Dating App Economics — Engineered Dissatisfaction

To understand the dating-app economy, begin with the cash flows. Match Group, parent company of Tinder, Hinge, OkCupid, Plenty of Fish, Match.com and The League, posted $3.43 billion in full-year 2025 revenue and operating cash flow of $1.08 billion [1]. Bumble Inc. reported $965.7 million [2]. Together with smaller competitors, the industry now monetises romantic attention at a scale comparable to professional sport — and through a subscription architecture that depends not on users finding partners, but on users continuing to look.

Tinder, the flagship asset, generated $1.9 billion in direct revenue in 2025 [1]. Yet the headline number conceals a structural tension. Tinder's paying-user count fell to 8.8 million in Q4 2025, an 8% year-over-year decline and the sixth consecutive quarter of contraction since the Q2 2023 peak of 10.9 million [1]. Revenue held up only because revenue-per-payer rose to $17.63 — the platform extracted more from each remaining subscriber even as the base shrank. The growth lever is no longer acquisition. It is monetisation density.

Hinge tells the inverse story. Q4 2025 direct revenue rose 26% year-over-year to $186 million; paying users grew 17% to 1.9 million; revenue-per-payer reached $32.96 — nearly double Tinder's [1]. Match Group has set a public target of $1 billion in annual Hinge revenue by 2027 [1]. The brand markets itself with the slogan Designed to be Deleted — a tagline whose mathematical viability inside a $1 billion ARR target is the central paradox of this report.

Bumble's 2025 trajectory is grimmer. Total revenue fell 10% to $965.7 million [2]. Paying users dropped 11.5% across the year and 20.5% by Q4 alone, ending at 3.3 million [2]. The company recorded a $1.039 billion non-cash impairment charge — an accounting acknowledgement that the business it acquired no longer commands the future cash flows previously assumed — producing a GAAP net loss of $906.6 million [2]. Adjusted EBITDA margin nonetheless reached 32.5%, confirming that even a contracting dating platform remains highly cash-generative for as long as recurring subscriptions hold.

The aggregate picture is a market that has matured, plateaued, and is now consolidating around price increases rather than new audiences. AppsFlyer mobile-attribution data show 65% of dating apps installed in 2024 were deleted within thirty days; the figure climbed to 69% in 2025 [11]. The combination of falling acquisition, rising churn and rising per-user revenue is the financial signature of a maturity stage in which firms harvest the existing base rather than expand it.

None of these numbers, taken individually, prove that the industry is engineered against its users. What they establish is the financial geometry within which every product decision is made. A subscription business with shrinking acquisition and rising churn is structurally incentivised to extract more from each retained user, to delay successful exits, and to convert dissatisfaction into upgrade opportunities. The marketing language of connection coexists with a capital-allocation logic that rewards friction.

The industry's defenders argue that revenue from love is no different from revenue from any other matched two-sided market. The counter-argument, which the remainder of this report will document, is that the matching produces measurable harm to the very users whose subscriptions sustain the cash flow — and that the harm is foreseeable, observable in internal documents, and inseparable from the design choices that generate the cash flow in the first place.

The first mechanism is desirability ranking. From at least 2014 until publicly disavowed in March 2019, Tinder operated an internal scoring system modelled on the Elo chess-rating algorithm [12]. Each swipe functioned as a chess match: a right-swipe from a high-ranked profile lifted the recipient's score; a left-swipe lowered it; visibility in the deck was conditioned on the score. Tinder confirmed the system to FastCompany in 2016 and abandoned the Elo terminology in 2019 — but the underlying mechanism, a ranked recommender that surfaces high-engagement profiles disproportionately, remains the only economically rational design for a swipe-based marketplace [12].

The behavioural consequence is documented. SwipeStats analysed 7,079 Tinder profiles in 2025 and found that women receive matches at 8.4x the rate of men: 44.4% versus 5.3% [12]. The asymmetry is not random. It is the predictable output of a ranked system in which a 76-78% male user base allocates limited swipe-budgets across a 22-24% female user base [12]. The matchmaking marketplace is mathematically two-sided, but only one side experiences abundance.

