an independently owned dealer network — 135 points in more than 50 countries at the last corporate count, in August 2020, with no later global figure published.

There is a second, quieter source of power, and the dossier is careful about it: several of the levers a hostility analysis cares most about sit one level up, at BMW Group — the aluminium body-in-white built at Dingolfing, the V12, the Spectre drivetrain, the ConnectedDrive telematics backend for which BMW AG is contractually joint data controller, the financial services division, and US recall administration, every campaign of which is filed by BMW of North America. The customer contracts with a dealer, is warranted by Rolls-Royce, and depends on BMW Group for a material part of what makes the car work.

The autonomy the customer surrenders. Access to the top tiers, which is by invitation and by disclosed relationship. Price transparency, comprehensively, before the dealer conversation. And — on paper — a substantial bundle of rights inside the owners' club and the connected-services stack: Whispers terms reserve joining, subscription and termination fees, admission at the sole discretion of a Membership Committee "for any reason or no reason," termination without notice on sale of the car, and a "strict no press policy" under which members "may be held accountable if they disclose, make negative statements about or identify other members" in public; the connected-services terms permit termination on technology change with no retrofit duty, impose AAA arbitration with a class waiver on the telematics layer, and cap aggregate liability at the greater of twelve months' fees or $100 while releasing claims for personal injury arising from failure of services that include emergency call.

The value received in return. The strongest written ownership offer among the checked comparators: a four-year, unlimited-mileage warranty with a four-year maintenance programme that includes wear items — brake pads and rotors, wiper inserts, oil services — at "no expense to you," plus four-year roadside assistance. Bentley's is three years with no currently published maintenance term — though a 2020 UK service-plan document bundled a plan into the list price of some models; Mercedes-Maybach's is 48 months capped at 50,000 miles with maintenance excluded. Ferrari's seven-year Genuine Maintenance programme is the single qualifying comparator and must always be named: it runs longer, on a shorter warranty, and excludes wear items.

Beyond the warranty: a certified pre-owned programme (Provenance) bundling up to two years' unlimited-mileage warranty including trim, two years' servicing, two years' roadside and club access; genuine authorship, with Bespoke content running around 40% of car value and Private Office commissions typically 25% higher again; and no digital extraction at all — no feature paywall or subscription exists in any fetched Rolls-Royce contract or press material, connectivity is included for four years — a term fixed by the 2022 Assist subscriber agreement rather than by the current North American terms, which state no duration — and over-the-air updates on Spectre are owner-launched rather than forced.

The most important alignment evidence. Three exhibits, in order of strength.

Honoured exclusivity, externally observable. Against a written undertaking that every Coachbuild project "will never be replicated," Sweptail remains one car, Boat Tail three, Droptail four, across five to nine years. This is the rare favourable claim that does not depend on believing the company: non-replication is checkable from outside, and it sits above an industry pattern of "one-off" design cues reappearing.

Representations that independent testing met or beat. Edmunds' instrumented loop returned 281 miles against a 266-mile EPA figure for the Spectre and 298 against 251 for the Black Badge — with the mandatory deflator that BMW Group EVs generally beat EPA on that gentle loop — and made the car "officially the quietest vehicle Edmunds has ever tested" at 35.9 dB idle and 56.4 dB at 70 mph. The Advertising Standards Authority rulings database returns zero entries for Rolls-Royce Motor Cars, in a period when sibling entity BMW (UK) Ltd was censured for a "Zero Emissions Cars" claim.

Retroactive early-adopter protection. On 5 February 2026 Rolls-Royce announced a 15-year, unlimited-mileage battery warranty "offered as standard to new and existing Spectres" — the longest documented term among Western luxury makers, codified in the MY2027 booklet with 80% state-of-health floors to year 10 and 70% to year 15. It is credited in full under the index's Fixed Under Pressure principle, with the pressure documented: it came after Spectre deliveries fell 47% to 1,002 in 2025, six weeks before the electrification reversal became public, and four months before a Series II battery-chemistry change with no retrofit path for Series I cars. The substance counts; the "confidence in the product" framing does not, and "unprecedented" is the company's own adjective, not adopted here.

The most important hostility or asymmetry evidence. Four items, none of which is conduct against an identified customer.

The 2030 reversal — an unretracted categorical commitment, reversed; no CHI pattern established. Rolls-Royce pledged unconditionally, twice, at primary-source level: "By then, Rolls-Royce will no longer be in the business of producing or selling any internal combustion engine products" (29 September 2021) and "Rolls-Royce will never again produce a new model with an internal combustion engine" (Spectre press kit, 4 July 2023). In March 2026 the chief executive abandoned the 2030 target in on-record interviews with two national newspapers. On re-adjudication against the index's published Promise Reversal definition, the classification does not hold and is excluded: the definitional threshold is not met, because a corporate statement about what the company will stop building is not a term of the customer bargain; and none of the three conjunctive recognition elements is established — no source shows the date presented as a customer proposition and no relying buyer is identified; no entitlement, feature, warranty term or specification was withdrawn from any delivered car; and no owner had to pay more, change behaviour or leave. No substitute pattern is asserted; no CHI Lexicon pattern is established for this company. What survives is scored, narrowly, as a stated-promise-versus-later-behaviour gap under Trust & Transparency, at 2 of 15 points. Three qualifications remain binding: the reversal was not silent; the maximum defensible criticism is that no corporate press release announcing it was located and the press archive still carries the superseded pledge, uncorrected — the most durable element of the finding; and the "never again" formulation has not been falsified, since no all-new internal-combustion model has launched. The most exposed cohort received a retroactive enhancement rather than a withdrawal.

The July 2023 access-control rhetoric. At the Spectre launch, the then chief executive told Car Dealer Magazine that customers reselling for profit "are going immediately on a blacklist and this is it — you will never ever have the chance to acquire again," and that "you need to qualify for a car and then you might get a slot for an order." Verbatim, corporate, unusually blunt — and supported only as a statement. No operating policy was established. The adjudicated formulation is the one carried here: the 2023 rhetoric was aimed at the resale stage, was not repeated by the succeeding chief executive, and has no documented enforcement instance. No written resale restriction was located anywhere; no written policy or contract clause was located in three-plus years; no customer is documented as refused or blacklisted; an independent trader publicly rejected the posture and bought two Spectres from customers anyway; and by December 2025 used Spectres were trading well over a hundred thousand dollars below original sticker. Conquest data cuts the same way: 80% of Cullinan buyers and 40% of Spectre buyers were new to the brand, which weakens a closed-clientele reading without disproving screening. The evidentiary boundary is binding and is stated rather than softened: this is a failure to find, not proof that no blacklist was ever operated. The statement was never withdrawn, was never given contractual force, was never shown to be enforced, and cannot be described as having lapsed, ceased to exist, or been formally retracted.

A drafted-but-unexercised legal architecture. The Whispers clauses above are the surviving exhibit, carried by the record as a latent-terms watchlist item. The connected-services stack is weaker still: on adversarial review it fell as a Rolls-Royce-specific signal, surviving only as low-weight industry context, because it is standard US telematics drafting on BMW Group's template. In both cases the finding is the option, not the exercise. No fee has been charged, no membership refused, no member sanctioned, no service terminated on technology change. Eligibility has only ever widened — in 2020 to all Goodwood-era owners and their partners, and later to Provenance certified pre-owned buyers.

Information asymmetries with real edges. Total price opacity — deliberate, systemic, and exactly sector-standard, with Ferrari's and Bentley's configurators also displaying no prices, and substantially cured in the largest market by Monroney labels that Rolls-Royce completes line by line down to $1,900 lambswool footmats. A Provenance programme with no published inspection-point count or accident-disclosure standard, whose one documented failure — Wynn Holdings v. RRMC NA & Towbin — turned on a dealer's withholding of accident documents, survived summary judgment, and settled without a liability finding, incidentally proving the Provenance promise is court-enforceable. A 40-month interval between an internal crash test that revealed the Ghost instrument-cluster glass could shatter and the recall notice to 1,305 owners — on a part that remained legally compliant throughout, recalled voluntarily, with NHTSA accepting the chronology and no injuries reported. One rebuffed attempt, in 2021, to compel a lemon-law claimant into a dealer's arbitration clause, which the court denied and which has not been repeated.

The resulting customer relationship. High control, low exercise, dense delivered value, and — the honest qualification — comparatively little independent visibility into how it feels from inside. The dossier's own summary is that the extraction surface a CHI analysis probes is "either absent, reserved-but-unexercised, or milder than segment peers," while roughly a fifth of the customer relationship sits behind sources the investigation could not reach.

↩ Back to the brief

Section 4

STANDARDISED COMPARISON MATRIX

4.1 The calibration vocabulary

Findings below use five bands, in ascending order of concern, plus one non-finding. They are not scores and they do not aggregate.

4.2 The matrix

Dimension 1 — Access and eligibility

Dimension 2 — Customer authorship

Dimension 3 — Pre-purchase corporate control

Dimension 4 — Post-purchase corporate control

Dimension 5 — Promise versus delivery

Dimension 6 — Price and value stability

Dimension 7 — Service, failure and remedy

Dimension 8 — Exclusivity and dependency

Dimension 9 — Disclosure and intelligibility

Dimension 10 — Reciprocity

4.3 How to read the matrix

Three cautions govern it.

Do not count the bands. Rolls-Royce accumulates more favourable cells than the others, partly because it has more assessed dimensions than Ferrari and a less adversarial evidence environment than Tesla. Counting cells would reward the record's shape rather than the companies' conduct.

Do not read UNASSESSED as clean. Ferrari's blank cells on Dimensions 6 and 7 are a scope decision made for a different investigation. They are the largest single gap in this comparison and are named again in Sections 8.7 and 10.7.

Do not treat these findings as equivalent magnitudes. Tesla's Dimension 5 is DEMONSTRATED CUSTOMER HOSTILITY: a decade-long representation record with a regulator finding attached. Rolls-Royce's Dimension 5 is ASYMMETRIC CONTROL: an unretracted categorical strategy commitment was reversed, but no material customer detriment, buyer reliance or owner consequence was established. The categories differ because the evidentiary thresholds differ. The band names the category; the severity is in the cell. (Revision 2 recorded the Rolls-Royce cell as "demonstrated at low harm weight"; that treatment is superseded and is not the current classification.)

