CHI Dating Comparison

A Dating App’s Best CustomerEventually Stops Being One.

Five companies, five ways of handling the fact that the product is supposed to end.

Read this part first. This page does not rank dating apps by quality, size, price, matchmaking skill or admissions standard. It does not argue that charging for dating is hostile, that scarcity is hostile, that selectivity is hostile, or that the largest app is the worst one. None of those propositions appears anywhere below.

What it examines is narrower: given that the customer’s stated objective is to meet a real, appropriate, realistically dateable person and eventually stop needing the app — whose product architecture behaves as though that outcome is the point?

HingeProgression-pricedCHI 57 · CVI 81 · CFS +24 RayaAccess-pricedCHI 39 · CVI 79 · CFS +40 LuxyAccess & position pricedCHI 76 · CVI 57 · CFS −19 BumbleUncertainty-pricedCHI 70 · CVI 68 · CFS −2 TinderAttention-pricedCHI 68 · CVI 67 · CFS −1

Each company below holds its own independently researched CHI assessment. This page does not re-score any of them. It sets five answers to the same commercial problem beside one another, and asks which architecture survives the customer succeeding.

Every other subscription wants you to renew. A dating app has to sell you something while quietly hoping you cancel.

The industry’s structural problem is not a scandal. It is arithmetic.

Almost every complaint written about dating apps assumes bad faith at the outset. That assumption is unnecessary, and it makes the analysis worse. The interesting thing about this sector is that its conflict of interest is structural — it exists whether or not anyone at any of these companies ever intends anything.

The customer’s objectiveMeet a real, appropriate, realistically dateable person — and stop needing the product. Success is defined by exit.

The company’s objectiveRevenue, which in a subscription and consumable business is generated by continued presence. Success is defined by retention.

The conflictNot an accusation. Arithmetic. Every company in this comparison operates inside it, and none of them chose it.

CHI does not begin at the conflict. Every dating company has it. CHI begins with what each company does about it — which behaviours the product architecture rewards, what gets priced when a customer is uncertain, and whether anything in the design would make sense only if the customer eventually left.

The framework

Five questions. Applied identically to all five companies.

The first and fourth questions credit the company. The second, third and fifth are where the comparison does its work.

What does it sell?

The core proposition, in the company’s own marketing language. What is the customer told they are buying?

Company’s own account

What does it optimize?

What behaviour does the product architecture actually reward — swiping, messaging, meeting, returning, upgrading? Mechanics, not mission statements.

Architecture

What does uncertainty cost?

When a customer is unseen, unsure, out of allowance or out of time, what is for sale at that exact moment? This is the sector’s most legible variable.

Monetized state

What genuinely differentiates it?

Every company here does something real that its competitors do not. Recorded at the same weight as the findings against it.

Credit where due

Does it work if you leave?

Would any part of the architecture make sense only on the assumption that the customer eventually stops using it? This is the alignment test.

The alignment test

The most customer-aligned dating app is not the one with the most profiles, the highest price, the strictest door or the cleverest algorithm. It is the one whose product still makes sense when you stop using it.

Published scores

Five companies, three numbers each.

CHI measures hostility. CVI measures delivered customer value. CFS is simply CVI − CHI. None of the three is an alignment measure, and the ranking further down this page deliberately disagrees with all of them in places.

Hinge

Progression-priced

CHI57

CVI81

CFS +24Customer-Friendly

Highest CVI in the dating cohort. The hostility is real and so is the value.

Raya

Access-priced

CHI39

CVI79

CFS +40Strongly Customer-Favorable

The strongest CHI/CVI balance in the group by a wide margin.

Luxy

Access & position priced

CHI76

CVI57

CFS −19Extraction exceeds value

Highest hostility and lowest CFS here — alongside a real, differentiated authenticity benefit.

Bumble

Uncertainty-priced

CHI70

CVI68

CFS −2Finely balanced

A capable mainstream product whose original differentiation has materially weakened.

Tinder

Attention-priced

CHI68

CVI67

CFS −1Finely balanced

Enormous marketplace value very nearly offset by the intensity of what sits on top of it.

These are not alignment scores, and the difference matters. Luxy holds the worst CFS in the group and finishes third on alignment. Raya holds the best CFS and finishes second. A CFS is a net of extraction against delivered value at a point in time. Alignment asks a different question — whether the architecture behaves as though the customer’s eventual departure is the objective — and the two answers are not required to agree.

What each app actually sells

Five propositions, compressed to the sentence the customer receives.

The meter measures one variable only: how much of what is being sold depends on the customer still searching. It is not a hostility ranking and it is not a product-quality ranking.

Hinge

“Get better at dating, and you will need us less. Here is what it costs to get there sooner.”

Progression

Raya

“There are fewer people in here, on purpose. Here is what it costs to see them faster.”

Access

Luxy

“These are real, reviewed people. Here is what it costs to reach them, be seen by them, and recover the ones who expired.”

Access + position

Bumble

“Somebody in here already chose you. Here is what it costs to find out who.”

Uncertainty

Tinder

“Everyone is in here. Here is what it costs to be noticed among them.”

Attention

Read down the meter and the sector’s economics become visible without anyone having to be accused of anything. At the top, revenue is attached to movement — communication, meeting, progress. At the bottom, revenue is attached to position — being seen, being noticed, being ahead. Movement ends when the customer succeeds. Position does not.