The second mechanism is the upgrade ladder. Tinder operates four paid tiers — Plus at $24.99 per month, Gold at $39.99, Platinum at $49.99, and the invite-only Select tier at $499 per month — alongside à-la-carte purchases of Super Likes and Boosts [13]. Each tier sells a partial relaxation of artificial scarcity: more visibility, more incoming likes shown, more daily swipes, ability to message before matching. The architecture is the architecture of any well-designed paywall — but the commodity gated is not access to information. It is access to other people's attention.

The third mechanism is personalised pricing. A 2022 investigation by the Mozilla Foundation and Consumers International found Tinder Plus prices varied by up to fivefold within the same country, driven by demographic signals including age, gender, location and inferred willingness to pay [10]. In 2019, Tinder paid $23 million to settle a California age-discrimination class action that demonstrated users over 28 were charged roughly twice the rate of users under 28 for the same product [10]. The settlement did not change the underlying practice. Court filings since suggest demographic-driven price differentiation continues, with age now subordinated to other inferred willingness-to-pay variables [10].

Match Group describes its tiers as offering different feature sets to different audiences [13]. The Mozilla investigation describes them as algorithmic discrimination engineered to extract maximum surplus from each user category [10]. The functional gap between these descriptions is the gap between marketing copy and revenue strategy. A subscription business optimising lifetime value on a personalised-pricing engine produces, by mathematical inevitability, two prices for the same hour of user attention.

Behind these three mechanisms sits a fourth — the loop. The product surfaces a profile, the user evaluates, the user swipes, the system updates, the next profile appears. Industry product literature refers to this loop as the "core engagement primitive." Behavioural-design literature refers to it as a variable-ratio reinforcement schedule, the same reinforcement architecture used in slot machines [3]. Its appearance in dating-app documentation is the proximate cause of the federal lawsuit examined in the next section.

The architecture is not concealed. It is documented in product roadmaps, investor presentations and patent filings. What is concealed is the framing: that a system optimised for engagement is mathematically optimised for not-yet-finding, and that the not-yet-finding is what the user pays for.

The case is Oksayan et al. v. Match Group, Inc., filed 14 February 2024 in the US District Court for the Northern District of California, assigned to Magistrate Judge Laurel Beeler [3]. Six named plaintiffs from California, New York, Georgia and Florida — all paying or former paying subscribers — assert claims under consumer-protection and product-liability statutes, breaches of warranties, and deceptive trade practices. The complaint runs to 121 pages [3]. Its core empirical assertion is documentary, not theoretical.

The complaint alleges that Match Group "employs recognised dopamine-manipulating product features to gamify the Platforms to transform users into gamblers locked in a search for psychological rewards that Match makes elusive on purpose" [3]. The mechanisms identified in the filing are specific. First, a content-presentation format that "gamified romance" by introducing intermittent variable rewards — the same reinforcement schedule found in slot machines, where the unpredictability of the reward is the primary driver of compulsive engagement [3]. Second, a push-notification system the complaint characterises as "preying on users' fear of missing out" with strategic timing designed to recapture attention at moments of high vulnerability [3]. Third, an incentive-reward structure that "punishes users from disengaging and rewards compulsive users" [3].

The plaintiffs' theory of liability is that Match Group's marketing — including Hinge's "Designed to be Deleted" tagline and Tinder's positioning as a relationship-formation tool — is materially false because the architecture of the products is incompatible with the marketing claim. A product engineered to maximise dopamine-driven retention cannot simultaneously be a product engineered for users to delete. The complaint does not ask the court to ban the products. It asks the court to require warning labels analogous to those required on regulated gambling, addictive consumer products, and tobacco [3].

The legal theory is novel. The factual basis is not. Federal Trade Commission proceedings establish parallel evidence: in August 2025 the FTC secured a $14 million settlement against Match Group for deceptive guarantees, retaliatory account suspensions of users who filed billing disputes, and obstructive cancellation flows [4]. The FTC found Match.com had deceptively induced subscriptions through a "free six-month if you don't meet someone special" guarantee bound by undisclosed onerous conditions, and had unfairly suspended accounts of users who challenged charges [4]. The FTC settlement does not adjudicate the gamification claims in Oksayan. It does establish, on a contested administrative record, that the parent company has been formally found to have used deceptive subscription mechanics — providing prima facie evidentiary weight for the broader allegations.

