← Back to Companies

Transportation

Uber

Concerning Primary pattern Two-Sided Spread Opacity Most serious finding The fare stopped being reconstructible Methodology CHI/CVI v2.0

Uber never hid the price. It hid what the price was made of.

This is not an assessment of a bad product. Uber is one of the most functionally successful consumer services ever built, and this page scores it that way: a CVI of 82 against category-leading satisfaction, near-universal trip completion, a free and expanding safety toolkit, and a documented improvement on a taxi system that failed a lot of people. Value case leads this page The charge is narrower and structural. Between 2016 and 2022 Uber rebuilt the transaction so that the price a rider pays and the compensation a driver receives became two separately managed instruments rather than two ends of one disclosed formula. Uber’s own filings define its service fee as “the difference between the amount paid by an end-user and the amount earned by Drivers.” Uber 10-K Neither side normally sees that difference, and outside the jurisdictions that have legislated it, neither side can check it.

CHI64/100Concerning
CVI82/100Exceptional
CFS+18Fair

CFS = CVI − CHI (82 − 64 = +18). Delivered value exceeds measured hostility, and that is the finding — but the margin is far narrower than the quality of the product on its own would predict. A service this good at the thing it does should clear its hostility score by more than eighteen points. It does not, because the excellence and the opacity are built into the same machine. Uber is not characterised here as customer-friendly and not characterised as a bad product. Both halves are the assessment. Methodology →

Pattern AnalysisAll patterns →
TYPE SPECIMEN

Two-Sided Spread Opacity

STRONG SUPPORT

Advertising Creep

STRONG SUPPORT

Reference-Price Erasure

STRONG SUPPORT

Loyalty Financialization

Pattern Heatmap
Two-Sided Spread Opacity*
Reference-Price Erasure
Advertising Creep
Loyalty Financialization
Responsibility Diffusion
Commercial Surface Expansion
Access Leverage**
Value Captivity**
Feature Fragmentation**
Lock-In / Exit Resistance
Preference Amnesia
Not established
Strongly supported
* Two-Sided Spread Opacity is a provisional pattern validated for this assessment and not yet a formal CHI Lexicon entry; a dedicated Lexicon page is a separate task and is deliberately not linked from here. ** Partially supported. Shown at reduced weight because the investigation specifically tested these and found substantial counterevidence — surge performs genuine marketplace allocation, documented emergency surges were reversed and refunded, and account deletion and multi-homing are straightforward.

The defining tension

Uber solved a real transportation problem, then built a marketplace sophisticated enough to price both sides of every transaction while making the middle increasingly difficult to see.

Both halves of that sentence are supported, and the page refuses to collapse them. The 2016 move to a single upfront number gave riders something they genuinely wanted: certainty before booking, fewer disputes, and overrun risk absorbed by Uber rather than the passenger. Uber newsroom

The same decision removed the surge multiplier the rider used to have to accept consciously, retired the published rate card, and converted Uber’s compensation from a disclosed commission into a residual. Analytical inference

The convenience is real. The opacity arrived inside it.

CHI v2.0 dimension breakdown
Trust & Transparency14/15
Revenue Extraction19/25
Information & Privacy10/15
Behavioral Manipulation14/25
Customer Restriction7/20
19 + 14 + 7 + 10 + 14 = 64/100. Shown in order of proportional severity rather than methodology order, because the ordering is the argument: Uber’s hostility concentrates in Trust & Transparency and Revenue Extraction — what the customer can see about the transaction, and how much of it Uber captures — and not in lock-in, restricted access, or a product that fails to work. Customer Restriction at 7/20 is the lowest hostility score on this page and is deliberate: a rider can delete the account, open Lyft or Bolt, hail a taxi or take the train, and lose nothing they had stored.

Read this first · Evidence against hostility

What Uber actually delivers.

A CVI of 82 is one of the highest value scores in this index, and it is not a courtesy. Uber replaced a system that failed measurably and unequally, and it did so at a scale almost nothing else in consumer services matches. This section leads rather than trails, because a reader who skips it will misread everything that follows.

01

It replaced a system that was failing people unequally. Independent research — not Uber-funded — found that Los Angeles taxis cancelled on Black riders at substantially higher rates and made them wait materially longer than white riders, while ride-hail completed nearly all trips with far closer parity. Brown, UCLA 2018 That finding is the single strongest item in Uber’s value case and it is used here precisely as the research states it: a documented disparity in the incumbent system that the app-dispatch model narrowed. It is not extended into a claim that Uber eliminated discrimination.