The distinction the page rests on

Monetizing the journey vs. monetizing the friction.

Two different objects. Conflating them produces both of this sector’s standard analytical errors — treating every paid feature as predatory, and treating every free core product as exoneration.

Monetizing the journey

Charging for things that help a customer move toward the outcome they came for. CHI records these; it does not treat them as findings on their own.

A subscription for a working product
Better filtering and search
Stronger signals of genuine intent
Verification and safety investment
Curation that costs money to perform
More conversations with people you matched with

CHI verdictNot inherently hostile

Monetizing the friction

Charging for the resolution of a state the product itself creates or withholds. This is where conduct becomes assessable.

Selling the identity of someone who already liked you
Selling position in a queue the company controls
Selling recovery from an expiry the company set
Selling relief from a cap the company defined
Selling as “unlimited” a capability reserved one tier up
Selling visibility inside a pool the customer already paid to enter

CHI verdictThis is where CHI begins

A company is entitled to charge for a dating product. What it chooses to withhold in order to have something to sell is a separate question, and it is the only one this page asks.

Primary comparison matrix

Sixteen dimensions, five companies, qualitative values only.

No dimension below is scored numerically and no company is re-scored on this page. Values are deliberately coarse — Strong, Present, Partial, Limited, Minimal, Reported, Not established, None published, Not assessed — because the underlying evidence does not support finer resolution. Columns run in the alignment order this page argues for, which is stated openly rather than smuggled in through the layout.

Neutral a recorded fact, not a finding Favourable cuts in the customer’s favour Limited Present Strong increasing customer-side burden Not assessed untested — never a clearance Not established means tested and unsupported. Where something simply is not published, the cell says None published, and Reported marks something evidenced at report level only.