Match Group has moved to compel arbitration on the basis of mandatory clauses in its terms of service [3]. The motion remains pending. Even if granted, the arbitration record will likely be sealed. What the complaint has already accomplished, however, is the public disclosure of internal-product framing — the language of "intermittent variable rewards" and "incentive rewards" comes directly from Match Group's product-management vocabulary, surfaced in court filings rather than corporate communications [3].

The pattern echoes prior platform disclosures. The 2021 Wall Street Journal Facebook Files surfaced internal Meta research demonstrating known psychological harm to teen Instagram users; the 2024 NPR disclosure of redacted TikTok internal documents surfaced an explicit habit-formation threshold of 260 videos. Oksayan is the dating-app entry in the same documentary genre — internal product logic, ordinarily hidden, made part of the public record by the friction of litigation [3].

The defence argument — that all engagement-driven products use behavioural design and that users consent through their continued use — is the same argument made by social-media platforms, mobile-game publishers and regulated-gambling operators. The legal question is whether the dating-app context, in which the product purports to deliver a specific outcome (a partner) and not merely entertainment, alters the materiality calculus for false-advertising claims. That question has not yet been answered.

The first observable harm is exhaustion. The Forbes Health/OnePoll survey, conducted between 27 March and 1 April 2024 with 1,000 US dating-app users, found 78% reported feeling emotional fatigue from dating apps "sometimes, often, or always." Among Millennials the figure rose to 80%; among Gen Z to 79%. Women reported higher fatigue than men (80% vs 74%) [6]. The top stated cause was "the inability to find a good connection" (40%), followed by rejection (27%) and "repetitive conversations while chatting with multiple matches" (24%) [6].

A peer-reviewed longitudinal study published in New Media & Society in 2024 — Sharabi, Von Feldt and Ha — moved the question beyond cross-sectional self-report. Tracking users over time, the authors found that emotional exhaustion and a sense of inefficacy increase the longer a user remains on the platform [9]. Critically, the predictors of burnout are not solely pre-existing distress (depression, anxiety, loneliness, all of which independently predicted higher burnout) but also problematic dating-app use itself — meaning the platform reliably converts even baseline users into exhausted users with sufficient exposure time [9].

The harassment data are particularly important because they are gender-asymmetric in the same direction as the structural match-rate asymmetry. The platforms produce a market in which one side experiences attention abundance — including substantial unwanted attention — while the other side experiences attention scarcity. The exit response is the same in both cases: deletion. AppsFlyer data showing 69% of dating apps deleted within thirty days describe a population unable to remain on the platform [11] — but who, the FTC settlement and class-action complaint argue, are then re-engaged through retention pricing, dark-pattern cancellation flows and re-acquisition advertising [4][3].

The third user-side data series is broader and more contested. Institute for Family Studies analysis of 2024 General Social Survey data documents what has become known as the "sex recession." The share of Americans aged 18-29 who report not having sex in the past year doubled from 12% in 2010 to 24% in 2024 [8]. The male sub-population shows a steeper trajectory — sexlessness rose from 9% in 2013-15 to roughly 24%, nearly tripling in nine years. The share of all 18-64 adults having weekly sex fell from 55% in 1990 to 37% in 2024 [8]. Pew Research independently documents that 25% of US 40-year-olds in 2023 had never been married, up from 20% in 2010 [15].

The dating-app industry's role in these trends is genuinely contested — Section 7 examines the dispute in detail. What is not contested is the temporal coincidence: the 14-year doubling of youth sexlessness occurred over the period in which dating apps moved from niche to dominant. Stanford's How Couples Meet and Stay Together data show that by 2017, 39% of US heterosexual couples and 65% of same-sex couples that formed had met online — overtaking introductions through friends, family, school and work for the first time in modern social history [7].