02

Reliability at a scale that is genuinely hard to build. Roughly 202 million monthly active platform consumers and 13.6 billion trips in 2025, across more than 70 countries and 15,000-plus cities. Uber investor materials Uber’s own published safety reporting puts the share of trips completed without any reported safety incident above 99.9%. Uber US Safety Report The pickup happens, the car arrives, the payment clears — and this is the part of the relationship the customer actually contracts for.

03

Upfront pricing gave riders something real. This page treats the 2016 change as genuinely two-sided rather than as a trick. A single number before booking removed post-hoc surprises and meter anxiety, cut disputes, and shifted overrun risk onto Uber — when a trip runs long, Uber’s fee absorbs part of it rather than the rider. Uber pricing documentation That is a durable customer benefit and it is credited in the CVI without qualification. What it also did is the subject of the rest of this page. Both are true simultaneously.

04

The safety toolkit is broad, expanding, and almost entirely free. RideCheck, PIN verification, share-trip, the in-app emergency button, audio recording where available, real-time driver ID verification, teen accounts (2023), senior accounts (2025), and Women Preferences rolled out nationwide (2025–26). Uber newsroom CHI specifically tested whether Uber paywalls safety and found that it does not. Reliability is monetised; safety is not. That distinction matters and Uber gets full credit for it. Uber has also published voluntary US safety reports, which is rare in transport and was not compelled.

05

Satisfaction is category-leading, measured externally. ACSI scored Uber 76 in 2026, tied first in its category. ACSI 2026 Customers who have every complaint on this page available to them keep choosing the product, and that is evidence, not noise.

06

Real product choice, including downward. Riders can pick shared and wait-and-save tiers priced well below standard UberX, standard UberX, Comfort and premium tiers, or Reserve for scheduled certainty. Uber product pages Since 2024 Uber has moved deliberately toward affordability — cheaper shared products, stated intent to hold prices relatively consistent, and insurance savings described as passed through. The direction of travel on price in the most recent period is toward the customer, and this assessment says so.

07

Switching cost is genuinely low, and this is why Customer Restriction is only 7/20. There is no data hostage, no meaningful export problem, no accumulated library to lose. Account deletion is straightforward. Multi-homing between Uber, Lyft, Bolt, taxis and transit is normal customer behaviour and Uber does nothing structural to prevent it. Analytical inference A rider who reads this page and dislikes what it describes can leave on Tuesday and has lost nothing. Very few companies in this index can say that.

08

Fee attribution is mostly honest. CHI expected to find widespread fee-washing and largely did not. Government surcharges, congestion charges, airport fees and driver-benefit levies are named and remitted rather than absorbed; the 2022 fuel surcharge went to drivers; the lost-item fee is a driver reimbursement; and tips have gone 100% to drivers since 2017. Uber fee documentation Where Uber is charging on behalf of somebody else, it generally says so. The transparency problem on this page is about the fare itself, not about the line items around it.

What changed

The fare used to be something you could check.

Nothing in this sequence is secret and none of it is alleged. Every step below is documented by Uber, in its own filings, blog posts and product announcements. The finding is not that any single step was hidden — it is what the five of them add up to.

2012–2016 · Reconstructible

base fare+ (minutes × per-minute rate)+ (miles × per-mile rate)+ booking fee× surge multiplier (displayed; rider had to confirm it)

Every input was published on a city rate card. A rider who wanted to could check the arithmetic afterwards, and a driver was paid on the same variables less a disclosed percentage commission. One formula, two ends, both visible.

2016–present · Residual

rider price — set independently government taxes, tolls, surcharges (remitted) driver compensation — set independently= Uber’s service fee

Uber’s 10-K defines the service fee as “the difference between the amount paid by an end-user and the amount earned by Drivers.” It is no longer a rate. It is whatever is left over — and what is left over is determined by two pricing systems the customer cannot see either half of.