Hinge, Raya, Luxy, Bumble and Tinder compared across sixteen dating-sector dimensions.
Dimension	HingeProgression-priced	RayaAccess-priced	LuxyAccess & position	BumbleUncertainty-priced	TinderAttention-priced
Core proposition	HingeRelationship progression. Its own hero line: “a dating app designed to get you off dating apps.”	RayaMembership of a restricted, curated social and dating environment. Entry, not reach.	LuxyA selective, affluent, human-reviewed pool. “All new profiles undergo a 24-hour review by the Luxy team.”	BumbleHistorically a safer, women-first mainstream alternative to Tinder. Now a three-tier freemium attention marketplace.	TinderMaximum dating-market liquidity — the largest available pool of people to consider.
Current CHI	Hinge57 / 100Concerning. Revenue Extraction 18/25 is the largest single component.	Raya39 / 100The lowest in the dating cohort. Behavioral Manipulation 6/25; Information & Privacy 2/15.	Luxy76 / 100The highest here. Finalized on the Luxy assessment when the ten-day BLACK field study closed.	Bumble70 / 100Information & Choice Manipulation 18/20 is its highest dimension, almost entirely Weaponized Curiosity.	Tinder68 / 100Behavioral Manipulation 19/25 is its highest dimension. Published as a working v2.0 assessment.
CVI	Hinge81 / 100The highest customer-value score awarded to a dating company in this index.	Raya79 / 100Trust, Safety & Reliability 14/15 — the strongest safety sub-score in the cohort.	Luxy57 / 100Depressed by geographic relevance in Core Product Value and by documented reliability problems.	Bumble68 / 100Published on the Bumble assessment at cohort-normalized level; its CHI dimension scorecard remains on an earlier five-by-twenty rubric.	Tinder67 / 100A usable free core plus real verification, safety and anti-fraud investment.
CFS	Hinge+24Customer-Friendly. 81 − 57.	Raya+40Strongly Customer-Favorable, and the best figure in the group. 79 − 39.	Luxy−19Extraction exceeds delivered value. 57 − 76.	Bumble−2Finely balanced. 68 − 70.	Tinder−1Finely balanced. 67 − 68.
Profile authenticity	HingePresentSelfie Verification and Face Check, phased where local regulations permit. Biometric retention disclosed in the privacy policy.	RayaStrongPre-entry human review, identity and accountability controls, anti-screenshot enforcement, active moderation.	LuxyStrong (firsthand)Documented 24-hour human review, plus a decade of firsthand use reporting dramatically fewer obviously fake profiles than Tinder or Bumble. Not an audited market-wide fake-profile statistic, and not asserted as one.	BumblePresentPhoto and ID verification, Private Detector explicit-image blurring, Deception Detector scam and spam blocking.	TinderPresentPhoto and identity verification including Face Check. Company-reported >60% lower exposure to suspected bad actors and >40% fewer bad-actor reports following Face Check.
Local relevance	HingeNot established as a failureDaters set where they date and adjust distance. Geographic Irrelevance was tested on the Hinge assessment and not established.	RayaSecondary concernRecorded as minor on Raya’s own assessment. Membership is small by design, which constrains local density everywhere.	LuxyThe defining failureLuxy documents its own behaviour: it “recommends people from further away if we think you fit.” Field study: 3 of 601 BLACK Match-feed profiles were in the customer’s city — 0.5%.	BumblePartialDiscovery is geographically constrained, but distance and advanced filtering sit behind Premium. Two advanced filters were free to every user in December 2018.	TinderPartialDistance controls exist; Passport and broader control are gated to Plus and above.
User control	HingeStrongProfile pausing — invisible in Discover and Standouts while keeping existing matches — plus granular notification categories. Voluntary, reversible invisibility is the opposite of retention architecture.	RayaLimited but honestThe daily profile limit is the product, and predates the paid workarounds. Screenshot prohibition is customer-protective rather than restrictive.	LuxyWeakFilters, anonymous and hide modes, Who Likes Me and Search By Location are all premium-gated — and the recommendation feed overrides the customer’s stated first-priority constraint anyway.	BumblePartialAll advanced filters now require Premium; Incognito is paid. Notification categories are adjustable and cancellation guidance is explicit.	TinderPartialMutual matching and messaging stay free, but unlimited Likes, Rewind, Passport and distribution advantages are gated.
Messaging access	HingeFree at the coreLikes and comments are free and profile-specific. In Standouts, only Roses can be sent — a paid channel into the most-liked feed, not a cap on ordinary conversation.	RayaPaid pre-match channelDirect Requests at $4.99, or three for $12.99 — a premium pre-match communication mechanism, priced but not disguised.	LuxyCapped and mis-describedThe Play listing sells “Unlimited messages to local millionaire singles.” The Help Center states BLACK members “can send BLACK Messages to up to 3 different singles daily,” with more via Coins, and reserves unlimited BLACK Messages to PLATINUM. Conversation with mutual matches is unlimited and free, and Luxy says so plainly elsewhere.	BumbleFree but timedCore matching and messaging are free. Matches expire in 24 hours; one free daily Extend remains, with unlimited extensions and Rematch paid. The timer existed at launch and is not later degradation.	TinderFree at the coreMutual matching and messaging remain free. What is priced is who you can reach and how quickly you can reach them.
Visibility monetization	HingePresentHingeX brings selected profiles to the top of Discover feeds and boosts the subscriber’s own profile; Priority Likes hold position for seven days; Boost and Superboost sell one hour and twenty-four hours of elevated visibility.	RayaNot establishedNo paid-position product is documented. Raya+ raises how many profiles a member may consume, not how prominently they are shown.	LuxyPresentPLATINUM profiles are “displayed to matching singles with greater priority,” with further Boosts purchasable in Coins. The intuitive split — Tinder sells visibility, Luxy sells curation — does not survive Luxy’s own documentation. It sells both.	BumbleStrongSpotlight and SuperSwipe are sold separately, and Premium+ explicitly promises to fast-track likes and prioritize a subscriber in other members’ feeds.	TinderStrongBoost, Super Like, Platinum’s Priority Likes and Select each buy a better position in a line Tinder itself controls. Relative standing, not a measured queue.
Premium-on-premium	HingePresentOne free Rose weekly, non-accumulating; more sold in packs of 3, 12 or 50 on top of a Hinge+ or HingeX subscription.	RayaPresentSkip the Wait at $7.99 (3 for $19.99, 5 for $29.99), Extra Likes at 30 for $10.99, Direct Requests — all above a $24.99 or $49.99 monthly membership.	LuxyThe densest in the groupA Coin currency funding at least six separate capabilities, in packs from $2.99 to $149.99, above App Store BLACK price points running $69.99 to $449.99 — and above that a PLATINUM tier at “Starting at $199.99/month.”	BumbleStrongBoost → Premium → Premium+, supplemented by separately purchasable Spotlight, SuperSwipe and Compliments.	TinderStrongFive rungs from Free to Select, with Boosts and Super Likes purchasable alongside. Select confirmed active at $499/month as of June 2026.
Curation / vetting	HingeAlgorithmic, not pre-entryNo admissions process. Standouts is proprietary algorithmic curation — and the assessment documents a severe historical failure in which that system elevated a user its safety system had already received a rape report about.	RayaStrong and genuineRoughly 2.5 million applicants against a membership in the low six figures. Independent reporting documents applicants — some with multiple referrals — waiting two, five, six and seven years.	LuxyDocumented, with a paid bypassHuman review is documented rather than merely claimed. But: “you can upgrade your account to a premium membership to skip waiting and gain access immediately,” with review still to follow. A first-party contradiction also exists on whether members vote on profiles.	BumbleNone pre-entryOpen entry; verification and safety tooling applied after the fact. Not a criticism — a description of the mainstream model.	TinderNone pre-entryOpen entry by design. Openness is the product: it is what produces the liquidity.
Advertising / upselling	HingeIn-product onlyNo third-party advertising business documented. Upsell is present and persistent.	RayaAd-freeAdvertising Creep was tested and returned “No.” The commercial relationship is customer-pays-company rather than advertiser-pays-for-attention.	LuxyThird-party ads inside a paid appA media kit sells splash screens where “All users will see your brands,” business-card swiping, banner and feed units, sponsored virtual gifts and “Push notifications for targeted users.” Firsthand: ads between profile swipes, and years of continuing in-app upgrade solicitation. Luxy never promised any paid tier would be ad-free, and no such promise is asserted.	BumbleIn-product upsellA layered consumable and tier architecture. No third-party advertising business documented on the assessment.	TinderIn-product upsellA six-prompt upsell architecture running from “Curious who likes you?” to “Hit a limit? → Upgrade.”
Technical reliability	HingeNot a findingOrdinary friction only.	RayaNot a findingTechnology & Performance 15/20; no reliability deduction of note.	LuxyDocumented, not moralizedSlow loading and a Match feed that sometimes returns nothing until the app is restarted — corroborated by Luxy’s own troubleshooting articles rather than by complaints. Recorded as ordinary friction and a CVI deduction, not converted into an ethical pattern.	BumbleNot a findingNo reliability finding on the assessment.	TinderNot a findingNo reliability finding on the assessment.
Incentive alignment	HingeStrongest evidenceWe Met asks whether the date happened. Your Turn Limits stop a dater who is not replying — and paying subscribers are not simply exempted. Signals cannot be purchased at any tier. A monetization-first company would have sold the exemption.	RayaStrongConsumption is capped by design and the cap predates the paid workarounds. Raya has reportedly resisted investor pressure and clear revenue opportunities to admit substantially more members.	LuxyMixedUnlimited free conversation with mutual matches is genuinely aligned. Pre-match reach, position, expiry recovery and geographic control are all priced, and the feed substitutes its own judgment for the customer’s stated first constraint.	BumbleWeak-to-moderateThe core match-and-chat loop still works without paying — Control & Dependency scores 9/20, and that gap is load-bearing. But the highest-scoring dimension is built on knowing who liked you and selling the answer.	TinderWeakestUncertainty, low visibility and continued participation are the most monetizable states in the product. Four consecutive reported quarters show Tinder payers down year-over-year while revenue per payer rose.
Genuine differentiation	HingeProduct-level alignmentThe only company here that has shipped mechanics which are hard to explain unless successful dating is an objective — and it accepted engagement cost to do it.	RayaReal restriction, honestly keptActual curation, a genuinely small pool, privacy enforcement, no advertising, and the cleanest promise-to-delivery relationship in the group.	LuxyApparent human authenticityThe strongest positive finding on its own assessment, and the one thing in this comparison a decade of firsthand use supports and the mainstream alternatives do not match.	BumbleSafety investmentReal and continuing. Opening Moves (April 2024) was a free, choice-expanding change rather than an upsell — but it also softened the founding rule.	TinderScaleLiquidity is genuine, valuable and extremely hard to replicate. No other company here can offer it, and this page does not dismiss it.
Customer-business alignment	HingeStrongestAggressive monetization sitting on top of genuinely exit-oriented mechanics.	RayaStrongSells restriction, delivers restriction, and prices relief from it openly.	LuxyMixedPossibly better humans, delivered in a geographically remote, opaque and heavily monetized way.	BumbleWeak-to-moderateStill useful; economically converging on the model it was founded to differ from.	TinderWeakestThe most valuable marketplace, attached to the revenue model least helped by customers leaving it.