↩ Back to the brief

Section 5

MATCHED COMPARATIVE ANALYSIS

Organised around eight common questions. Each subsection asks why mechanisms that look alike produce outcomes that are not alike.

5.1 Permission to buy

The apparent similarity. Ferrari and Rolls-Royce both operate invitation ladders. Both decide, above a certain product tier, which customers are offered what. Both have chief executives on record describing selection in terms that would read as arrogant anywhere else: Ferrari's tiers 03 to 05 are invitation, curated list and unpublished selection; Rolls-Royce's Coachbuild is "by invitation only… offered to clients with a deep affinity for Rolls-Royce design," and its then chief executive said in July 2023 that "you need to qualify for a car and then you might get a slot for an order."

Why the consequence differs. The two ladders rank on different things, and that is the entire difference.

Rolls-Royce's disclosed criteria are affinity, engagement with the design process and willingness to participate — "you can't send your chauffeur"; "we don't do that, because I want to do Coachbuild with clients we really know." Those are criteria about the commission. They select for a client who will be a good counterparty to a bespoke project running from months to years — standard commissions at six months to two years, coachbuilt one-offs at around four. Meeting them costs the customer time and attention, not additional purchases. And Rolls-Royce's own conquest data undercuts any closed-shop reading: 80% of Cullinan buyers and 40% of Spectre buyers were new to the brand, with an incoming Spectre buyer average age of 35.

Ferrari's reported criteria are prior ownership, number of cars owned, retention rather than rapid resale, service history with the dealer, participation in client activities and general brand engagement. Every one of those is commercially valuable to Ferrari and separable from the scarce car. Ferrari's disclosed sales mix confirms the channel's dominance rather than merely its existence: 84% of 2025 sales to existing owners, 56% to multi-Ferrari owners (company figure via Reuters, June 2026).

That is the step at which allocation becomes leverage. Not selection — selection is unavoidable arithmetic when 799 cars exist and more than 799 qualified buyers want one. Not loyalty preference — rewarding a collector who carried depreciation and supported weaker models is ordinary commerce and appears across many industries without scoring. The step is that "loyalty" is measured in commercially valuable behaviour whose price the customer cannot see.

Tesla's inversion is the control that proves the point. Tesla's access architecture is the most customer-aligned in the comparison: a posted national price, no dealership, no finance-and-insurance office, no markup over list even through the 2021–22 shortage when buyers of competing electric vehicles paid $10,000–$30,000 over sticker, and litigation for direct sales in states that barred it. A company can be near-perfect on permission to buy and still hold the deepest hostility finding in the comparison — because permission to buy is a pre-transactional dimension and Tesla's concerns are almost entirely post-transactional. Any framework that weighted access heavily would rank Tesla first and be wrong.

The Rolls-Royce statement that was not established as a practice. The July 2023 "blacklist" quote is the most-quoted hostility exhibit in the Rolls-Royce corpus, and the adjudicated formulation is that the 2023 rhetoric was aimed at the resale stage, was not repeated by the succeeding chief executive, and has no documented enforcement instance: no written resale restriction was located anywhere; no written policy, clause or enforcement instance in over three years; the speaker retired five months later; an independent trader publicly rejected it and bought two Spectres from customers anyway; and by December 2025 used Spectres were listed at $385,575 against a $521,575 original and $385,150 against $543,150. That is a failure to find, not proof that no blacklist was ever operated, and it is neither a withdrawal of the statement nor a finding that the posture ceased to exist. The comparison CHI must resist is the easy one — both companies restrict resale, therefore both are equivalent — because one company said something that was never shown to have been acted on, while a written instrument with a heavy remedy exists in the other's market at the dealer layer. Those are different facts with different customer consequences, and Section 5.6 takes the remedy point up directly.

Finding. Ferrari's permission architecture is the only one of the three where the criteria for permission reach back into unrelated purchasing behaviour. That, not restriction itself, is the discriminating fact.

5.2 Authority to define the product

The apparent similarity. All three companies define what the customer may have. Rolls-Royce holds design authority over every Bespoke commission; Ferrari controls specification of its limited series absolutely; Tesla defines the feature set, the hardware requirement and what is enabled at delivery.

Why the consequence differs. The question is whether the company's definitional authority is exercised toward the customer's stated preference or toward the company's revenue, and the record separates the three cleanly.

Rolls-Royce is the clearest case in the corpus of definitional authority used for authorship. Bespoke content runs around 40% of car value and is spent voluntarily by a price-insensitive clientele, itemised line by line on the US window sticker. Client-developed paint and leather hues are "reserved exclusively for their use." Refusals are documented as safety- or legality-based only — "We very rarely have to say no to anything… we're not the taste police." And the uniqueness undertaking is the rare corporate promise a third party can check: Sweptail remains one car, Boat Tail three, Droptail four, over five to nine years. The honest residue is that the commercial terms behind Bespoke — deposits, cancellation, change orders, whether a reserved colour stays reserved — are published nowhere, and the dossier records that as an information asymmetry that may be fully cured at the dealer's quotation and cannot be confirmed either way.

Tesla holds the same authority and uses it in both directions through one mechanism. Physically installed, software-disabled capability is a standing architecture, not an incident: battery capacity has been software-locked since the Model S 40 and 60; acceleration has been sold as a $2,000 unlock; from October 2025 Autosteer hardware ships in every new US Model 3 and Model Y while lane centering requires the $99 monthly subscription. Disclosed in advance and priced accordingly, that is a legitimate commercial architecture — the CHI relevance is what it demonstrates about where the product's boundary sits. What crosses the line is the reverse operation: deleting a control and then selling it back. A $595 retrofit for a deleted stalk and $700 for a round steering wheel are not personalisation; they are restoration priced as an option.

Ferrari* is unassessed here, and the gap should be stated rather than filled. The Ferrari page examines allocation, not configuration. Tailor Made appears once, in an auction citation. **Nothing in this comparison establishes how much authorship a Ferrari customer holds, and nothing establishes that they hold little.

Finding. The same authority produces authorship at Rolls-Royce and monetised complexity at Tesla's margins. The discriminating variable is not the amount of definitional power but whether the customer's specification or the company's price list is the thing being served.

5.3 Authority retained after purchase

This is the dimension on which the three companies are least alike, and it is the dimension that decides the piece.

Ferrari retains relational authority. Standing in the hierarchy can be lost — by flipping a car, or by never accumulating it. That is a genuine and consequential outcome, and the record is careful that it is not a "ban": no formal blacklist of named individuals was established, every widely circulated version traces to lifestyle aggregators repeating each other, and Ferrari has stated on the record that it does not blacklist individuals from buying standard production cars. Ferrari retains no documented technical, software or service authority over a car it has delivered. An owner who ceases to matter to Ferrari still owns a fully functional Ferrari.

Rolls-Royce retains drafted authority it has not used. The Whispers terms reserve fees, discretionary refusal, termination on sale and a no-press clause reaching "negative statements" about other members, partners and events. The connected-services terms permit termination on technology change with no retrofit duty, impose arbitration with a class waiver, cap aggregate liability at the greater of twelve months' fees or $100, and release claims for personal injury from failure of services that include emergency call — though that second set is BMW Group's standard US telematics template, and on adversarial review it fell as a Rolls-Royce-specific signal, surviving only as low-weight industry context. None of it has been exercised against any documented customer. No fee charged, no membership refused, no member sanctioned, no service terminated. Meanwhile the exercised record runs the other way: connectivity included for four years at the generous end of industry practice; no feature paywall or subscription on any car anywhere in the fetched contract stack; over-the-air updates owner-launched rather than forced; the warranty expressly tolerating non-genuine parts; club eligibility only ever widening. The car is fully drivable without the app.

Tesla retains everything, and the customer retains nothing inside it. The mechanism is not disputed and is stated in Tesla's own documents: reverting to a previous software version is not possible; warranty coverage is conditioned on accepting updates; harm from failing to install an update is not covered. The customer cannot decline the channel, cannot audit what an update changed, and cannot return to the prior state.

The exercised record is where this stops being architecture and becomes conduct. Voltage caps applied over the air to delivered Model S packs — undisclosed as a loss of range and charging speed — produced a $1.5 million US class settlement at $625 per owner and, in Norway, a Supreme Court judgment in April 2026 awarding NOK 50,000 to each of 115 owners after a five-year, four-instance fight over claims worth roughly $4,600 each. Model 3 Standard Range owners were emailed in June 2019 that an update would remove range, acceleration and Autopilot delivered with their cars unless they paid $4,500. Roughly $8,000 of purchased Autopilot was removed from a used Model S after Tesla's own auction, restored only under press attention, with no consolidated policy on what survives resale ever published. A used Model S was re-locked by roughly 80 miles at a modem retrofit and $4,500 demanded to restore capacity the owner had bought — reversed after the case went viral. Salvage vehicles lose Supercharging and third-party fast charging under a policy the internal memo recorded as unchangeable once applied, whether or not the vehicle passes inspection.

Why the same channel produces the opposite result at the other end. The over-the-air channel is genuinely dual-use, and the Tesla page's own framing is the accurate one: the channel that shipped Sentry Mode free to a million already-sold cars is the channel that capped battery voltage on aging packs. The distinguishing pattern is not the technology. It is where the withdrawals cluster — precisely at the points where Tesla's financial interest conflicts with the installed base: warranty-cost exposure, paywalling, and resale control. And it is how the corrections arrive: the remedy came because the case became public, not because a process existed. A company that improves your product without asking is exercising the same right as a company that degrades it without asking. What separates them is not the act but whether you had any standing to object to either.

Finding. Ferrari retains influence over your next purchase. Rolls-Royce retains rights it has not used over your membership. Tesla retains authority over the object itself, exercisable at any time, with no procedural right of any kind available to the owner. These are three different things and CHI should never score them on one axis.

5.4 Promise and delivery

What is actually being purchased is different in each case, and the correspondence between representation and delivery tracks that difference.