2012–2016
Published rate cards and a visible multiplierBase, time, distance, minimum and a surge multiplier the rider had to type in or tap to accept. Fares fell repeatedly across this period on published cards. The price was often unpredictable in total — but it was reconstructible in composition. Contemporaneous rate cards
2016
Upfront pricing — “no math and no surprises”US pilots in spring, global by December. One number before booking. Genuine value: certainty, fewer disputes, overrun risk moved to Uber. Simultaneously: the multiplier left the rider display, the rate card stopped being a check, and Uber’s compensation became a residual rather than a rate. Uber newsroom
2017
Route-based pricing, acknowledged publiclyUber confirmed it prices routes using machine-learned estimates of what riders on those routes will pay: “we price routes differently based on our understanding of riders’ choices.” It has never retracted this. Bloomberg, May 2017 This page states that at the level Uber stated it — route and segment level. It is not evidence that Uber calculates an individual rider’s personal maximum, and this assessment does not claim that it does. Individual personalisation: not established
2018–2019
Driver surge is formally decoupled from rider surgeUber replaced multiplier-based driver surge with flat additive amounts keyed to the driver’s location rather than the rider’s. Uber’s own explainer notes drivers “may earn less on longer surged trips.” Uber, Under the Hood After this point the premium a rider pays at peak and the premium a driver receives at peak are two different numbers by design.
2021–2022
Driver Upfront Fares complete the separationAlgorithmic per-trip offers replaced formula-based driver pay across US markets. Uber told drivers they “will make less money for longer trips but should earn more on shorter trips.” The Markup, March 2022 From here, rider price and driver compensation are two separately optimised instruments and the gap between them is a variable, not a rate.
One disclosed formulaTwo managed instruments

What Uber retains is not profit. The residual between rider payment and driver compensation absorbs commercial insurance, payment processing, fraud, support, safety and technology costs, driver incentives, and regulatory and operating expense. Uber has made exactly this argument in court, and it is correct. CHI states it here, before the criticism, because any page that treats the gap as Uber’s margin is wrong on the arithmetic and this one does not.

Defining pattern · Type specimen

Two-Sided Spread Opacity.

A platform independently controls or materially influences the price paid by one side of a transaction and the compensation received by the other, while preventing either side from meaningfully observing the relationship between those amounts — allowing the platform to capture a variable and difficult-to-evaluate economic spread. Uber is the type specimen.

Rider pays

One number

Quoted before booking. Accurate, binding, and genuinely useful. The rider can accept it or decline it.

What the rider cannot do is reconstruct it. There is no published rate to check it against, no multiplier shown, and no visible relationship to what the trip costs to supply.

The rider seesThe total, and nothing behind it.

Uber pricing & intermediation layer

Variable spread

Two separate pricing systems operate here: one estimating what the rider will pay for this route at this moment, one estimating what compensation will get this trip accepted.

The gap between them is not a commission rate. Uber’s filings define its fee as the difference itself. It varies trip to trip, and the distribution is not published.

Neither side seesThe relationship between the two amounts.

Driver receives

A separate number

An algorithmic per-trip offer, presented before acceptance. The driver can accept it or decline it.

What the driver cannot normally do is see what the rider paid, what promotions applied to that rider, or how the platform’s share on this specific trip was arrived at.

The driver seesThe offer, and nothing in front of it.
Read the middle column carefully. It is a spread, not a margin, and certainly not profit. Out of it come commercial insurance — the largest single component, and one Uber says has risen sharply per trip — plus payment processing, fraud, safety, support, technology, regulatory compliance and driver incentives. Uber reports a global take of roughly 21% and a US figure it puts well under 20% once insurance is netted out; independent trip-level studies measuring the raw gap between rider payment and driver earnings arrive at higher numbers because they are measuring a different thing. Both can be accurate. That is the point. The reconciliation between those two ways of cutting the same pie is detailed work and it is reserved for the Special Investigation rather than compressed here. CHI does not publish, and this page does not claim, that rider payment minus driver compensation equals Uber profit.
1. Price control on both sides Uber determines the rider’s price and, since 2021–22, the driver’s per-trip compensation, through separate systems with separate objectives. This is not contested; Uber describes both. Uber product documentation
2. Material opacity between them The rider is not shown the driver’s share. The driver is not shown the rider’s price or the promotions applied to it. Colorado is the exception, and it exists because a legislature created it. Colorado SB24-075
3. Platform benefit from the spread Uber’s per-transaction compensation is defined residually rather than as a disclosed rate, which makes the spread itself an optimisable variable rather than a fixed consequence of the fare. Uber 10-K

All three elements must be present to assign this pattern. Ordinary retail markup has one visible price and fails element two. A marketplace where the supplier sets their own asking price fails element one — which is why this pattern is not automatically assignable to home-sharing or third-party retail platforms, and is not being assigned to them here.