How to read the colouring. Severity styling marks the direction of the customer-side burden on that dimension only. It is not a verdict on the company and it is not a product-quality signal: Tinder carries the group’s strongest differentiation row and its weakest alignment row, and both are accurate. Where evidence differs between a company’s documentation and a customer’s report, the cell says so — firsthand observation is never written here as company policy. Nothing in this table is a score. “Not established” appears only where a proposition was tested and did not survive.

Company by company

The same five questions, five times.

The order below follows this page’s alignment ranking. Each company’s score is shown exactly as its own assessment publishes it.

01 — Strongest alignment

Hinge

Progression-priced

CHI 57/100 · CVI 81 · CFS +24

What it sells: Relationship progression. Hinge’s own positioning is that the app exists to be left, and its assessment’s hero line puts it plainly: “a dating app designed to get you off dating apps — with an increasingly sophisticated business built around what happens before you leave.”
What it actually optimizes: Communication and dating behaviour more than swipe volume, and the mechanics are checkable. Your Turn Limits stop a dater from sending or receiving new Likes when too many people are waiting on a reply. Signals are awarded on participation behaviour — looking before liking, commenting, reviewing likes, messaging, confirming dates — and the Help Centre states eligibility is “based on requirements related to participation on Hinge, not your subscription status.” We Met asks, days after numbers are exchanged, whether the date happened and whether the dater would see the person again, and feeds the answer back into recommendations. Liking is profile-specific rather than whole-person, which is slower and more effortful than swiping.
What uncertainty costs: A great deal. Standouts is a feed of the profiles catching the most attention, and in Standouts only Roses can be sent — one free weekly and non-accumulating, more in packs of 3, 12 or 50. HingeX sells being seen sooner, seen by more people, and staying at the top of someone’s Likes You page for seven days. Priority Likes hold that position for seven days. Boost and Superboost sell an hour and a day of elevated visibility. Hinge is not the low-hostility option in this group. It is the aligned one.
Genuine differentiation: Hinge is the only company in this comparison that has shipped product mechanics which are difficult to explain unless successful dating is an actual objective — and which cost it engagement to ship. The decisive detail is not that Your Turn Limits exist; it is that Hinge+ and HingeX subscribers are not simply exempted from them, and that a Signals badge cannot be bought at any price. A company optimizing purely for monetization would have sold the exemption. It is also the highest CVI awarded to a dating company in this index.
How aligned with exit: Strongest in the group — and that is a statement about architecture, not about hostility. Its strongest product mechanics are built to get customers into real relationships. Its strongest revenue mechanics sell better position in the race to get there.