Rolls-Royce sells a bespoke process and delivers it. Range and noise claims were met or beaten under instrumentation. Spectre's promised Q4 2023 first deliveries were met. Uniqueness undertakings were kept. Marketing copy is, by segment standards, unusually careful: superlatives are framed as reputation rather than measurement, range claims are asterisked, and the complaints target — "We aim to close 50% of our customer complaints within 7 business days" — is honest to the point of weakness, since it commits to nothing for the other half and names no ADR route.

Two accuracy debts stand against that. "Hand-built at Goodwood" elides a body-in-white built at BMW Dingolfing, a roof pressed in a BMW plant and a V12 built in Germany — an elision, not a concealment, since Rolls-Royce invited Autocar to Dingolfing and its Phantom chief engineer explained the arrangement on the record. And the 2030 pledge is an unretracted categorical commitment, reversed: twice stated at primary-source level, abandoned within five years. It is not a Promise Reversal: on re-adjudication against the index's own definition the threshold and all three recognition elements fail, and the classification is excluded with no substitute pattern asserted. Three qualifications are binding and travel with it everywhere: the reversal was announced by a named chief executive to two national newspapers and was therefore not silent; the maximum defensible criticism is that no corporate press release was located and the press archive still carries the superseded pledge, uncorrected; and the strongest formulation — "never again produce a new model with an internal combustion engine" — has not been falsified.

Tesla sells a future capability and has not delivered it. This is not the same category of failure and should not be graded on the same scale. Rolls-Royce reversed a strategic pledge about its own product plan, on which no buyer reliance is documented, which withdrew nothing from any delivered car, and which harmed V12-preferring customers not at all — continued combustion production suits roughly 82% of 2025 buyers. Tesla sold a priced product feature, from October 2016, at $3,000 to $15,000, under messaging that the price would rise as delivery approached, with a launch demonstration established as staged by the sworn testimony of its own Autopilot software director, and with hardware sufficiency represented across three generations. In January 2025 the chief executive conceded that HW3 — roughly four million cars — "does not have the capability," promised free computer upgrades, and added that he was "kind of glad that not that many people bought the FSD package." At the July 2026 earnings call there was still no programme, timeline or cost; owners were told to be patient; a stripped "v14 Lite" build was shipped as a software consolation; more than 3,000 HW3 owners across 29 countries have organised collective claims.

The single most important difference between the two reversals is the existence of a remedy path. Rolls-Royce paired its retreat with a retroactive 15-year battery warranty for the most exposed cohort — pressure-timed, credited in full with the pressure documented. Tesla has never had a refund policy for Full Self-Driving. The refunds that exist were won individually: a Washington small-claims order in December 2022, an arbitration reimbursement in July 2025, a $10,000 court award in May 2026 that Tesla continued to contest. A remedy a single customer can win is not a remedy the company offers. Each win shows the claim had merit and shows that no general remedy was available.

Ferrari makes few checkable promises of this kind, which is itself an observation rather than a clearance. Its allocation representations are not made in a form that could be tested. The one tested representation is favourable — the June 2026 rejection of Luce-conditioned allocation — and CHI records it as a company position, never as proof of the underlying fact, the same treatment applied to company denials elsewhere in the index.

The legal discipline required, stated once. Tesla's representation record is contested rather than adjudicated fraud. One state administrative forum found the marketing false and Tesla is litigating to reverse it; a German appellate court ruled substantially for Tesla in 2022, reversing an earlier 2020 ruling against it; two US juries returned defence verdicts in Hsu and Molander; the Benavides plaintiff verdict is one verdict on appeal, not a general finding on Autopilot; LoSavio is certified with the merits untried; the NHTSA engineering analysis covering 3.2 million vehicles is an open allegation, not a finding of defect. What is not in dispute is what was promised on dated pages, what was paid, what was delivered and when. This piece asserts nothing beyond that.

Finding. All three companies made a promise they did not keep in the form stated. The difference is not honesty; it is whether the promise was priced, and whether a remedy exists that the customer does not have to litigate for.

5.5 Price and value stability

Rolls-Royce and Tesla are close to opposites on this dimension, and the reason is instructive: it is about who bears timing risk.

Rolls-Royce publishes no price for anything, which sounds like the worse position and is not. Base prices trailed or tracked US inflation across the decade — the Ghost down 7% to 13% in real terms depending on base year, the Phantom down 4% to 9%, the Cullinan roughly at CPI with its only real-terms rise coinciding with a facelift that added content. Destination and gas-guzzler charges were unchanged from 2019 to 2026. Revenue growth came from itemised, voluntary Bespoke content, disclosed line by line on a statutory window sticker the company completes in full, down to $1,900 lambswool footmats. When demand collapsed, the company cut the Spectre Series II price by 5.3% — a rare segment price reduction — and funded a $5,000 lease credit. (A separate $15,000 of dealer cash on aging 2023 Ghost and Cullinan stock, circulated in Q1 2024, is not counted here: it was undisclosed to consumers with discretionary pass-through, and the record expressly declines to credit it as generosity.) Opacity that is not masking gouging is a disclosure problem, not an extraction problem, and the record establishes which it is.

Tesla publishes one national price for everyone, which sounds like the better position and produces the more damaging finding. The direct-sales architecture genuinely removed the most extractive institution in car retail: no dealer markup, no finance-and-insurance office, no negotiation asymmetry, and buyers never paid over list even in the 2021–22 shortage. The same architecture concentrated all repricing power in one actor, and the one instrument that would share the resulting timing risk — a price-protection policy — has never existed in any market.

The consequences are documented in owner equity. Global cuts of up to about 20% on 13 January 2023 pushed used Model 3 values down 21.5%, about $11,300, between September 2022 and February 2023 — the worst of any vehicle in a 1.4-million-listing study; Hertz booked roughly $245 million of incremental depreciation and exited its electric fleet; Tesla's China spokesperson said there were "no plans to compensate buyers who took delivery before the cut," and there was silence in the US. FSD purchasers at $15,000 saw the price fall to $12,000 and then $8,000 with nothing returned. Five-year depreciation ran −55.9% on the Model 3, −60.4% on the Model Y and −65.2% on the Model S against a −45.6% industry average. CHI does not score the fact that prices moved. It scores that customers carried all of the timing risk while Tesla retained all of the repricing freedom.

The Spectre counter-case, carried honestly. Rolls-Royce's Spectre owners took a heavy residual loss too — used cars listed $136,000 and $158,000 below original stickers by December 2025. The difference is causation and response. Rolls-Royce did not cut list prices under owners; the correction was a market outcome in a collapsing luxury-EV segment, and the company's responses — the price cut on a new model variant, a lease credit, and a retroactive battery warranty extended to existing cars — moved value toward existing owners rather than away from them. Depreciation is not conduct. What is conduct is what a company does to owners when value falls, and the two records differ.

Ferrari is unassessed on this dimension — the page examines no price series and expressly declines to treat secondary-market premiums as evidence, on the ground that a resale premium is a consequence of excess demand and says nothing about whether the shortage behind it was manufactured. That discipline is correct for the Ferrari page's purpose and leaves this comparison with a hole.

Finding. Price transparency and price stability are different goods, and a company can lead on one while producing the comparison's worst finding on the other.

5.6 Service and remedy

The mechanism that matters here is not service quality. It is whether a customer who is wronged has somewhere to go.

Rolls-Royce. The US warranty booklet contains no binding arbitration, no class-action waiver and no jury waiver. It contains a Magnuson-Moss informal dispute mechanism that is free, capped at roughly 40 days, and whose decision "is binding on ROLLS-ROYCE MOTOR CARS NA if you decide to accept it" — binding on the company only, never on the consumer. Any statement that Rolls-Royce forces warranty customers into binding arbitration is false. Warranty service is portable to any authorised dealer, and the manufacturer, not the dealer, is the warrantor. The Provenance promise was proved enforceable in court in Wynn Holdings, where warranty, deceptive-practice and Magnuson-Moss claims survived summary judgment before the case settled without a liability finding.

The friction that exists is real and small: routine removal of California lemon-law suits to federal court, which is lawful and industry-standard; and one 2021 attempt to compel a lemon-law claimant into a selling dealer's arbitration clause as a non-signatory, which the court denied and which has not been repeated. That case belongs on the friction side of the ledger, and the dossier's own adversarial pass corrects an earlier draft that had inverted it into a fairness exhibit.

Tesla.* The warranty terms themselves benchmark at or above par — four years basic, eight-year battery cover with an objective 70% capacity floor that has no equivalent in combustion vehicles, twelve-year corrosion cover. The architecture around them is the finding. AAA arbitration with a class waiver, a real 30-day opt-out and a small-claims carve-out looks milder than the category's worst on its face; its documented effect is class extinguishment — the range class compelled in March 2024, the camera-privacy class in October 2023 in an order that bound a non-signatory minor through equitable estoppel, most FSD purchasers in 2023. What happens after compulsion is confidential by design, so no win rate can be inferred, and none is. **The opacity is itself the finding.

Around it sits a documented pattern of burden transfer. Touchscreen failures on a component with a finite write life were charged to owners at $2,500–$4,000, with Tesla's VP of Legal writing to the regulator that components "are not deemed defective if they fail due to age and wear," until a formal recall demand produced a 135,000-vehicle recall with reimbursement in February 2021. A document-based investigation described a 2019 internal memo instructing staff to attribute suspension failures to "vehicle misuse" against internal data showing known defects and four part redesigns, with roughly 31,000 of some 120,000 control-arm replacements customer-paid — journalism, not an adjudication, and presented as such, with the countervailing fact that roughly three-quarters of those repairs were warranty-paid. Goodwill repairs were once conditioned on silence until the federal regulator publicly called the practice troublesome, and Tesla revised the language within days. There is no telephone path into service by design, so there is no human escalation route. And the two 2020 remedy contractions — the seven-day return policy deleted by removing its web page and the used-vehicle warranty cut from up to four years to one, in the same month, neither announced — have never been reversed.

Against all of that, the service cost finding runs the other way and belongs in the judgment at full weight: the lowest ten-year maintenance and repair cost of any brand in Consumer Reports' member data, no scheduled maintenance regime at all, no dealer service-upsell channel, 30% to 50% of service visits handled by mobile technicians, free service manuals from 2022 ahead of several legacy manufacturers, and over-the-air recall remedies that complete at rates the rest of the industry cannot approach.