Supporting pattern · Strongly supported

Reference-Price Erasure.

A modern Uber fare is hard to evaluate not because it is high, but because there is nothing stable to evaluate it against. The same trip varies with route, time of request, live marketplace conditions, product tier, and whatever account-specific promotions happen to be live. Dynamic pricing in real-time transportation is legitimate and this page does not treat price movement as manipulation. The concern is the loss of an intelligible reference point.

Uber’s own defence makes the point Responding to a 2026 consumer-testing report, Uber argued that “a trip is defined not only by where it starts and ends, but when it is requested.” Uber rebuttal, June 2026 That is a fair rebuttal to a like-for-like comparison claim. It is also a concession: if identity of route is insufficient to define identity of product, then there is no stable normal price for a rider to compare against, and the customer has no way to know whether today’s number is ordinary or not.
And then Uber sold the reference point back In May 2025 Uber launched Price Lock at a monthly fee — a product whose entire function is to give a rider a stable price on a chosen route. Uber product launch This is the cleanest available illustration of the pattern anywhere in this index. Price certainty became a saleable product precisely because the marketplace stopped supplying it as a property of the fare. Uber did not invent volatility to sell the cure — but it is now charging for the thing the 2016 rate card gave away.

Deliberately not claimed here: that Uber sets an individual rider’s base price from their personal behavioural profile. That allegation is unproven, the popular versions of it — low battery, iPhone, card type — are folklore that Uber has specifically denied, and CHI does not publish them. What is supported is narrower: route-level yield pricing that Uber has acknowledged, plus account-specific promotions that mean two riders standing in the same place can legitimately see different numbers.

Supporting pattern · Strongly supported

Advertising Creep.

A rider is paying Uber for transportation. Uber is simultaneously monetising that rider’s attention and trip context, on the same screen, during the same paid transaction — and there is no way to opt out of advertising itself.

2019–2021
Sponsored listings inside Uber EatsThe most defensible surface Uber advertises on. Promoted merchant placement in a marketplace is ordinary discovery advertising and CHI treats it as the weaker half of this finding.
Oct 2022
Uber Advertising launches; Journey Ads run “throughout the entire ride process”The ride surface — not Eats — is the strong case. This is a paid journey the customer has already bought. Uber, Oct 2022
2024–2025
Programmatic buying; JourneyTV screens inside vehiclesIn-car tablet inventory deployed at fleet scale. The advertising follows the customer into the car.
2026
Journey Takeovers on the map, home-screen ride offers, destination offers, off-platform audiencesUber’s own advertising materials describe the moving vehicle icon on the map as “one of the most attention-grabbing spaces in the app.” Uber Ads, Jan 2026 The navigation surface — the one functional element the rider is actually watching for — became inventory.
Described to ridersAdvertising that should “add value” and not interrupt the experience.
Described to investorsPremium CPMs, a “very, very high margin” business, and a run-rate described as well over $2 billion annualised. Earnings remarks

Riders can opt out of ad personalisation. They cannot opt out of advertising, and there is no evidence advertising revenue reduces fares. The customer is paying twice in different currencies for one trip. The $2B figure is the one supported by the dossier; a later and larger number circulating publicly is deliberately not used here because it has not been independently confirmed.

Supporting pattern · Strongly supported

Loyalty Financialization.

Uber ran a free loyalty programme, reduced its benefits, closed it, and then reproduced parts of its value proposition inside paid products. That sequence is documented and Uber’s own language marks the transition.