Counterevidence and limits. Hinge’s CHI of 57 is not low, and the assessment does not pretend otherwise: Revenue Extraction scores 18/25. It also documents a severe historical safety-integrity failure in which the Standouts curation system elevated a user its safety system had already received a rape report about — proprietary algorithmic elevation is a statement, and it made that statement about him. A firsthand CHI account was deactivated after an apparent report with no meaningful explanation ever provided; the recent premium purchase was refunded without resistance. Both are recorded at equal weight. And the conclusion the Hinge assessment expressly does not draw: that Hinge deliberately wants customers to stay single. That was tested and is not established.

Read the full Hinge assessment →Artificial Scarcity →

Raya

Access-priced

CHI 39/100 · CVI 79 · CFS +40

What it sells: Restricted access to a controlled social and dating environment. Not reach, not volume, not liquidity — entry.
What it actually optimizes: Scarcity, privacy, controlled membership and reduced consumption. The daily profile limit is not a paywall dressed as philosophy: it predates the paid workarounds and was Raya’s stated anti-infinite-scroll position. Roughly 2.5 million applicants sit against a membership in the low six figures, and independent reporting documents applicants — some with multiple referrals — waiting two, five, six and seven years. Raya has reportedly resisted investor pressure and clear revenue opportunities to admit substantially more members.
What uncertainty costs: Relief from Raya’s own restrictions, priced openly. Skip the Wait at $7.99, or three for $19.99 and five for $29.99, bypasses the delay after the daily profiles run out. Raya+ at $49.99 raises the allowance itself. Extra Likes run 30 for $10.99 and Direct Requests $4.99, or three for $12.99. Its own assessment’s formulation is the sharpest available: Raya didn’t invent the wait. It found a price for skipping it.
Genuine differentiation: Actual curation and controlled membership, and the group’s cleanest relationship between what a company says it sells, what it delivers, and how its restrictions serve the proposition. Raya is more comfortable than any mainstream app with a smaller pool, less consumption and restricted access. It enforces anti-screenshot rules, identity and accountability controls and active moderation — the strongest Trust, Safety & Reliability sub-score in the cohort at 14/15 — and it takes no advertising money: the commercial relationship is customer-pays-company, not advertiser-pays-for-attention.
How aligned with exit: Strong. A product that caps consumption, refuses to scale its pool and sells no visibility is not structurally dependent on the customer continuing to search. What it does depend on is the customer continuing to belong.

Counterevidence and limits. Raya’s restrictions are not automatically benign and its assessment does not treat them so. Pricing has moved from roughly $7.99/month in 2018 to $24.99 standard and $49.99 for Raya+ today — a Price Creep finding, confirmed. Waitlist status is indefinite and uninformative, recorded as the provisional concept Application Opacity, which describes the silence rather than the rejection: capping or refusing applicants is the product working as designed and scores nothing. Published terms indicate unused subscription value may not necessarily be refunded after termination. And Artificial Scarcity was tested at Raya and not supported — the evidence does not establish that the limit was invented in order to sell the exception.

Read the full Raya assessment →Convenience Tax →

Luxy

Access & position priced

CHI 76/100 · CVI 57 · CFS −19

What it sells: A selective, affluent, supposedly vetted dating environment. The vetting is documented rather than merely claimed: “All new profiles undergo a 24-hour review by the Luxy team,” covering “your profile bio, your occupation, and your profile pictures.”
What it actually optimizes: Several things at once, which is the difficulty. Premium access, curation, profile volume, monetization ladders, positional advantage — PLATINUM profiles are “displayed to matching singles with greater priority” — and an algorithmic substitution of the company’s compatibility judgment for the customer’s stated preferences. Luxy documents that substitution itself: it “recommends people from further away if we think you fit.”
What uncertainty costs: The densest ladder in this comparison. A BLACK subscription at App Store price points running $69.99 to $449.99; BLACK Messages capped at three per day with more bought in Coins; a Coin currency funding at least six separate capabilities in packs from $2.99 to $149.99 — Boosts, extra swipes, additional BLACK Messages, Super Likes, Read Receipts, video calling, and reconnection with matches that expired after 72 hours; a PLATINUM tier above it starting at $199.99/month which is where “unlimited access to BLACK Messages” actually lives; visibility priority; a third-party advertising business sold inside the paid app on the promise that “All users will see your brands”; and, firsthand, years of continuing in-app upgrade solicitation.
Genuine differentiation: Apparent human authenticity, and this comparison credits it. In long-term firsthand use by a paying subscriber of roughly ten years, obviously fake or non-human profiles are dramatically rarer than on Tinder or Bumble. That is a real, differentiated benefit that neither mainstream alternative matches. It is not an audited market-wide fake-profile statistic, no authoritative comparative measurement establishes one, and this page does not assert a market-wide ranking.
How aligned with exit: Mixed. Luxy may well be delivering better humans while doing an unusually poor job of delivering them in a geographically relevant, transparent, customer-controlled way.
: Its defining CHI characteristic is Intellectual Disrespect: broad customer-facing promises that collapse into narrower technical definitions when examined closely. The Google Play listing sells “Unlimited messages to local millionaire singles”; the Help Center caps BLACK at three per day and reserves unlimited to the tier above. The company demonstrably knows how to write the smaller, true version — it does so for the free tier, which “allows you to hold unlimited conversations with your matches.”