Ferrari is unassessed on service and remedy entirely — the page's scope is "not vehicle quality." What the Ferrari record does contain is the harshest single documented remedy in this entire three-company comparison — and it was not imposed by Ferrari N.V. Ferrari of Houston sued a customer over the sale of a $429,000 Purosangue under an "Opportunity Agreement" giving the dealer a right of first refusal if the car was sold within 18 months and making a third-party sale inside that window trigger liability for the profit made plus the dealer's legal fees. CHI's test is proportionality, not existence: a time-limited right of first refusal is a proportionate anti-flipping measure; disgorgement of profit plus the dealer's legal costs is materially heavier than the objective requires. The agreement is a dealer instrument, Ferrari dealers are independent businesses, and it is not attributed to Ferrari N.V. The litigation's outcome was not established.

Finding, and it is one of the piece's more counter-intuitive results. The most concentrated control architecture (Tesla) and the most diffuse (Ferrari's independent dealer network) both produced remedy problems, by opposite routes. Tesla's concentration removed the customer's forum. In the Ferrari market, diffusion meant the harshest instrument in the record was written by an independent dealer rather than by the manufacturer — and it is not attributed to Ferrari N.V. Whether Ferrari publishes dealer conduct standards governing such terms is a question the record neither answers nor reports as searched. Control that is neither concentrated nor supervised is a third failure mode, and the framework in Section 6 has to accommodate it. Rolls-Royce, which sells through independent dealers too, is the counter-example: it warrants centrally, publishes the ADR route, and its one documented certified-pre-owned failure was a dealer's non-disclosure against a manufacturer promise the customer was able to litigate — claims surviving summary judgment before a settlement that carried no liability finding.

5.7 Exclusivity

Ferrari and Rolls-Royce both sell exclusivity as part of the product. The question CHI asks is whether it is customer-beneficial, mutually accepted, performative or extractive.

Rolls-Royce: customer-beneficial and mutually accepted. Volume is capacity-bound at 25 to 26 cars a day and the £300 million Goodwood extension is expressly "not… more capacity for volume." Volume rose roughly 60% between 2018 and 2022 while the rarity language continued, and fell to 5,664 in 2025 — the scarcity is curated and commercially calibrated, and it is openly disclosed as such, which is the Hermès verdict, not the Artificial Scarcity finding. Crucially, exclusivity has been honoured rather than merely asserted, and the honouring is externally checkable. And it has been widened rather than tiered downward: club eligibility went from Goodwood-era owners and their partners in 2020 to certified pre-owned buyers later. The one documented withdrawal of access runs the other way and is small: public factory tours ended, which the record calls a real but low-harm enclosure. Nothing was taken from owners to make ownership feel more exclusive.

Ferrari: customer-beneficial and extractive simultaneously. The physical scarcity is genuine and CHI rejects Artificial Scarcity outright rather than merely failing to prove it — 799 and 599-unit runs are published numbers, and hand-built low-volume manufacturing genuinely cannot be scaled to meet Icona demand without destroying the thing being sold. Exclusivity is part of what a Ferrari customer buys. The extraction is not in the scarcity; it is in the ranking that governs access to it. A customer can read Ferrari's scarcity strategy in the company's filings before spending a euro, and cannot determine their own position, what would improve it, or whether any given purchase counted.

Tesla: no exclusivity finding at all — but the strongest dependency finding.* The Tesla page records Artificial Scarcity as not established: no commercially overrideable scarcity exists, and scarcity effects are limited to delivery-window pressure, scored as manufactured urgency rather than scarcity. What Tesla has instead is dependency that increases after purchase and reaches the exit. Free unlimited Supercharging, sold as a benefit of the car, was made non-transferable and stripped from Tesla's own used inventory. FSD is non-transferable by default, with transfer permitted only through recurring time-limited windows conditioned on buying another Tesla — and those "one-time" windows were repeated at least five times with deadline messaging. Salvage vehicles are permanently locked out. **What survives a change of owner is decided by Tesla rather than by the sale.

Why exclusivity and dependency must not be conflated. They look similar — both restrict what the customer can freely do — and they run in opposite directions in time. Exclusivity restricts entry and is priced into the purchase; the customer knows what they bought. Dependency restricts exit and accrues afterwards; the customer discovers it when they try to leave. A customer who accepts exclusivity has made a bargain. A customer who encounters dependency has had one made for them. That distinction, more than any other in this piece, is why a hostility index cannot treat "restriction" as a single category.

5.8 Reciprocity

All three companies return substantial value for the autonomy their customers surrender. The finding is that this does not settle anything.

Ferrari's return is access, membership, rarity as a purchased attribute, disclosure of strategy in advance, and the protection anti-flipping gives to buyers who intend to keep the car. Nine separate counterevidence findings support Ferrari against one central concern, and the assessment says plainly that this ratio is the honest description of the case. The reciprocity is real. It does not reach the concern, because the concern is not about what Ferrari gives; it is about what the customer is quietly paying in a second currency.

Rolls-Royce's return is the densest documentary value package in the comparison and it is genuinely delivered: the leading warranty-plus-wear-items package among checked peers — with Ferrari's seven-year scheduled-maintenance programme running longer than Rolls-Royce's four — a retroactive 15-year battery term, a certified pre-owned programme whose substance beats its segment even where its disclosure trails, authorship, kept promises, and the complete absence of a digital extraction surface. The reciprocity here is balanced. Its weakness is evidentiary, not analytical — it is documented more thoroughly than it is observed.

Tesla's return is the most extensively independently measured value in the comparison, and the most important finding in this section. CVI 81, "Strong," built on things that are independently verified rather than claimed. And it moved the hostility score by nothing. CHI 68 and CVI 81 are scored on separate axes because they are separately true, and the page does not average them. A customer who received a world-leading charging network, the lowest running costs of any brand and a decade of free features still had no procedural right inside the update channel, still had no refund path for a $15,000 software purchase, still had a class waiver between them and public adjudication, and still could not audit the telemetry that decided their dispute.

The methodological consequence, stated as plainly as it can be:

Reciprocity is necessary for a control architecture to be legitimate and insufficient to make it legitimate. Value delivered does not purchase the right to retain control the customer cannot contest.

That is the single most transferable result in this document, because it is the argument every high-control company makes — look what we give you — and this comparison shows exactly where it stops working. It stops working at the point where the customer has no standing to object to an exercise of the control, however generous the average exercise has been.

↩ Back to the brief

Section 6

THE CONTROL-LEGITIMACY FRAMEWORK

Derived from the three-company evidence, not imposed on it. No numeric formula is proposed, because neither the evidence nor the existing CHI methodology supports one — Ferrari is deliberately unscored, and the two frozen score sets that do exist were produced by separate assessments against different evidence environments rather than by a common instrument applied to all three. A weighted formula built on that would look precise and mean nothing.

6.1 The question the framework answers

When does exceptional corporate control protect customer value, and when does it become customer hostility?

The eleven candidate legitimacy factors were tested against all three records. Six discriminate. Two are necessary but weak on their own. One is untestable on any of the three records. Two behave in ways the original thesis did not anticipate. Sorting them is the framework's actual output. Two structural amendments follow from the sorting, and only two.

One housekeeping point governs the whole framework: latent power — a capability drafted or technically available and never used against a customer — is carried inside Factor 1 as its middle state, not as a separate amendment. It is a watchlist condition: neither a hostility finding nor a clearance.

6.2 The factors that discriminate

1. Unilateral mutability — can the company change what the customer already has?

The strongest single discriminator in the record. It resolves into three states, and keeping them apart is what lets the factor carry latent power without either scoring it as hostility or writing it off:

• Tesla: adversely exercised, over a channel that cannot be declined, audited or reversed, with multiple documented instances.
• Rolls-Royce: present but unused. The technical capability sits one level up at BMW Group and the contractual capability is drafted into live terms; no adverse exercise is documented anywhere — updates are owner-launched, no recall has been remedied over the air, and no over-the-air-removed feature exists in the record. This is the watchlist state, and it is where Rolls-Royce's reserved architecture belongs. It is not scored as hostility, and it is not treated as a clearance: CHI's own Tesla record shows how quickly a reserved capability becomes exercised conduct, which is exactly why the state is named rather than ignored.
• Ferrari: unassessed. No such capability appears anywhere in the record, and the record does not examine the question — the governing artifact's scope is allocation, not ownership.

Test: does the company hold the technical or contractual power to alter a delivered product, and has it used it against an owner's interest? A yes to the first alone places the company on the watchlist. A yes to both is the finding.

2. Post-purchase dependency — does the customer's exposure grow after payment?

• Tesla: grows. Seven customer systems; an account layer whose exit is "none"; non-transferability rules that reach the resale itself; an irrevocable salvage lockout.
• Ferrari: plateaus, on the assessed record. Standing matters for the next purchase, not for using the present car. Technical and service dependency are unassessed.
• Rolls-Royce: shrinks. Four years of connectivity included, then optional; no paywall; independents supplied with parts and dealer-grade technical information; club eligibility widening from owners and partners to certified pre-owned buyers.

Test:* between delivery and disposal, does the customer need the company more, the same, or less? **Increasing dependency is the reliable early signal of a control architecture turning against the customer, and it is observable before any adverse exercise occurs.

3. Remedy accessibility — does a remedy exist that the customer does not have to fight for individually?

• Rolls-Royce: yes. No binding arbitration in the warranty; a free pre-suit process binding on the company only; a manufacturer promise proved enforceable in court.
• Tesla: structurally, no. A class waiver with documented class-extinguishing effect; confidential outcomes; remedies won one at a time in small claims and arbitration; the recurring pattern that remedies arrive with publicity rather than through process.
• Ferrari: unassessed.

Test:* if the company is wrong, what does the customer do, and how many of them can do it at once? **A remedy a single determined customer can win is evidence the claim had merit and evidence that no general remedy exists.