The displacement matters. Pew's January 2025 analysis of friend-time data documents that average weekly time spent with friends among young US adults fell from 12.8 hours in 2010 to 6.5 hours in 2019 — a roughly 50% collapse [15]. The decline in friend-mediated introduction is not balanced by app-mediated introduction at the population level: more couples meet through apps now, but fewer couples form overall, and the user pool reports rising fatigue and burnout. The system is producing matched pairs, but at lower throughput than the system it replaced.

None of this proves that apps cause the broader trends. What it establishes is that the same population using the apps reports measurable distress, that the distress increases with use, and that the displacement of prior introduction pathways is documented. The evidence is consistent with a dating-app industry that monetises a frustration it does not cause but perpetuates.

The United States is the dating-app industry's home market and its analytic baseline. By 2017, online introduction had become the dominant pathway for new heterosexual couples; by 2024, Match Group's Tinder, Hinge, Match.com and Bumble together accounted for the overwhelming majority of US dating-app subscriptions [1][2]. The 2024 Forbes Health survey, the 2023 Pew Research harassment study, the IFS sexlessness data and the Stanford HCMST series describe a market in which the apps have achieved near-saturation among the under-40 population while delivering measurable distress to that same population [6][5][8][7].

The United Kingdom shows analogous penetration with a sharper recent contraction. AppsFlyer data tracked between January 2024 and January 2025 record Tinder losing 594,000 UK users, Bumble 368,000 and Hinge 131,000 over the period [11]. In total, paying-user counts in the UK declined faster than the US, and faster than Match Group's global average [1]. The UK Online Safety Act 2023, which came into force progressively through 2024-2025, applies general user-to-user platform duties of care to dating apps in respect of harmful content, including the obligation to address unwanted sexual content — the harassment surface documented by Pew Research in the US [5].

France approaches the industry through general consumer-protection law. The Direction générale de la concurrence, de la consommation et de la répression des fraudes (DGCCRF) has applied unfair-commercial-practices rules to subscription auto-renewal, ambiguous cancellation flows and personalised pricing — the same enforcement vector used against Match Group in the US by the FTC [4]. The EU Digital Services Act, in force since February 2024, imposes additional transparency obligations on recommender systems, including those used in dating apps once their EU user base crosses the very-large-online-platform threshold. Tinder has not yet been formally designated a VLOP, but contestation is ongoing.

Japan and South Korea present a different category. South Korea's total fertility rate fell to 0.78 in 2023 — the lowest figure ever recorded for any nation in modern demographic statistics [14]. Japan's stood at 1.20. The 2024 survey by Japan's Children and Families Agency found that 25% of married Japanese under 39 had met their spouse through a dating app [14]. The economic pressure to increase that share is now governmental. Tokyo Metropolitan Government has launched its own AI-assisted matchmaking platform; Kochi Prefecture subsidises commercial-app subscription fees for residents aged 20-39; Pairs (the Japanese dating-app market leader) launched in South Korea in 2024 specifically to address the cross-national fertility crisis [14].

The cross-country comparison reveals two distinct regulatory postures. Western jurisdictions — the US, EU, UK — treat dating apps as consumer-protection and platform-safety problems, applying enforcement against deceptive subscription mechanics, harassment-handling deficiencies and personalised-pricing discrimination. East Asian jurisdictions — Japan, South Korea — increasingly treat dating apps as demographic-policy instruments, subsidising or directly operating them to address fertility decline.

Both approaches share a common premise: that the existing commercial dating-app market is producing outcomes the state finds inadequate. The Western response treats the market as harmful and seeks to regulate the harm; the Eastern response treats the market as insufficient and seeks to extend it. Neither has yet produced an evaluation framework that distinguishes when apps cause demographic and psychological outcomes from when they merely correlate with them.

That evaluation gap is the regulatory frontier. The next two sections examine the regulatory tools currently being deployed and the empirical debate they will need to resolve.

The Federal Trade Commission's August 2025 enforcement action is the most consequential to date. Match Group, the operator of Match.com, OkCupid, Plenty of Fish and The League — branded properties separate from Tinder and Hinge — agreed to pay $14 million and to "permanently halt" the practices found deceptive [4]. The findings are specific: Match.com had induced subscriptions through a six-month free-extension guarantee bound by undisclosed onerous conditions; the company had unfairly suspended accounts of users who filed billing-dispute claims through their banks, retaining the disputed payments without delivering services; and the cancellation flow had been engineered to maximise dropout from the cancellation process [4].