Counterevidence and limits. Luxy is not the easy case its price point invites, and several accusations against it were tested and did not survive. Exit Resistance: low — cancellation runs through the app stores and is straightforward. Loyalty Penalty: not established, and reversed — the long-term customer holds a materially better evergreen rate than the headline price, and at that rate he says the exchange works. “BLACK was promised ad-free”: not established — no first-party ad-free promise for any paid tier was found, and none is asserted. Reported violations are generally acted on. Technical problems are recorded as ordinary friction and a CVI deduction, not converted into an ethical pattern. And a subscriber of roughly a decade continues to prefer Luxy to Tinder and Bumble with full knowledge of everything above.

Read the full Luxy assessment →Intellectual Disrespect →Algorithmic Replacement →

Bumble

Uncertainty-priced

CHI 70/100 · CVI 68 · CFS −2

What it sells: Historically, a safer and more controlled mainstream alternative to Tinder, distinguished by a rule about who speaks first. Currently, a three-tier freemium marketplace layered with separately purchasable consumables.
What it actually optimizes: Mainstream participation, plus a growing premium discovery and visibility economy. Payment increasingly determines how much interest a customer can express, whether they can identify who already likes them, whether expiring opportunities can be recovered, how widely and precisely they can search, and how prominently they themselves are shown.
What uncertainty costs: This is Bumble’s defining CHI concept and the reason its highest-scoring dimension is Information & Choice Manipulation at 18/20. Weaponized Curiosity: Bumble necessarily knows which members have liked a customer, because that expression of interest is required to form a match. Free customers see that interested people exist — a blurred grid of admirer profiles — while the identity behind the blur resolves through Premium or through organic discovery. Around it sit Spotlight, SuperSwipe, Compliments, Premium and Premium+, which explicitly promises to fast-track likes and prioritize a subscriber in other members’ feeds, plus paid unlimited match extensions and Rematch above the one free daily Extend. The interest was already real. The uncertainty is the product.
Genuine differentiation: Still a capable mainstream product with meaningful and continuing safety and verification investment: photo and ID verification, Private Detector explicit-image blurring, Deception Detector scam and spam blocking, in-app voice and video calling, date sharing and tracked reporting. Opening Moves in April 2024 was a free, choice-expanding change rather than an upsell. Bumble has publicly criticized its own swipe model and announced a redesign addressing dating-app fatigue.
How aligned with exit: Weak-to-moderate. Bumble remains genuinely useful, and the core match-and-chat loop still works without paying — Control & Dependency scores 9/20, and that gap is load-bearing. But its economics increasingly resemble the mainstream swipe-app model it was founded to differ from.
: The differentiation decay is documented rather than asserted. The women-initiate rule was softened by Opening Moves in 2024, and a more fundamental reversal was announced for late 2026 but had not been broadly deployed as of the assessment’s research cut-off. That is a weakening of the founding proposition, recorded at the level the evidence supports — not a claim that the model has already been abolished.

Counterevidence and limits. Several of the harsher available claims were tested at Bumble and refused. Covert desirability scoring: not counted — anecdotes and forum theories exist; Bumble-specific evidence sufficient for scoring does not. Deliberate shadowbanning to induce purchase: not counted — paid prioritization is documented, secret punitive suppression is not. Individualized or age-based pricing: not counted for Bumble — the credible age-pricing findings concern Tinder and were deliberately not imported. Fabricated likes: not counted. Deliberately preventing successful matches: not counted — financial data alone cannot establish intent. Two structural mitigations also stand: the beeline has been a paid feature since 2016, so the concern is the packaging of partial knowledge rather than a removed free entitlement, and the 24-hour timer existed at launch and is not later degradation.

Read the full Bumble assessment →Promise Reversal →

05 — Weakest alignment

Tinder

Attention-priced

CHI 68/100 · CVI 67 · CFS −1

What it sells: Maximum dating-market liquidity. More people, considered faster, than anywhere else.
What it actually optimizes: Competition for attention inside that marketplace. Tinder created not only a dating marketplace, but a second marketplace for being noticed inside Tinder — and the second one is where the pricing lives.
What uncertainty costs: Five rungs. Free matches and messages; Plus removes limits and adds Rewind and Passport; Gold reveals who already likes you; Platinum buys priority distribution; Select, invite-only, adds enhanced visibility and access, confirmed active at $499/month as of June 2026. Alongside them, Boost sells temporary exposure and Super Like a stronger signal. Each is a different way of buying a better place in a line Tinder itself controls. The upsell architecture is legible in one column: curious who likes you → Gold; want more visibility → Boost; want a stronger signal → Super Like; want priority → Platinum; want access → Select; hit a limit → upgrade.
Genuine differentiation: Scale, and it is real. Liquidity is the single hardest thing to build in a two-sided market and Tinder has more of it than anyone. No other company in this comparison can offer a customer that many people, and nothing on this page dismisses the network effects that produce it. Tinder has also invested materially in verification and anti-fraud: it reports over 60% lower exposure to suspected bad actors and over 40% fewer bad-actor reports following Face Check.
How aligned with exit: Weakest in the group — which is a statement about incentive architecture, not about product quality. The careful formulation is the defensible one: continued uncertainty, low visibility and ongoing participation create highly monetizable product states. No claim of intent is required, and none is made.