2018
Uber Rewards launches — freeFree tiers carrying price protection on nominated routes, priority airport pickup, flexible cancellation and preferred drivers. Uber newsroom
2020
Benefits reducedPrice protection removed from the free tier; cash-out of accumulated value withdrawn.
Nov 2022
Uber Rewards closedUber told press at the time: “Nothing is replacing Uber Rewards. However we’ve built out a great membership program called Uber One.” Trade press quoting Uber The two halves of that sentence are the pattern.
2021–2025
The former perks reappear, pricedUber One as a paid membership; Price Lock as a paid restoration of route price protection; UberX Priority as a per-ride fee for a likelier faster match, which Uber correctly notes is “not guaranteed.” Uber product pages

Uber One benefit drift

20215% discountA straight reduction in what the rider pays. Spendable anywhere, because it is not a currency — it is a lower price.
May 20236% Uber CashHeadline number rises; the benefit converts to closed-loop credit redeemable only inside Uber. Announced as making benefits easier to use across Uber’s suite.
SubsequentlyExpiring creditsThe ride benefit became Uber One credits carrying a 60-day expiry and revocable on cancellation. Exact transition date unresolved in the research
Aug 2026Cut to 5%Rate reduced in an email framed as an improvement to the programme. Back to the original headline percentage — but as expiring closed-loop credit rather than a discount.

Uber One is not worthless and this page does not pretend otherwise. The fee has held at its launch price while the categories it covers expanded, and for a heavy delivery user it clears break-even quickly. That value is credited in the CVI. The pattern is the direction: a benefit that began as money off has become money back, inside a closed loop, on a clock, at a rate the company can adjust. Separately, the FTC and 21 states plus DC allege non-consensual enrolment, savings claims that ignore the membership fee, and obstructive cancellation; most of those claims survived a motion to dismiss in April 2026 and are set for trial. Allegations only · Uber denies them and no finding of liability has been made

Supported patterns

What the evidence actually supports.

Primary pattern

Two-Sided Spread Opacity Provisional · type specimen Uber independently determines what the rider pays and what the driver receives, and its own filings define its compensation as the difference between the two. Uber 10-K Neither party normally observes the other’s number or the split, which makes the platform’s per-transaction margin a residual rather than a disclosed rate — and therefore an optimisable variable. Scored primarily under Trust & Transparency with a secondary contribution to Revenue Extraction, and deliberately not counted again across the other dimensions. This is a provisional pattern validated for this assessment; its formal CHI Lexicon entry is a separate task and is intentionally not linked from this page. Analytical inference

Major supporting patterns

Reference-Price Erasure No Lexicon page yet No stable normal price exists for a rider to compare against, by Uber’s own account of what defines a trip — and price certainty is now sold separately as Price Lock. Real-time transportation genuinely requires dynamic pricing, so the finding is about the missing reference point, not about price movement. Uber rebuttal · product pages
Advertising Creep → An ad-free ride surface became an advertising environment spanning the ride flow, the in-car screen, the map icon, the home screen and off-platform audiences, at a run-rate well over $2 billion, with no opt-out from advertising itself and no evidence of fare offset. The Eats case is ordinary merchant discovery; the ride surface is the strong finding. Uber Ads · earnings remarks
Loyalty Financialization No Lexicon page yet Free Rewards devalued then closed, with its price protection, priority and flexibility resurfacing inside Uber One, Price Lock and UberX Priority; the membership ride benefit drifted from discount to closed-loop expiring credit. Genuine value remains for heavy users and is credited in the CVI. Uber newsroom · product terms
Responsibility Diffusion → Automated and increasingly AI-mediated support with no rider phone line, disputes adjudicated on data only Uber holds, discretionary refund policy, cancellation and wait rules that default toward charging, regulatory and labour cost increases surfaced as named customer fees, and a compensation model in which tipping carries part of the load. Supported but secondary — Uber is not primarily a Responsibility Diffusion story, and it has corrected several of these when found against. ACCC 2022 · DOJ 2022

Secondary and qualified

Commercial Surface Expansion Supported · secondary A ride-hailing app deliberately rebuilt as a multi-product commercial funnel: rides, Eats, grocery, delivery, travel and reservations, membership and advertising, with investor materials describing cross-sell and materially higher spend among multi-product users. Investor materials One-app convenience is real and this is kept subordinate to the pricing thesis.
Access Leverage Partially supported Airports, events, transit failures and late nights concentrate dependency, and paid certainty products sit exactly there. But surge performs genuine allocation, emergency surges have caps dating to 2014, and every documented emergency surge was reversed and refunded. CHI does not characterise surge itself as hostile and does not score it as such.
Value Captivity Partially supported Confined to closed-loop mechanics: Uber One credits that expire and are revocable on cancellation, and refund-to-credit options in some contexts. Purchased Uber Cash does not expire and ride refunds are stated to return to the original payment method, so the finding is deliberately narrow.
Feature Fragmentation / Paid Restoration Partially supported Free route price protection returned as paid Price Lock; free priority returned as paid products; the standard free wait window was cut while Comfort sells extended wait. CHI specifically did not establish that standard UberX reliability was deliberately degraded to create demand for Reserve or Priority, and no such claim is made. The evidence for it does not exist.