The settlement does not adjudicate gamification or personalised pricing. It does establish, on a contested administrative record, that a major dating-app operator engaged in deceptive subscription mechanics across multiple branded products simultaneously — providing the FTC with prima facie grounds to scrutinise the broader product portfolio. The $14 million figure is itself negligible against $3.43 billion in annual Match Group revenue [1]. As a behaviour-changing instrument, the settlement's value lies less in the cash and more in the prohibition: any continuation of the practices found deceptive would expose the company to contempt and to enhanced civil penalties.

The European Union's Digital Services Act, in force since February 2024, imposes obligations on recommender-system transparency, notice-and-action mechanisms for harmful content, and dark-pattern prohibitions [4]. As applied to dating apps, the DSA's most consequential provision is Article 25, which prohibits the design of online interfaces that "deceive or manipulate" users in ways that impair autonomous decision-making — language whose application to swipe-based gamification has not yet been tested in enforcement, but whose textual reach plainly extends to it.

The structural limitation of these instruments is the click-through gap. Each addresses a specific harm — deceptive guarantees, hard-to-cancel subscriptions, dark-pattern interfaces — without confronting the upstream design principle that produces them. A subscription business optimising lifetime value of a churning user base will continually generate new dark patterns at the rate that old ones are prohibited. The regulatory machinery is reactive by construction. Each enforcement action restates the prohibition; the prohibition does not change the optimisation function.

The risk landscape combines high-severity harms (personalised pricing, gamification) against which existing instruments are weakest, with medium-severity harms (cancellation friction) against which existing instruments are strongest. The regulatory inversion is structural: it is easier to mandate a cancellation-button location than to mandate the redesign of an entire reinforcement schedule.

The most consequential development of 2025-2026 is therefore not enforcement but discovery. Oksayan v. Match Group, if it survives the motion to compel arbitration, will produce the dating-app industry's first court-ordered disclosure of internal product documentation [3]. The Facebook Files emerged from whistleblower disclosure rather than litigation; the dating-app industry has not yet had its Frances Haugen. The court system, rather than the regulator, may be the institutional pathway through which the architectural questions are eventually addressed.

Until the discovery record is public, regulatory analysis necessarily operates on the model that exists: enforcement against documented deceptive practices, with the engineered-frustration core of the business model formally untouched.

The case for causation rests on three pillars. First, temporal coincidence: the doubling of 18-29 sexlessness from 12% (2010) to 24% (2024) [8] maps onto the period in which dating apps moved from minority to majority introduction channel [7]. Second, mechanism: the ranked-recommender architecture documented in Section 2 produces measurable match-rate asymmetry [12] and the burnout architecture documented in Section 4 produces measurable exhaustion [9][6]. Third, the displacement evidence: Stanford HCMST shows apps have eaten the introduction share previously held by friends, family and workplace [7] — a substitution that, if the apps then deliver lower-throughput pairing, mathematically reduces aggregate relationship formation.

The case against causation rests on four counter-pillars. First, confounders dominate: housing costs, real-wage stagnation, post-pandemic social atrophy, rising self-reported anxiety, and the documented collapse of friend-time (12.8 hours per week to 6.5 hours, 2010-2019) [15] all independently predict relationship-formation decline. Second, dating apps still produce relationships at scale — 39% of 2017 heterosexual pairings and 65% of same-sex pairings [7] are not a population in which the technology has failed. Third, the sex-recession trend predates app dominance: the 1990 figure of 55% of adults having weekly sex [8] establishes a long-running decline that cannot be attributed to a 2012 technology. Fourth, country-level variation is inconsistent with simple causation: South Korea has the lowest fertility in the world but lower app-penetration than the US [14].