Counterevidence and limits. Tinder’s own assessment refuses more claims than any other page in this cohort, and this comparison inherits every one of those refusals. Not established: that Tinder fabricates Likes to induce subscriptions; that it universally shadowbans customers; that it deliberately suppresses customers after they subscribe; that paying customers are secretly penalized in ranking; that a “Likes waiting but no profiles left” state proves intentional withholding. Gold provides genuine efficiency by revealing real inbound interest — the objection is to the mechanism, not its authenticity. The reported figures on payers and revenue per payer do not prove hostility by themselves: higher RPP is consistent with pricing, duration, mix, reduced discounting, currency effects or product changes. What is supported is narrower — four consecutive reported quarters of fewer payers year-over-year alongside rising revenue per payer.

Read the full Tinder assessment →Upgrade Escalation →

Field-test callout

601 profiles. Three of them were in the city.

One quantitative exhibit inside a broader argument, reproduced here in summary only. The full methodology belongs on the Luxy assessment, not on a comparison page.

10days logged

601BLACK Match-feed profiles

3profiles in Prague

0.5%local relevance

Every profile served into the paid BLACK Match feed was logged prospectively over ten days. Three of 601 were in the customer’s city. That is 0.5%.

The number matters because of what preceded it. This customer had told Luxy repeatedly that location is his single most important filter, above every other compatibility factor. Luxy’s own Help Center describes the behaviour that produces the result: the app “recommends people from further away if we think you fit.” And Luxy’s recurring support answer has been to point at Search By Location — a premium tool for connecting with singles worldwide.

That answer resolves a different question. The complaint is not how do I manually search Prague. It is why the paid recommendation feed keeps substituting the company’s compatibility judgment for the constraint the customer ranked first. That substitution is Algorithmic Replacement; answering it with an adjacent tool is Intellectual Disrespect; and the resulting feed is the clearest support yet for Artificial Abundance as an observation — a feed that is full without being useful.

What this exhibit is not. It is one customer, one city, one ten-day window and one tier. It is not a measurement of Luxy’s global inventory, not evidence about any other city, and not a claim that Luxy misrepresents its pool size. Sixty remote profiles is not a better answer than ten local ones. It is a different answer — and that is the whole of the finding.

The full field test on the Luxy assessment →Algorithmic Replacement →

The ranking

Customer-business alignment, first to fifth.

This ranks one thing: how closely each company’s product and economic architecture tracks the customer’s stated objective of meeting someone and eventually leaving. It is not a quality ranking, not a popularity ranking, not a value-for-money ranking, and — as the second column shows — not the CFS ranking either.

HingeStrongest alignment

The strongest evidence in this sector that successful dating is itself a product objective — mechanics that cost engagement, that paying subscribers cannot buy their way out of, and that measure whether a date actually happened. It monetizes attention heavily anyway. Both facts are true at once.

RayaStrong alignment

The cleanest relationship between promise, restriction and delivered environment. It sells a smaller room and delivers a smaller room. What it monetizes is relief from limits it defined — priced openly, on top of a product that takes no advertising money and sells no position.

LuxyMixed alignment

Real differentiated value — the authenticity benefit is the strongest single positive in this comparison — undermined by a promise-versus-delivery gap, the densest monetization stack in the group, and a recommendation feed that overrides the customer’s stated first constraint. Third on alignment while last on CFS, and both placements are correct.

BumbleWeak-to-moderate

A useful mainstream product converging economically on the model it was built to differ from. Its safety investment is real and its free core still works. Its founding differentiation has been softened by its own product decisions, on its own published timeline.

TinderWeakest alignment

The strongest liquidity in the sector attached to the revenue model that benefits least from customers leaving. Not the worst product here — arguably the most valuable single asset in the comparison — and last on this axis regardless.

Alignment ranking

What this page argues

Hinge
Raya
Luxy
Bumble
Tinder

CFS ranking

What the published scores say

Raya +40
Hinge +24
Tinder −1
Bumble −2
Luxy −19

The two columns disagree twice, and the disagreements are the point. Raya leads on CFS and places second on alignment, because a superb extraction-to-value ratio is not the same thing as an architecture organised around the customer leaving. Luxy places last on CFS and third on alignment, because delivering a genuinely more authentic pool is directly relevant to the customer’s objective even when the surrounding commerce is the heaviest in the group. Tinder and Bumble are effectively tied on CFS and separated on alignment. If a single number could answer this page’s question, the page would not exist.

Five distinctions

What this comparison is careful not to collapse.

Each of these is a separate variable. Treating any two of them as one produces a simpler page and a worse one.

Scale ≠ alignment

Tinder can hold the largest and most valuable marketplace in the sector and still finish last on this axis. Liquidity is a genuine customer benefit. It is simply not the variable being measured.

Scarcity ≠ hostility

Raya’s restrictions largely support its proposition rather than manufacturing a shortage to sell relief from. Its consumption cap predates the paid workaround, and Artificial Scarcity was tested there and not supported.

High CHI ≠ low utility

Luxy carries the highest hostility score here and still delivers the group’s strongest differentiated benefit. A decade-long subscriber prefers it to the mainstream alternatives knowing everything on this page.

Good design ≠ clean hands

Hinge is the most aligned company in this comparison and monetizes attention aggressively. Alignment is not absolution, and a CHI of 57 is not a low number.