4. Predictability — is the bargain stable after commitment?

• Rolls-Royce: high — a decade of base prices trailing inflation, fixed pass-through charges, no functional mutability.
• Tesla: low — no price-protection policy in any market, a feature price ladder that halved twice with nothing returned, and remedies contracted without announcement.
• Ferrari: unassessed on price, which the governing artifact does not examine; predictable on product, since nothing in the record permits post-delivery alteration; and unpredictable on access by design — a different and more defensible kind of unpredictability, since it is disclosed as the operating model.

Test:* what can change after the customer commits, and who bears the risk of the change? **Volatility is market conduct. Asymmetric allocation of timing risk is scored conduct.

5. Proportionality of enforcement — does the sanction fit the objective?

• The two documented failures here are the Ferrari dealer's profit-disgorgement-plus-legal-fees remedy and Tesla's irrevocable salvage lockout applying regardless of inspection outcome. Both pursue legitimate objectives — protecting allocation integrity, protecting network safety — with instruments materially heavier than the objective requires.
• Rolls-Royce has no documented enforcement instrument of any kind, proportionate or otherwise.

Test:* would a lighter instrument achieve the same legitimate end? **Anti-flipping controls, resale restrictions and network eligibility rules are not hostile in themselves; the remedy attached to them is where the finding lives.

6. Direction of benefit — what is the control pointed at?

The factor that most cleanly separates Rolls-Royce from Ferrari, since both are ultra-luxury access architectures.

• Rolls-Royce's control is pointed at the specification — at producing the object the customer described, uniquely, and at supporting it afterwards.
• Ferrari's control is pointed at commercial behaviour — at ranking customers by spending, retention and engagement that are separable from the scarce car.
• Tesla's is pointed both ways through one channel, which is why the company scores at the top of the corpus on both indices at once.

Test: if you removed the control, who would be worse off? Where the answer is the customer, the control is aligned. Where the answer is only the company, the label is asymmetric at best.

6.3 The factors that are necessary but weak alone

7. Advance disclosure. Necessary — nothing undisclosed can be legitimate — and it does much less work than the thesis assumed. Tesla discloses that reverting to a previous software version is not possible. Full disclosure of an unrefusable term does not make the term consensual; it makes it clear. Ferrari discloses its scarcity strategy fully and its ranking not at all, and it is the ranking that carries the finding. Rolls-Royce discloses every term it has published — warranty, dispute route, repair rights, club and telematics terms — while publishing no price at all and no Bespoke commercial terms, no certified-pre-owned inspection standard and no non-US warranty booklet. The price opacity produces the weakest finding against it; the unpublished term-sets are the residue. Disclosure earns credit — CHI treats it as a mitigating factor and Ferrari's transparency is the strongest single item in its favour — but disclosure of a control is not a justification of it.

8. Meaningful consent. Necessary, and structurally impossible in one of the three cases. The Tesla customer cannot decline the update channel; the choice is between accepting alterations they cannot review and losing features and warranty coverage. The Ferrari customer consents to entering a hierarchy and cannot consent to their position within it, because they cannot see it. The Rolls-Royce customer accepts terms that are published, complete and — so far — unused. Consent is discriminating only where it is possible; where the architecture removes the option to decline, consent stops being a factor and mutability takes over the analysis.

6.4 The factor that is untestable on this record

9. Customer comprehension. No direct evidence exists for any of the three companies. No survey, no comprehension study, no owner-population data. The nearest available proxy is inferential and belongs to Ferrari: its own top clients found "buy this to protect your standing" a plausible description of their relationship, which suggests they comprehend the hierarchy's incentive perfectly well without disclosure — and that comprehension did not protect them from it. That is a useful hint and not a finding. CHI should either build an instrument for this factor or stop listing it, because at present it cannot be scored and its presence in a legitimacy list implies an evidence base that does not exist.

6.5 The two factors that behaved unexpectedly

10. Reciprocity — does not discriminate. Established in Section 5.8 and restated here because it is the framework's most important negative result. Tesla's delivered value is the most extensively verified of the three and it holds the deepest hostility finding. Reciprocity is a precondition, not a defence. In the framework it should be recorded as a threshold test — is anything being returned at all? — rather than as a weighted factor, because above that threshold it stops separating cases.

11. Ability to exit — discriminates, but only in combination. Exit costs are real in all three cases and mean different things. Leaving Ferrari costs standing, which is a cost in a currency the customer chose to hold. Leaving Rolls-Royce costs nothing documented — no written resale restriction was located, the one spoken threat was never enforced, and depreciation is market physics. Leaving Tesla costs transferable value the customer paid for: free unlimited Supercharging made non-transferable, FSD non-transferable by default, a salvage lockout, a Cybertruck resale clause. Exit friction is only a hostility signal when the thing lost at exit was sold to the customer as theirs. That refinement matters, because otherwise every loyalty programme and every depreciating asset registers as exit resistance.

6.6 The two structural amendments this comparison forces

Amendment A: concentration is itself a factor.

The original thesis holds that the amount of control is irrelevant. The Tesla record does not permit that in its strong form. Tesla holds the product, the price, the evidence, the forum, the parts, the repair, the insurance, the account and the exit. No single one of those has to be exercised hostilely for the customer to have nowhere to stand. The Tesla page's own formulation is exact: the single accountable party and the single point of unilateral power are the same fact seen from opposite sides — and the customer argues against a database they cannot see, in a forum whose outcomes are confidential, about a product the other party can alter.

The framework therefore needs a concentration test, distinct from any individual exercise:

Does any independent party hold the evidence, the forum, or the remedy? If the answer is no on all three, the architecture has removed the customer's recourse, and that is a harm no individual act produced.

Amendment B: unsupervised diffusion is a third failure mode.

The mirror image, and the reason Amendment A is not simply "concentration is bad." The harshest documented individual remedy in this three-company record — profit disgorgement plus the dealer's legal fees on an 18-month resale window — came from an independent dealer that Ferrari does not control, in a market where published, audited dealer conduct standards on allocation appear on the Ferrari record only as evidence that would reduce concern if it existed — the assessment asserts neither that such standards exist nor that they do not.

Rolls-Royce sells through independent dealers too and produces the opposite result on the same structure, because it warrants centrally, publishes the dispute route, makes servicing portable across the network, and — in the one documented certified-pre-owned failure — carried a promise the customer could enforce against the manufacturer even though the non-disclosure was the dealer's.

Diffused control without published standards is not the safe alternative to concentrated control. It relocates the power to a party with fewer reputational constraints and no obligation to publish anything.

This amendment also disposes of a lazy comparative move that this piece explicitly rejects: treating a manufacturer as innocent simply because the harsh instrument was a dealer's. Attribution discipline requires that the instrument not be assigned to Ferrari N.V.; analytical honesty requires noting that the record leaves the manufacturer's supervisory role in that layer entirely open, in either direction.

6.7 The framework, assembled

A control architecture is legitimate to the extent that:

The six that discriminate

1. It cannot alter what the customer already holds — or, where it can, the customer holds a procedural right inside that mechanism. (unilateral mutability; read in three states — capability not present / present but unused / adversely exercised. The middle state is a watchlist condition, not a finding either way.)
2. The customer's dependency does not grow after payment. (post-purchase dependency)
3. A remedy exists that the customer does not have to win individually. (remedy accessibility)
4. The commercial bargain is stable after commitment, and timing risk is shared rather than transferred. (predictability)
5. Enforcement instruments are no heavier than their legitimate objective requires. (proportionality)
6. The control is pointed at an outcome the customer wanted. (direction of benefit)

The two that are necessary and insufficient

1. Everything material is disclosed before commitment. (advance disclosure)
2. Declining is possible, where the architecture permits it at all. (meaningful consent)

The one that is currently untestable

1. The customer understands what they are agreeing to. (comprehension — no instrument exists for this on any of the three records)

The two that behaved unexpectedly

1. Something substantial is returned. (reciprocity — a threshold test, not a weighted factor)
2. Exit does not cost the customer something that was sold to them as theirs. (ability to exit — discriminates only in that specific form)

The two structural amendments this comparison forces

• Amendment A. Some independent party holds the evidence, the forum, or the remedy.
• Amendment B. Where control is delegated, standards are published and the manufacturer remains answerable.

Applied to the three:

Rolls-Royce satisfies factors 2, 3, 4, 5, 6, 8, 10 and 11 on the documented record, and satisfies Amendment B. It sits in factor 1's middle state — capability present but unused: no adverse alteration of a delivered car is documented anywhere, while the technical capability sits one level up at BMW Group and the contractual capability is drafted into live terms. That is a watchlist condition, and it is the only place in this framework where latent power is carried at all — carried, not scored. It is split on factor 7: warranty, dispute route, repair rights, club and telematics terms are all published, while no price, no Bespoke commercial term, no certified-pre-owned inspection standard and no non-US warranty booklet is. It satisfies Amendment A only partially: its warranty dispute route is real and non-binding on the consumer, but the telematics layer imposes arbitration with a class waiver and a $100 liability cap — a BMW-template stack that the record's adversarial pass found fell as a Rolls-Royce-specific signal and survives only as low-weight industry context. Factor 9 is unassessed, as it is for all three.

Ferrari satisfies factors 10 and 11, and — on the assessed record — factor 2. It is split on factor 7, and the split is the finding: the scarcity strategy is disclosed exceptionally well in the company's own investor communications, and the ranking that governs access to it is disclosed not at all. It fails factor 8, because a customer can consent to the existence of a hierarchy and cannot consent to a position within it they cannot see. It fails factor 6 at the ranking layer, which is the central Ferrari finding. Factors 1, 3, 4 and Amendment A are unassessed, because the governing artifact examines allocation rather than ownership. Factor 5 cannot be scored against Ferrari N.V. at all: the only enforcement instrument in the record is a dealer's, expressly not attributed to the manufacturer, and no Ferrari corporate enforcement instrument is documented. Amendment B is where the Ferrari market's exposure sits, and the manufacturer's role in it is undetermined.

Tesla satisfies factor 10 emphatically and factor 6 partially, is mixed on factor 7 — terms published, material changes made silently — and fails factors 1, 2, 3, 4, 5, 8, 11 and Amendment A outright. Amendment B barely arises in the vehicle business, which delegates almost nothing; it is live in Tesla Energy, where installation was handed to third parties after 2023.