The methodological literature has now begun to take the question seriously. Sharabi et al. (2024) is the most rigorous longitudinal evidence currently in print: it establishes that prolonged dating-app use predicts increasing emotional exhaustion, controlling for baseline distress [9]. The study's effect sizes are modest but statistically robust. It does not estimate population-level causal contribution to the sex recession. No study currently in print does — the natural-experiment data required (a population with otherwise-equivalent characteristics that lacks dating-app access) is not available within the developed economies in which the recession is observed.

This hierarchy matters for policy. If apps are a passive mirror of broader trends, regulating them addresses a symptom; if apps are the sole cause, regulating them is decisive; if apps are a contributing accelerator, regulating them is necessary but not sufficient. The accelerator framing is the framing best supported by the current evidence, and it is the framing within which the policy debate has begun to converge — explicitly in the EU's recommender-system transparency obligations, implicitly in the FTC's pursuit of subscription deception [4].

The class-action discovery process, if it produces the document disclosure that whistleblower-driven processes produced for social media, will sharpen the answer. Until then, the policy posture must accommodate uncertainty. The next section synthesises what the existing evidence supports.

Take the four findings the evidence most clearly supports and assemble them. First, the dating-app industry generates billions of dollars in annual revenue from a subscription architecture in which retention, not partnership formation, is the optimisation target [1][2]. Second, the gamification architecture documented in court filings — intermittent variable rewards, push-notification timing, incentive rewards — is the same reinforcement architecture used in regulated gambling [3]. Third, the user population reports measurable distress: 78% emotional fatigue [6], 38% unwanted sexual content [5], increasing burnout over time [9], and 69% deletion within thirty days [11]. Fourth, the demographic outcomes that the apps were marketed to improve — relationship formation, partnership rates, sexual activity — have moved in the opposite direction over the period of app dominance [8][15].

The structural mismatch is not a moral failing of any specific platform. It is the inevitable output of placing a partnership-formation product inside a subscription-revenue chassis whose financial physics reward continued search. Match Group is not a uniquely cynical operator; it is responding rationally to capital-market expectations expressed in its share price. Bumble is not less aligned with users; it is the same model with worse 2025 execution. The investor-relations materials of both companies describe lifetime-value-per-user growth as the principal value-creation driver — a metric that, by definition, increases with longer engagement [1][2].

The cross-country comparison from Section 5 reinforces the analytic conclusion. Where the United States, the United Kingdom and the European Union treat the apps as a consumer-protection and platform-safety problem, Japan and South Korea treat them as a fertility-policy instrument [14]. Both responses concede, by their existence, that the unregulated commercial model produces inadequate outcomes — that the market alone, without state intervention either to constrain or to subsidise, does not deliver the outcomes participants seek.

What does this analytic posture imply? Three implications follow directly. First, regulatory effort focused on the worst-documented harms — personalised-pricing discrimination, dark-pattern cancellation, harassment-handling failure — is well-targeted and should continue. Second, regulatory effort focused on the gamification architecture is necessary but currently lacks the evidentiary base required for direct intervention; the discovery record from Oksayan may, but has not yet, supplied it [3]. Third, the demographic-policy framing in Japan and South Korea risks regulatory capture by exactly the operators whose architecture is in dispute — a state that depends on commercial dating-app outcomes for fertility metrics will be slow to constrain those operators [14].

The deeper question for users is not whether to use dating apps. It is whether to use them aware of the financial geometry within which the experience is delivered. The platforms are not neutral matching infrastructure; they are subscription businesses whose interests align with users only at the moment of acquisition. After that, the alignment inverts. A user who recognises this is in a stronger position to use the products as transactional infrastructure rather than relational projects — to time their use, limit their exposure and exit on their terms rather than the platform's.

The Attention Economy report in this series concluded that engagement-optimised social-media platforms generate measurable harm because their economic incentives are mismatched with user welfare. The dating-app industry is the same architecture applied to a different commodity. The commodity here is not attention. It is hope. The financial mechanics that convert one into recurring revenue are equally capable of converting the other. The evidentiary record now in print does not yet permit a final verdict on causation. It permits, with high confidence, a verdict on contradiction.

A platform engineered for retention cannot be a product engineered for departure. The marketing slogan and the income statement cannot both be true. The slogan is what the company tells the user. The income statement is what the company tells the market. The user pays for both.