Differentiation decays

Bumble’s founding proposition has been softened by its own product decisions on its own timeline. A company can weaken its differentiation without any single decision being hostile.

None of the five companies here is condemned by having a business model, and none is exonerated by having a good feature. What separates them is narrower: whether the architecture would still make sense if every customer got what they came for.

The exit test.

CHI does not object to

Charging for a dating product
Subscription tiers
Selective admission
Small, curated pools
Verification and safety spending
Consumables sold openly

CHI asks about

What is withheld in order to be sold
Who is told an entitlement is “unlimited”
What a customer’s stated priority is worth
Whether position can be bought in an unpublished queue
Whether the product still works after success

If a customer names their first constraint, respect the constraint.

If a capability is capped, say the cap out loud where the product is sold.

If position is for sale, acknowledge that position is for sale.

If success means departure, build something that survives the departure.

Every other subscription profits from renewal.This one profits from the search.The best dating app, on this axis, is the one that works best when you stop opening it.

Evidence treatment

What this page refuses to say.

Every statement above is drawn from the five published CHI assessments. A comparison page is the easiest place in an index to lose precision, because compression rewards the strong sentence over the accurate one. These are the compressions this page declined to make.

Fact documented company behaviour, disclosure or published figure Reported journalism, trade reporting or customer accounts Firsthand directly observed by a CHI investigator, labelled as such CHI finding CHI’s own reasoning about those facts Not established tested and unsupported — never a synonym for “does not exist”

Refusals carried forward from the assessments

Not establishedThat any company here wants its customers to stay single. It is not claimed for Hinge, where the proposition was tested and rejected; it is not claimed for Tinder, where deliberate suppression of subscribers and fabricated Likes were both tested and not established; and it is not claimed for Bumble, where “deliberately preventing successful matches” was refused on the ground that financial data alone cannot establish intent. The supportable statement about the sector is structural: continued uncertainty, low visibility and ongoing participation are highly monetizable product states. That requires no intent and none is asserted.
Not establishedThat Luxy has the lowest fake-profile rate in dating. No authoritative comparative measurement supports a market-wide ranking. What is supported is narrower and still meaningful: a clearly labelled, long-term firsthand experience of markedly fewer obviously fake profiles than Tinder or Bumble. This page makes the narrower claim and refuses the wider one.
Not establishedThat Tinder is the worst product in this comparison. It is not said anywhere above, because the evidence does not support it and the page is not measuring it. Tinder holds the strongest liquidity in the sector, real verification investment and a usable free core. It finishes last on one axis: alignment with the customer’s eventual exit.
Not establishedThat Raya manufactures its scarcity in order to sell the exception. Artificial Scarcity was tested at Raya and not supported: the consumption limit predates the paid workarounds and reflects a stated anti-infinite-scroll position. Capping or refusing applicants is the product working as designed and scores nothing on any CHI page.
Not establishedThat Bumble’s women-message-first model has been abolished. The accurate statement is that it was softened by Opening Moves in 2024 and that a more fundamental reversal was announced for late 2026 but had not been broadly deployed as at the assessment’s research cut-off. Decay is recorded; abolition is not asserted.
Not establishedThat paid tiers were promised to be advertising-free. No first-party ad-free promise for any Luxy tier was found and none is asserted. The advertising finding rests on Luxy’s own media kit and on labelled firsthand observation, not on a broken promise.
CHI findingThat a high CHI is a verdict on product quality. It is not, and the ranking on this page deliberately demonstrates why: Luxy holds the highest hostility score here and places third on alignment, while Tinder holds a mid-table CHI and places last.
CHI findingThat aligned mechanics cancel monetization. They do not. Hinge ships the strongest exit-oriented mechanics in this sector and scores 18/25 on Revenue Extraction. This page credits the mechanics and does not use them as an eraser.
FirsthandThat one customer’s experience is a company-wide fact. The Luxy field-test result, the authenticity observation, the reliability observations and the support-response pattern are all recorded at the level at which they were evidenced — one long-term paying subscriber, in named cities, over stated periods. None is written here as a company policy, a global inventory measurement, or a statement about any other customer.

Scores are those published by each company’s own CHI assessment as at 23 August 2026. Luxy: CHI 76, CVI 57 and CFS −19 are the figures its assessment publishes following the close of the ten-day BLACK field study; the earlier provisional 70 / 60 / −10 is superseded and appears on that page only as recorded history. Dimension rubrics differ across this cohort, and the dimension rows in the matrix above should not be read across without reconciliation: Bumble’s CHI scorecard uses an earlier five-by-twenty rubric, Tinder’s assessment is expressly published as a working v2.0 draft, and Hinge, Raya and Luxy carry full v2.0 splits. The headline totals are comparable; a direct 70-versus-68 dimension reading is coarser than the numbers make it look. No company is re-scored on this page, no new score is assigned, and nothing in the alignment ranking should be read as adjusting any published figure. Pricing, tier composition and feature availability change frequently in this sector and should be reconfirmed before republication.

CHI AssessmentHinge → CHI AssessmentRaya → CHI AssessmentLuxy → CHI AssessmentBumble → CHI AssessmentTinder → Company ComparisonLuxury → Company ComparisonDropbox vs. Zoom → All CHI editorialArticles →