None of the three fails everything, and none passes everything. A framework that produced either result on this evidence would be miscalibrated.

↩ Back to the brief

Section 7

FINDINGS AND METHODOLOGICAL SIGNIFICANCE

What the comparison demonstrates about the index's ability to make five distinctions that critics reasonably doubt it can make.

7.1 Premium constraint versus exploitation

The test. Can CHI tell the difference between a company that constrains supply because it genuinely cannot make more, and a company that uses constraint as leverage?

The demonstration. Both answers came from the same company, which is the strongest possible form of the test. Ferrari builds 799 F80s; that constraint is physical, published and admitted, and Artificial Scarcity is rejected on the Ferrari page, not merely left unproven — a stronger clearance than the index gave Rolex on the equivalent question. And the same page finds Scarcity Leverage supported, because access to the constrained product structurally rewards commercially valuable behaviour.

Three concepts that popular argument collapses into one accusation are held apart: Scarcity Allocation (a genuinely insufficient product exists and someone must be chosen — not hostile, deliberately unscored), Opaque Scarcity (the customer cannot understand availability, position or criteria — partially applicable, at the ranking layer only), and Scarcity Leverage (access induces additional spending, retention or engagement — the finding). Separating them is what allows two of the three to be cleared and the third to stand.

Why it matters. An index that could not separate them would have to score Rolls-Royce and Ferrari identically, since both build few cars on purpose and both say so. It scores them differently, and the reason is legible: Rolls-Royce's constraint binds at 25–26 cars a day and its ladder ranks on affinity; Ferrari's constraint is real too, and its ladder ranks on spend.

7.2 Exclusivity versus extraction

The test. Can CHI distinguish a company that sells rarity from one that monetises the customer's desire for it?

The demonstration. Rolls-Royce's exclusivity architecture is more elaborate than Ferrari's in some respects — five directly operated Private Offices, a members' club with a Membership Committee, invitation-only Coachbuild, no published price anywhere — and produces a milder finding. The discriminating facts are specific and checkable: membership is free in practice; eligibility has only widened; no fee, refusal or sanction has ever been documented; and the one-off undertakings have been externally observably honoured. Meanwhile the reserved right to charge fees is recorded as a watchlist item precisely because CHI declines to score capability as conduct.

Why it matters. It shows the index scoring the exercise, not the apparatus. If the apparatus were the finding, Rolls-Royce's Whispers terms would score worse than Ferrari's unwritten hierarchy, which would be the wrong answer.

7.3 Personalisation versus monetised complexity

The test. Both look like "more options for the customer." Can CHI tell them apart?

The demonstration. They appear in this comparison as near-mirror images.

Rolls-Royce's Bespoke is roughly 40% of car value, spent voluntarily, itemised line by line on a statutory label, producing objects with client-named colours reserved for their use, under an undertaking of non-replication that has held for five to nine years. The customer ends up with something that did not exist before and that nobody else can have.

Tesla's marginal cases run the other way: a $595 retrofit for a deleted stalk, $700 for a round steering wheel, $2,000 to $2,500 to keep infotainment features working, a $1,000–$1,500 hardware toll to subscribe to FSD on cars sold as hardware-complete. The customer ends up back where they started, having paid.

Why it matters. The test is not the price and not the number of options. It is whether the transaction adds something to the customer's product or restores something taken from it. Stated as a rule: personalisation creates; monetised complexity restores.

7.4 Innovation versus unilateral post-purchase control

The test. The hardest of the five, because the two run through one wire.

The demonstration. The over-the-air channel produced, on the same architecture: Sentry Mode, Dog Mode, dashcam, a blind-spot camera, a 5% power increase and holiday updates still reaching decade-old cars, free, on cars already sold; recall remedies installing fleet-wide within days at completion rates the rest of the industry cannot approach; and software-limited packs unlocked remotely for hurricane evacuations. It also produced voltage caps on delivered packs, $4,500 demanded to keep delivered range, purchased Autopilot removed after Tesla's own auction, capacity re-locked at a modem retrofit, and an irrevocable salvage lockout.

CHI's resolution is to score the two on separate axes — CVI 81 and CHI 68 for the same company in the same year, with the page explicitly declining to average them — and to locate the finding not in the technology but in two observations. First, the withdrawals cluster where the company's financial interest conflicts with the installed base: warranty-cost exposure, paywalling, resale control. Second, the customer holds no procedural right inside the channel in either direction, which is why the hurricane unlock and the voltage cap are the same capability seen from two sides.

Why it matters. This is the distinction most consumer-protection frameworks fail. A framework that scored over-the-air capability negatively would penalise the best recall-completion mechanism ever built. A framework that scored it positively would have nothing to say about battery-gate. Separate axes plus a procedural-rights test resolves it without pretending the channel is two channels.

7.5 Brand power versus customer hostility

The test. Does the index punish companies for being powerful, admired or expensive?

The demonstration. Rolls-Royce Motor Cars commands unusual brand deference and produces a frozen result at the customer-favourable end — CHI 23 / CVI 85 / CFS +62, "Low hostility," with no established CHI pattern. Ferrari, another marque with formidable brand standing, receives a named concern with the score deliberately withheld rather than manufactured, because scoring a control case against subscription-economy companies without cross-sector normalisation would produce a number that looks precise and means nothing. Tesla, whose brand deference is the most contested of the three, holds a frozen CHI of 68 alongside the CVI of 81. Two frozen results at opposite ends of the index, from two companies that both hold exceptional control, is the demonstration — and it is a demonstration about mechanisms, not a placing. The Rolls-Royce figure is what its own assessment publishes; it is not a score this comparison assigned, and it is not weighed against Tesla's as though a common instrument had produced both.

High price is nowhere treated as evidence of hostility. A multi-million-pound coachbuilt commission and a $99 monthly subscription are both assessed on what the company retains, not what it charges. (No coachbuild price is stated anywhere in this piece: every reported figure in the record is unconfirmed and on the do-not-publish register.) Ferrari's product value is expressly protected from contamination: "a finding about how a car is allocated says nothing about how good the car is, and CHI does not let one contaminate the other."

Why it matters, and the honest caveat. The result refutes the "CHI dislikes powerful companies" objection on its own terms. It does not refute a subtler objection — that CHI can only find what is visible, and visibility correlates with scale, litigation and press attention. Section 8.5 takes that seriously rather than dismissing it.

7.6 What the comparison contributes to CHI methodology

Four transferable results, in order of usefulness:

1. Reciprocity is a threshold, not a weight. Value delivered does not offset retained control the customer cannot contest. This is the argument every high-control company makes, and the Tesla record shows exactly where it fails.
2. Concentration and unsupervised diffusion are both failure modes. The framework needs a test for whether any independent party holds the evidence, the forum or the remedy — and a test for whether delegated control comes with published standards.
3. Exit friction only counts when the thing lost at exit was sold to the customer as theirs. Without that refinement, ordinary depreciation and ordinary loyalty programmes register as hostility.
4. Unassessed is not clean, and it must be visible in the output. Ferrari's blank cells on price stability and remedy are the largest hole in this comparison, and a matrix that rendered them as blanks or dashes would read as favourable. They are labelled.

And one methodological duty this piece hands back to the index: customer comprehension is currently unmeasurable across the entire corpus. It should be instrumented or removed from the legitimacy factor list.

↩ Back to the brief

Section 8

COUNTERARGUMENTS AND LIMITATIONS

Each objection is stated at its strongest, answered, and — where it survives the answer — carried as a limitation on the conclusion rather than argued away.

8.1 "The companies serve different markets and different customers"

The objection at full strength. A Model 3 buyer and a Phantom commissioner are not in the same market, do not have the same alternatives, do not have the same bargaining power, and are not protected by the same expectations. Comparing them is category error dressed as analysis.

The answer. The comparison is not of markets, customers or products. It is of architectures, and the architecture question is identical in all three cases: what power does the company hold over the customer, when does it hold it, what is it pointed at, and what standing does the customer retain? That question is as meaningful at the bottom of the price range as at the top, and the answers are directly comparable because the mechanisms are the same kind of thing — terms, channels, criteria, forums, remedies.

The comparison also deliberately excludes everything market-dependent. It does not compare prices, specifications, performance, market share, or which company makes the better car. Where a product fact enters, it does so because it tests a representation (Rolls-Royce's instrumented range) or because it is a component of the customer relationship (Tesla's charging network).

What survives as a limitation. Market position genuinely shapes what a customer can do about an architecture. A Tesla owner facing a class waiver has fewer practical alternatives than a Rolls-Royce client who can call a named Client Relations contact and, if necessary, retain counsel. The architectures are comparable; the customers' capacity to resist them is not, and that asymmetry runs in Tesla customers' disfavour in a way the architecture comparison alone does not capture. This limitation makes the Tesla finding, if anything, understated.

Entity controlRolls-Royce here means Rolls-Royce Motor Cars Limited, the BMW-owned automobile manufacturer — not Rolls-Royce plc / Rolls-Royce Holdings plc, the separate aerospace, defence and power-systems group, which is out of scope and is described nowhere on this page.

8.2 "Ferrari and Rolls-Royce customers voluntarily accept unusual conditions"

The objection at full strength. Nobody needs an F80. A customer who submits to an unpublished ranking system in exchange for the chance to buy a 799-unit hypercar has made a free bargain with full knowledge that a bargain is being made. Treating that as hostility is paternalism, and treating the customer as a victim is condescension.

The answer. The objection is substantially correct and this piece adopts most of it. Ferrari's customers are not treated as irrational for accepting restrictive conditions; the assessment states explicitly that exclusivity is part of the product being sold and that a customer buying a Ferrari is buying, among other things, the fact that not everyone has one. Loyalty preference is expressly not a finding against Ferrari: rewarding a collector who carried depreciation and supported weaker models is ordinary commerce.