Score breakdown

Why CHI is still 64.

Because the transaction remains extremely useful while becoming substantially harder to understand. Uber’s hostility does not come from lock-in, restricted access, or a product that fails to deliver — on all three of those it scores well. It comes from what the customer can see about the exchange they are in, and how much of that exchange Uber captures on terms it does not disclose.

CHI v2.0 · 64 / 100 · Concerning
Revenue Extraction19/25
Behavioral Manipulation14/25
Customer Restriction7/20
Information & Privacy10/15
Trust & Transparency14/15
19 + 14 + 7 + 10 + 14 = 64. Trust & Transparency at 14/15 is the highest proportional score on the page and carries the spread finding. Revenue Extraction at 19/25 reflects a residual, undisclosed and variable platform take rather than a high price level — Uber’s fares are not scored as excessive. Customer Restriction at 7/20 is the lowest, and is the number that keeps this assessment out of the hostile band.
CVI v2.0 · 82 / 100 · Exceptional
Core Product Value26/30
Feature & Capability Improvements20/25
Technology & Performance17/20
Trust, Safety & Reliability11/15
Innovation8/10
26 + 20 + 17 + 11 + 8 = 82. Core Product Value is near the ceiling because the core product genuinely works. Trust, Safety & Reliability at 11/15 is the softest CVI dimension — the safety toolkit is strong and free, but reliability of the commercial relationship is where value delivery thins.
CHI64
BandConcerning
CVI82
BandExceptional
CFS+18
ReadingFair

CFS = 82 − 64 = +18. Under CHI v2.0, scores from 0 to +19 are classified as Fair. Uber therefore still delivers more measured customer value than hostility, but only narrowly relative to the exceptional strength of the product itself. The interpretive point is the gap between the two halves: an exceptional product carrying a materially hostile commercial architecture nets out to a modest customer surplus, and a reader who saw only the CVI would badly overestimate the relationship.

What we refused to count

CHI is not simply compiling complaints.

Uber charges more when your phone battery is lowFalse as a pricing claim The origin of this is an Uber executive saying the company had studied the correlation and explicitly does not use it. Repeatedly denied, and in a decade no controlled evidence has appeared. It is the most durable piece of folklore about Uber and it is not published here.
Uber charges iPhone users, or certain cardholders, moreUnsupported Viral single-comparison screenshots, no controlled testing, denied by Uber. Not used.
Uber calculates each rider’s personal maximum willingness to payNot established Uber acknowledged route-level pricing in 2017 and that is what this page states. Individual base-fare personalisation is denied by Uber and no independent test has controlled for timing, GPS and promotions. This is the single most consequential open question about Uber pricing, and CHI is not going to pre-empt it.
Uber One members are deliberately shown higher base faresUnsupported No evidence beyond the observation that members do not receive win-back promotions, which is a different mechanism entirely.
Uber takes 50–70% of the fareExaggerated as an average Individual trips at those levels are documented and real. They are tail cases in a wide distribution, not a mean, and presenting them as typical would be exactly the kind of number this page criticises Uber for not disclosing.
Uber deliberately profiteers during disasters and terror attacksNot supported Surge in these events has been automated, and every documented instance was reversed and refunded. Emergency caps have existed since 2014. A specific widely-repeated claim about a 2024 global IT outage could not be substantiated at all.
It takes 23 screens to cancel Uber OneWorst case, not the standard path This figure comes from regulatory filings describing a worst case within a specific window, not the ordinary cancellation route. Uber disputes it. CHI reports cancellation friction as an allegation under active litigation and does not present the number as the typical experience.
Uber keeps part of the tipFalse since 2017 Tips pass through in full. Stated plainly because the belief persists.
Uber’s take of the fare is Uber’s profitWrong on the arithmetic The residual carries insurance, payments, fraud, support, safety, technology, compliance and incentive costs. Uber makes this argument and it is correct. CHI’s finding is about visibility of the mechanism, not the size of the margin.
Uber’s spread is illegal, or an FTC violationNot asserted Independent legal validation concluded the variable spread appears largely lawful. Exposure attaches to specific representations and to statutory disclosure duties, not to the existence of the spread. This page is a transparency assessment, not an allegation of unlawful conduct.
Surge pricing is inherently hostileRejected Demand-responsive pricing is how a real-time marketplace avoids collapse. What CHI scores is the decoupling of the rider’s premium from the driver’s, and the removal of the multiplier from view — not the existence of dynamic pricing.
Uber degraded standard service to sell Reserve and PriorityNot established An intuitive theory the research specifically tested and could not support. No time series of standard UberX wait times was located that would demonstrate it, so no claim is made.
Uber paywalls safetyNot supported Tested directly and rejected. The core safety toolkit is free. Uber monetises reliability — speed, certainty, scheduling — which is a different and much more defensible thing.
Uber is expensive, and drivers are underpaidNot the finding Both are contested empirical questions with genuine cost drivers behind them, including sharply rising commercial insurance. CHI scores neither as the Uber story. The story is that neither party can evaluate the exchange they are agreeing to.