The finding survives the objection on a narrower point. Voluntary acceptance is only meaningful where the terms are knowable. A Ferrari customer can read the scarcity strategy in the company's filings and cannot determine their own position, what specifically would improve it, or whether any given purchase counted. What is consented to is the existence of a hierarchy; what cannot be consented to, because it is not visible, is the exchange rate between behaviour and standing. Consent to an unpriced obligation is thinner than consent to a priced one, and that is the whole of the finding.

For Rolls-Royce the objection largely succeeds and the assessment says so: the invitation ladder is disclosed, sector-universal, capacity-grounded patronage, and no refusal, blacklisting or arbitrary exclusion was identified in the available record — which the later adjudication is careful to state as a failure to find rather than as proof that none occurred.

What survives as a limitation. There is no evidence, for any of the three companies, about what customers actually understand. The comprehension gap named in Section 6.4 cuts both ways: it weakens any claim that customers are misled and any claim that they consent knowingly.

8.3 "Tesla's software model generates substantial customer benefit"

The objection at full strength. The over-the-air channel is the most significant consumer-benefit mechanism in modern automotive history. It has added capability to sold cars for a decade, executed recalls at completion rates the industry cannot approach, and unlocked range for people fleeing hurricanes. Penalising it is technophobia, and no other company in this comparison has given its customers anything comparable after the sale.

The answer. Accepted, in full, and it is priced in. CVI 81 — "Strong" — rests on independently verified evidence rather than claims. (Rolls-Royce's CVI is now frozen at 85, "Exceptional," on its own assessment. The two are not ranked against each other here: they were produced by separate assessments against very different evidence environments, and the comparison draws no conclusion from the four-point gap.) The Tesla page's countercase runs to a full section carried at full weight, and the assessment states plainly that a page which only prosecuted would be a worse analytical instrument and a less accurate one.

The finding is not that over-the-air control is hostile. It is narrower and better evidenced: the customer holds no rights within it, and the documented withdrawals cluster precisely where Tesla's financial interest conflicts with the installed base. A version of Tesla that ran the identical channel with a procedural right — advance notice of what an update changes, an audit trail, a rollback option, a published policy on what survives resale — would deliver every one of the benefits above and would score materially lower on CHI. The benefit and the hostility are separable in practice, which is why they are separable in the scoring.

What survives as a limitation. Nothing in this piece establishes that a rights-preserving version of the channel is commercially or technically straightforward, and it may not be. That is a real constraint and it does not change who currently bears the risk.

Entity controlRolls-Royce here means Rolls-Royce Motor Cars Limited, the BMW-owned automobile manufacturer — not Rolls-Royce plc / Rolls-Royce Holdings plc, the separate aerospace, defence and power-systems group, which is out of scope and is described nowhere on this page.

8.4 "Rolls-Royce's attentive service may be inseparable from extreme pricing"

The objection at full strength. A company that charges half a million dollars a car can afford a four-year warranty that includes brake pads. The favourable finding may be measuring price, not virtue — and CHI has committed to not treating high price as evidence of hostility, which cuts the other way too: it must not treat high price as evidence of alignment.

The answer. The objection is serious and is partly correct, so the comparison is drawn against same-price peers, not against the mass market. Against freshly fetched comparators, Bentley's warranty is three years with no currently published maintenance term (a 2020 UK service-plan document bundled a plan into the list price of some models); Mercedes-Maybach's is 48 months capped at 50,000 miles with maintenance excluded; Ferrari's is three years (four in the UK), with a seven-year scheduled-maintenance programme that runs longer than Rolls-Royce's four and excludes wear items. Every one of those companies can also afford it. Rolls-Royce's package leads the checked set on warranty-plus-wear-items; Ferrari leads on maintenance duration; the comparison is close and it is a comparison of equals.

More importantly, the strongest Rolls-Royce findings are not affordability findings at all:

• Non-replication of coachbuilt designs costs nothing to promise and something real to keep, and it has been kept for five to nine years where it is externally checkable.
• The absence of a digital extraction surface — no feature paywall, no subscription on any car, four years of included connectivity — is a choice, not a budget. BMW Group experimented with feature subscriptions on its own brands; those experiments never crossed to Rolls-Royce on any evidence.
• Base prices trailing inflation for a decade is the opposite of what price power would predict.
• A warranty containing no binding arbitration and no class waiver is a legal choice available to any company at any price point.

What survives as a limitation. The retroactive battery warranty and the Series II price cut were demand-collapse responses, and the record says so: they came after a 47% fall in Spectre deliveries, weeks before the electrification reversal and months before a chemistry change with no retrofit path. Under the index's Fixed Under Pressure principle the substance is credited in full and the pressure is documented. Any reading of this piece that implies spontaneous generosity misstates the record.

8.5 "Publicly observable evidence differs significantly across the three companies"

The strongest objection in this section, and the one that most constrains the conclusion.

The objection at full strength. Tesla has millions of customers, an adversarial press corps, class actions, multiple regulators on three continents, leaked internal documents and sworn depositions. Rolls-Royce has roughly 5,664 customers a year who complain to a concierge rather than a federal portal, sealed dealer agreements, confidential settlements, and — in this investigation specifically — blocked access to Reuters, Bloomberg, the New York Times, the Wall Street Journal, Consumer Reports, the Internet Archive, Reddit, Trustpilot, the Better Business Bureau and CourtListener, plus robots-walled owner forums and unqueried non-US regulators. Ferrari was researched as a control case for a different investigation and stopped when that question was answered. A comparison of documented incidents across those three evidence environments measures journalism, not conduct.

The answer, in four parts.

First, the comparison is deliberately built on the least visibility-dependent evidence available.* Terms, contracts, warranty booklets, published criteria, dispute forums, recall filings and regulator records are equally observable regardless of a company's press profile. Rolls-Royce's Whispers clauses, connected-services liability cap and arbitration terms were found because the investigation read the contracts, not because anyone complained about them. Tesla's owner's manual, warranty conditions and order agreement were read the same way. **Both companies' most structurally revealing facts came from documents, not incidents.

Second, incident counts are explicitly demoted. This piece does not conclude that Tesla is worse because more things have happened to Tesla customers. It concludes that Tesla holds mutability, dependency and forum control that the other two do not — architecture findings that would hold if not one incident had ever been reported.

Third, the asymmetry is recorded in the finding itself rather than corrected for silently. Rolls-Royce's verdict is stated throughout as documentary-strong and observationally thin, which is the dossier's own characterisation: the direction of every reachable indicator is favourable, and roughly a fifth of the customer relationship sits behind sources the investigation could not reach.

Fourth, where absence of evidence could be mistaken for evidence of absence, it is labelled. The Rolls-Royce record contains explicit evidence-insufficiency findings — on discrimination, on dealer-floor conduct, on non-US recall records, on dispute volume ("no visible red flag" is supportable; "low litigation volume" is not) — and this piece carries them as insufficiency, not clearance.

What survives as a limitation, and it is the piece's principal one. A favourable finding built primarily on documents is less robust than an adverse finding built on documents and conduct. If the unreachable venues were opened — owner forums, confidential settlement terms, dealer agreements, UK and EU regulators, the Bespoke contract stack — the Rolls-Royce finding could move. The Tesla finding could not move much in the favourable direction, because its adverse evidence is largely primary and its countercase has already been carried at full weight. The two verdicts are therefore not equally falsifiable, and readers should weight them accordingly.

8.6 "Customer expectations are not identical"

The objection at full strength. A Rolls-Royce client expects to be looked after and pays for it. A Tesla buyer expects a technology product and knows technology products change. Judging both against one standard imports a set of expectations that only one group of customers holds.

The answer. The framework in Section 6 is deliberately built from factors that do not depend on expectation. Can the company change what you already hold? Does your dependency grow? Does a remedy exist you needn't win alone? Is the bargain stable? Is enforcement proportionate? What is the control pointed at? None of those asks what the customer hoped for.

Where expectation does enter, it does so as the company's own stated standard rather than an imported one — which is the fairest available test and, notably, the one that produces the adverse Rolls-Royce finding. The 2030 pledge is scored against Rolls-Royce's own words, not against a general standard of corporate constancy. Tesla's representation record is scored against dated pages Tesla published, not against what a reasonable buyer might have assumed.

What survives as a limitation. The mass-market/luxury divide does affect one thing the framework cannot fully neutralise: the severity of a given failure. Losing $4,500 of range matters differently to a $40,000-car owner than a comparable proportion would to a commissioner of a car costing an order of magnitude more. This piece does not convert findings into a severity-weighted score, partly for that reason.

8.7 Limitations not raised as objections, stated for completeness

• One of the three records is deliberately unscored, and the two that are scored are not commensurable. Ferrari's CHI, CVI and CFS are reserved by design. Tesla's are frozen at 24 August 2026; Rolls-Royce's at 29 August 2026. The two frozen sets come from separate assessments against different evidence environments — one adversarial and conduct-rich, one documentary and conduct-thin — and no conclusion in this piece rests on comparing the numbers, which are shown only as calibration context.
• The Ferrari record does not assess price stability, service, failure or remedy at all. This is the largest single gap and it is named in Sections 0.2, 4.3, 5.5 and 5.6.
• A material conflict between the artifacts was resolved rather than averaged. The Rolls-Royce dossier's claim that Ferrari imposes written corporate no-resale terms is not supported by the Ferrari page, which attributes the documented written instrument to a dealer and declines to attribute it to Ferrari N.V. This piece follows the Ferrari page. See Section 0.3.
• Several Rolls-Royce quotations are second-hand pending a primary anchor, including the March 2026 reversal quotes, whose originals in The Times and The Guardian were never fetched. The substance is consistent across at least six reproductions; the wording is not verbatim-safe and is not treated as such here.
• Three live matters could move findings. The Tesla HW3 remedy remains unresolved; the California DMV finding is adopted and under Tesla's court challenge; the Rolls-Royce Spectre battery-backorder pleading is untested with the company's side entirely unreported, and is the primary monitoring item on the now-frozen Rolls-Royce assessment.
• No external research was conducted for this piece. Where the three records are collectively silent, this document says so rather than filling the gap.

↩ Back to the brief

Section 9

FINAL CONCLUSION

Returning to the question

When does exceptional corporate control protect customer value, and when does it become customer hostility?