Reserved

The Number in the Middle

This module is intentionally empty. A CHI Special Investigation into Uber’s two-sided pricing architecture is in preparation and no part of it is represented here.

Reserved for that investigation: full transaction anatomy for a ride and a delivery order · the dealer-spread analogy and its limits · the take-rate definition dispute and why Uber and its critics produce different numbers from the same pie · Colorado’s transparency statute and Uber’s federal challenge to it · paired rider and driver receipts · regulatory policy implications · and where Two-Sided Spread Opacity does and does not transfer to other industries.

No link is provided because the investigation does not yet exist. This page deliberately explains the mechanism only as far as is needed to justify a CHI of 64.

Evidence & methodology

Research statusComplete
Legal validationComplete
Methodologyv2.0
CHI64
CVI82
CFS+18
  • SEC filingUber Technologies, Inc. — Annual reports on Form 10-K (SEC EDGAR)The controlling source for the entire spread finding. Uber’s own definition of its service fee as “the difference between the amount paid by an end-user and the amount earned by Drivers” is what converts this from an inference into a documented structure. Also the source for segment revenue, take-rate presentation and risk-factor language.
  • Investor materialsUber Investor Relations — quarterly results, prepared remarks and presentationsSource for advertising run-rate and margin characterisation, membership scale and cross-sell framing, monthly active platform consumers and annual trip volume, and the investor-facing description of pricing as a managed variable. Where this page contrasts what riders are told with what investors are told, both halves come from named company sources.
  • Company documentationUpfront pricing — UberUber’s current rider-facing explanation of how a fare is produced, including the acknowledgement that price reflects demand patterns for the route and time rather than a published rate card. Read directly rather than paraphrased from criticism of it.
  • Company documentationSafety — UberPrimary source for the CVI safety case: background screening and annual rescreening, emergency assistance with live location, RideCheck, commercial liability insurance, 24/7 trained safety agents, and number masking. Used to establish that the core safety toolkit is not paywalled.
  • Company documentationUber NewsroomAnnouncement source for upfront pricing, the 2017 product changes including in-app tipping, Uber Rewards and its closure, Uber One, Comfort, Reserve, UberX Priority, Price Lock, and the teen, senior and Women Preferences accounts. The chronology on this page is built from Uber’s own announcements.
  • LegislatureColorado SB24-075 — Transportation Network Company TransparencyThe most important external validation of the pattern. A legislature examined precisely this information asymmetry and required disclosure connecting what the consumer paid to what the driver received. It establishes that the disclosure is possible, not that the prior arrangement was unlawful — a distinction this page keeps.
  • RegulatorFTC takes action against Uber over deceptive billing and cancellation practicesThe source for every Uber One litigation reference on this page, including the savings-claim and cancellation allegations. Cited as allegations: Uber denies them, most claims survived a motion to dismiss in April 2026, and no finding of liability has been made. This case concerns the subscription, not the ride spread.
  • CourtUber Technologies, Inc. v. Moss (D. Colo., filed January 2025)Uber’s federal challenge to portions of the Colorado transparency regime, in which it argued among other things that presenting rider price alongside driver compensation could mislead because the difference is not pure profit. The preliminary relief Uber sought was denied and the disclosure requirements took effect. Uber’s argument is credited on this page — it is the reason the spread is never labelled profit here.
  • Company documentationUber, “Under the Hood” — driver surge (August 2019)Uber’s own explanation that driver surge became a flat additive amount keyed to the driver’s location rather than the rider’s multiplier, including the acknowledgement that drivers “may earn less on longer surged trips.” This is the decoupling stated by the company itself.
  • JournalismBloomberg — Uber’s route-based pricing (19 May 2017)Source of the route-level willingness-to-pay acknowledgement: “we price routes differently based on our understanding of riders’ choices.” Never retracted. Used strictly at route and segment level, and explicitly not extended to individual personalisation.
  • JournalismThe Markup — driver Upfront Fares (1 March 2022)Documentation of the shift to algorithmic per-trip driver offers, including paired rider-payment and driver-earnings examples and Uber’s statement that drivers would make less on longer trips and more on shorter ones. Individual paired receipts are reserved for the Special Investigation rather than reproduced here.
  • AcademicBrown, A. — ride-hail and taxi service disparities in Los Angeles (UCLA, 2018)Independent, non-Uber-funded research finding materially higher taxi cancellation rates and longer waits for Black riders relative to ride-hail. The strongest single item in Uber’s value case, and used at exactly the scope the research supports.
  • RegulatorACCC — Uber penalty for misleading conduct (December 2022)Uber admitted contraventions of Australian Consumer Law over cancellation warnings shown to more than two million users inside the free-cancellation window, and over inflated fare estimates for a taxi product. An admitted finding rather than an allegation, and the strongest regulator-established deception in the file.
  • RegulatorUS Department of Justice — Uber wait-time fee settlement (2022)Resolution of claims that wait-time fees were charged to riders who needed additional time to enter a vehicle because of disability. Relevant to Responsibility Diffusion: a facially neutral reliability charge produced a discriminatory outcome until it was corrected under compulsion.
  • Company statementUber response to independent consumer price testing (16 June 2026)Source of the verified quotation that a trip “is defined not only by where it starts and ends, but when it is requested.” Presented here as a fair rebuttal that simultaneously concedes the Reference-Price Erasure point.
  • Company documentationUber Advertising — Journey Takeovers announcement (January 2026)Source of Uber’s description of the moving vehicle icon on the map as “one of the most attention-grabbing spaces in the app.” The company’s own characterisation of the navigation surface as advertising inventory.

Entries without a hyperlink are cited to a specific dated document rather than to a URL, because no stable public canonical link was verified at build time; every one of them is sourced in full in the underlying CHI research dossier and legal validation memo. That register runs to roughly 15,800 words and is deliberately not reproduced on this page. Where this assessment could have used a more dramatic claim and the evidence did not support it, the claim was dropped — the section above headed What we refused to count is a list of exactly those.

Final verdict

EXCEPTIONAL PRODUCT · OPAQUE MARKETPLACE

CHI64/100Concerning
CVI82/100Exceptional
CFS+18Fair

Uber works. That is not a concession offered to soften what follows — it is a finding, scored at 82, and it is the reason this company is interesting rather than easy. The car arrives. The payment clears. The trip is tracked, the driver is accountable, the safety tools are free, and the system it replaced was demonstrably worse for people who were already getting worse service. A rider who has never thought about any of this is not being foolish. They are receiving something valuable.

What changed underneath them is narrow and specific. In 2016 the fare stopped being a formula and became a number. In 2017 Uber acknowledged pricing routes by what riders on them would pay. Between 2018 and 2022 the driver’s compensation was detached from the rider’s price and made algorithmic in its own right. What remains between the two is defined, in Uber’s own filings, as the difference — which means the platform’s share of any given trip is not a rate anyone agreed to but a residual left over between two numbers neither party can see.

None of that appears to be illegal, and this page does not say that it is. Uber is right that the gap is not profit; it carries insurance and a great deal else. Uber is right that riders see a total before they book and can decline it. Uber is right that a real-time marketplace needs dynamic pricing. The assessment survives all three defences intact, because it was never an argument about legality or price level. It is an argument about whether the customer can see the mechanism they are inside — and a company sophisticated enough to independently price both sides of a transaction is plainly sophisticated enough to show both sides what happened to the money.

Uber never hid the price. It hid what the price was made of.
The product works remarkably well. The commercial system built around it has become remarkably difficult to inspect — and it took a state legislature, not a market, to make any part of it visible.