Control protects customer value when it is pointed at the object the customer asked for, cannot reach the object after it has been delivered, and leaves the customer standing somewhere — a forum, a remedy, an independent holder of the evidence — from which an exercise of the control can be contested.

Control becomes customer hostility when it survives the transaction, grows afterwards, and closes every venue in which it could be challenged. At that point the company does not have to do anything hostile for the customer to have been harmed, because the loss of recourse is the harm.

That is the answer the three records support. It is narrower than "control is bad" and it is more demanding than "disclosure makes control legitimate," and both of those are the answers a less careful comparison would have produced.

The thesis, adjudicated

The working thesis — hostility is determined not by how much control a company possesses but by how that control is exercised, disclosed, reciprocated and remedied — is supported in its main claim and complicated in two respects. The two complications are numbered here as complications, not as amendments, to keep them distinct from the framework's two structural amendments (A: concentration removing independent recourse; B: unsupervised diffusion through delegated parties), which are propositions about architecture rather than about the thesis.

Supported: two companies holding exceptional control over their customer relationships produced opposite findings. And the market architecture with the least documented manufacturer post-purchase control nevertheless exposed the harshest individual remedy in the record — through an independent dealer, not through Ferrari N.V. Where control sits, and whether it is supervised, predict more than how much of it a company holds.

Complication one — reciprocity leaves the list of determinants. It becomes a threshold test. Tesla's delivered value is the most extensively and independently verified in the comparison and it holds the deepest hostility finding, and the value moved the hostility score by nothing.

Complication two — concentration re-enters as a determinant, in one specific form. Where a single party holds the product, the evidence, the forum and the remedy simultaneously, accumulated control converts into the absence of customer recourse. That is a harm no individual exercise produced, and it is what the framework encodes as Amendment A.

Latency needs no further amendment. A power drafted and never used is already inside the thesis: the phrase "how control is exercised" accommodates it once Factor 1 is read in its three states — capability not present, capability present but unused, capability adversely exercised. Rolls-Royce's reserved architecture occupies the middle state. It is not hostility and it is not nothing; it is a watchlist condition, and the reason for naming it is that CHI's own Tesla record shows how quickly a reserved capability becomes exercised conduct.

The verdict

The architecture that most strongly aligns corporate control with customer value is Rolls-Royce Motor Cars' — and the reason is specific rather than general. It is not that Rolls-Royce is admired, expensive or well-mannered. It is that the company has not built most of the mechanisms that would allow control to be used against a delivered car, and has directed the control it does hold at producing the object the customer specified. No feature paywall or subscription exists on any model. Connectivity is included for four years and then optional. Updates are owner-launched. The warranty tolerates non-genuine parts and contains no binding arbitration or class waiver. The pre-suit dispute process binds the company and not the consumer. Base prices trailed inflation for a decade while revenue growth came from voluntary, itemised content. Uniqueness undertakings have been kept where anyone outside the company can check. Club eligibility has only ever widened.

The freeze does not do this work, and must not be read as doing it. Rolls-Royce's scores were frozen on 29 August 2026 at CHI 23 / CVI 85 / CFS +62 with no established CHI pattern, and that is the company's own published result rather than a placing this comparison awarded. The verdict below rests on the architecture, exactly as it did before the numbers existed.

Two qualifications are load-bearing and are not footnotes to that verdict.

First, the verdict rests on a thinner observational base than the others, and freezing the scores did not change that. It is documentary-strong and conduct-light, because the venues where adverse conduct would be visible were unreachable. The Rolls-Royce assessment states the point itself: low hostility here means low observed hostility, and a low hostility score built on partial visibility is a different claim from a low hostility score built on complete visibility. If those venues were opened, the finding could move. Its adverse counterpart at Tesla could not move much, because Tesla's adverse evidence is largely primary and its countercase has already been credited in full. These verdicts are not equally falsifiable and should not be read as though they were.

Second, alignment achieved by not having built the mechanism is real and it is contingent. Rolls-Royce's cars increasingly run BMW Group electronics and telematics; the Spectre and Project Nightingale are electric; the capability that makes remote alteration possible exists one level up in the corporate structure; and the company has already drafted — in live terms — the fee, expulsion, speech and service-termination powers it has never used. Nothing suggests it intends to use them. Everything about the Tesla record suggests that this is exactly the moment at which an index should say so and keep watching.

The greatest unresolved asymmetry is Tesla's — and the phrase is chosen precisely, because it is not the same as saying Tesla is the worst company here. Tesla's delivered value is the most extensively and independently verified in this comparison, and its conventional extraction surface is one of the cleanest in the entire index: no advertising in the product, no evidence of data sales, one posted national price, no dealership standing between the customer and the manufacturer. The asymmetry is not about what Tesla takes. It is that the Tesla customer is the only one of the three who holds no procedural rights inside the mechanism that governs what they own, and who cannot decline, audit or reverse a change to a product they hold title to. It is compounded by the fact that the corrections that arrived came overwhelmingly from regulators, courts and publicity rather than from the company deciding first — and by the trajectory: the operational record has genuinely improved while capability that once shipped as standard is now rented, deleted controls are sold back, and the purchase option for the flagship software has been removed.

Ferrari sits between them, and its asymmetry is of a third kind that the other two do not exhibit. It is pre-transactional, relational and voluntarily entered — and it is the least curable by disclosure alone, because publishing full allocation criteria would partly destroy the thing being allocated. Ferrari deserves two specific credits that no other company in this comparison earns. It is more honest about its scarcity strategy than almost any company in the index, stating exclusivity and controlled volume in its own investor communications where others deny the equivalent. And when a report suggested its electric model was functioning as a ticket to future allocations, its most senior commercial executive publicly called a mandatory-purchase policy a "huge mistake" within days — placing Ferrari's own line in almost exactly the place CHI draws it.

What survives against Ferrari is narrow, structural and hard to dismiss: it ranks its customers, the ranking determines access to the things they most want, the ranking is fed by spending and retention, and the criteria are not published in terms a customer could check. That architecture creates the incentive whether or not anyone states a requirement — which is why Ferrari's own clients found the allegation believable. And Ferrari's remedies here are cheap: publishing allocation principles for limited series, stating a prohibition on linking any purchase to standing, or giving waiting customers position information would each cost the company disclosure and not a single additional car.

What this is not

This is not a ranking of three companies by desirability, and it does not tell anyone what to buy. Rolls-Royce's favourable finding is about an architecture, not about whether a Phantom is worth its price. Tesla's adverse finding coexists with the highest delivered value in the comparison and with a genuinely clean record on the extraction behaviours most consumer indices measure. Ferrari's finding is about allocation and says nothing about the cars, which the index expressly refuses to let one contaminate.

Nor is it a ranking that a fourth company could be slotted into. What the comparison produces is a framework, not a league table. It has eleven tested factors, sorted into four groups: six that discriminate, two that are necessary and weak, one that is currently unmeasurable, and two that behave unexpectedly. Above them sit exactly two structural amendments, about concentration and unsupervised diffusion. Latent-but-unused capability is carried as the middle state of Factor 1, and nowhere else. The framework is the transferable output. The three verdicts are what testing it produced.

The formulation, final

Ferrari's control decides whether you are permitted to buy — and quietly prices what that permission costs in behaviour.

Tesla's control decides what the thing you already bought will be allowed to remain.

Rolls-Royce's control is aimed, so far, at building what you asked for — and holds in reserve powers it has drafted and not used.

Three architectures of control, exercised by three companies whose exceptional power over their customers sits in three different places in the relationship, producing three materially different customer positions.

The total quantity of power was never something this comparison could hold constant, and it was never what decided the outcome. Where the power sits, where it points, how long it lasts, and whether the customer can stand anywhere and object — those were.

↩ Back to the brief

Section 10

EVIDENCE AND CLAIM-CONTROL APPENDIX

10.1 Artifact codes

10.2 Evidence classes used in this appendix

FACT — established by primary documentation. COMPANY REPRESENTATION — the company's own statement, recorded as its position and never as proof of the underlying fact. REPORTED ALLEGATION — journalism or pleading, untested. CUSTOMER EVIDENCE — an individual account, used as pattern evidence only. LITIGATION / REGULATORY STATUS — procedural posture stated, merits not asserted. ANALYTICAL INFERENCE — a conclusion drawn by CHI from facts, labelled as such. CHI JUDGMENT — an assessment or calibration decision of the index. NOT ESTABLISHED — tested and unsupported. UNRESOLVED — open in both directions.

10.3 Claim-to-source matrix — Ferrari

10.4 Claim-to-source matrix — Tesla

Entity controlRolls-Royce here means Rolls-Royce Motor Cars Limited, the BMW-owned automobile manufacturer — not Rolls-Royce plc / Rolls-Royce Holdings plc, the separate aerospace, defence and power-systems group, which is out of scope and is described nowhere on this page.

10.5 Claim-to-source matrix — Rolls-Royce Motor Cars

10.6 Cross-cutting and comparative claims

10.7 Unresolved evidentiary limitations carried into the conclusion

1. Ferrari's price stability, service, failure and remedy are unassessed. The conclusion's treatment of Ferrari is therefore confined to access, allocation, exclusivity and disclosure.
2. Rolls-Royce's conduct record is thin where its documents are thick, and remains so after the freeze. Owner forums, confidential settlements, sealed dealer agreements, non-US regulators and the Bespoke contract stack were all unreachable. The published assessment carries the same limitation in its own words: low hostility here means low observed hostility.
3. Three live matters could move findings: Tesla's unresolved HW3 remedy; the California DMV finding under Tesla's court challenge; the Rolls-Royce Spectre battery-backorder pleading, untested and one-sided.
4. Several Rolls-Royce quotations require primary anchoring, above all the March 2026 reversal quotes, whose originals were never fetched.
5. Retroactivity of the 15-year Spectre battery warranty to MY2024–25 cars rests on a press release alone. No contract-level instrument was located.
6. Ferrari's scores are reserved by design and remain so. Rolls-Royce's are now final and frozen at CHI 23 / CVI 85 / CFS +62; its own assessment records that they will be revisited if its monitoring items resolve, chief among them the pleaded Spectre battery-and-parts matter.

↩ Back to the brief