# Dating Apps Are for Access, Not for Dating
> The apps are measurably good at one thing and measurably bad at several others. Here is what the survey data and the psychology research actually establish.
Written by Ahmet Abiç (Founding editor). Reviewed by Noah Cortez (Stanford Journalism Program). Published August 18, 2026.
Source: https://gritwright.com/social/dating-apps/
---
import Figure from '../../../components/Figure.astro';
import ResearchNote from '../../../components/ResearchNote.astro';
import PressNote from '../../../components/PressNote.astro';
import PracticeNote from '../../../components/PracticeNote.astro';
import appsPhone from '../../../assets/guides/social/apps-phone.webp';
import appsSwipe from '../../../assets/guides/social/apps-swipe.webp';
import appsMeeting from '../../../assets/guides/social/apps-meeting.webp';
Search for whether dating apps work and you get pages of statistics with no visible source, mostly published by companies that sell dating advice. The numbers move between sites, gain a decimal place, and lose their origin.
The real evidence exists and it is better than the summary you will have picked up. It is also less flattering to the apps in one specific place and more flattering in another, and the difference between those two places is the whole practical point.
## The short answer
Apps do one job extremely well and several other jobs badly, and most frustration comes from expecting the second set.
| The claim | What holds up |
| --- | --- |
| What they are good at | Access to people you would never otherwise meet |
| What they claim and cannot show | That an algorithm can match you on compatibility |
| What the profile cannot carry | Anything that only exists when two people are in a room |
| The measured mistake | Messaging for weeks before meeting |
| What the experience looks like for men | Structurally message-poor, and that is documented |
| The right use | Get to a meeting quickly, then stop using the app for a while |
## How much of dating actually happens here now
Before arguing about whether apps are good, it is worth establishing how central they have become, because [the data here is unambiguous](https://web.stanford.edu/~mrosenfe/Rosenfeld_et_al_Disintermediating_Friends.pdf) and the shift is genuinely historic.
Using a nationally representative 2017 survey of American adults, the authors found that for heterosexual couples in the US, **meeting online has become the most popular way couples meet, eclipsing meeting through friends for the first time around 2013.**
About **39% of heterosexual couples who met in 2017 met online.** For same-sex couples, who adopted these services much earlier, the figure was about **65%.**
Their broader finding is about what got displaced: **internet meeting is displacing the roles that family and friends once played in bringing couples together.** Among couples who meet online, the share who met through the mediation of a third person has declined over time.
From the end of the Second World War until 2013, if you met a partner, the most likely route was somebody you both knew. That is no longer true, and it stopped being true recently enough that most advice about dating was written when it was still the case.
[Pew's survey work](https://www.pewresearch.org/internet/2023/02/02/key-findings-about-online-dating-in-the-u-s/) fills in the personal side of that. Three in ten American adults have ever used a dating site or app, rising to 53% of adults under 30. Around one in ten partnered adults met their current partner on one, and among partnered adults under 30 that rises to one in five.
So the honest answer to whether they work is that they work at the scale you would expect of a tool that a third of the country has tried. Not a miracle, not a scam. A common route that produces a real share of real relationships.
## What they are good at, and what they are not
The definitive assessment of this is [a 64-page review in Psychological Science in the Public Interest](https://faculty.wcas.northwestern.edu/eli-finkel/documents/2012_FinkelEastwickKarneyReisSprecher_PSPI.pdf), which separated what apps offer into three services and assessed each on the evidence. It is the most useful framing available because it stops the argument being about whether apps are good or bad in general.
The review splits what dating sites sell into **access, communication and matching**, and reaches a different verdict on each.
On **access**: online dating offers unprecedented and remarkably convenient levels of access to potential partners, which is especially helpful for people who would otherwise lack it. This is the clear win.
Also on access, the cost: encountering people through profiles **reduces three-dimensional people to two-dimensional displays of information**, which fail to capture the experiential aspects of social interaction that are essential to evaluating compatibility. A large pool can also elicit **an evaluative, assessment-oriented mindset** that leads users to objectify potential partners, undermine willingness to commit, and make lazy, ill-advised decisions.
On **matching**: **no compelling evidence supports matching sites' claims that mathematical algorithms work**, in the sense of producing romantic outcomes superior to other ways of pairing people.
Read the access finding twice, because it contains both the reason to use these things and the reason they feel bad.
The access is real. If you have left education, work somewhere without many peers, moved city, or simply exhausted your existing network, there is no offline mechanism that puts several hundred available people in front of you. That is not a small thing and no amount of complaining about apps replaces it.
The assessment mindset is also real, and it is the thing people are describing when they say apps have made them worse at dating. A profile invites you to evaluate. Evaluating hundreds of people in a row trains a habit of evaluation, and that habit then arrives at the table with you.
## The algorithm question, settled
The single most commercially important claim these companies make is that their system knows who suits you. It is worth being precise about what the evidence says, because it is not a matter of opinion.
The review's conclusion above is one half. The other half comes from [the speed-dating work](https://pauleastwick.com/s/JoelEastwickFinkel2017PSci.pdf) covered in [what actually happens on a first date](/social/first-dates/), where researchers took more than a hundred self-report measures from each participant and used machine learning to try to predict attraction. They could partly predict how much a person tends to like others, and how much others tend to like them. Whether two specific people would want each other was not predictable from any combination of traits and preferences reported before they met.
Those two findings fit together into one statement, and it is the most useful sentence in this article. An algorithm can sort a pool. It cannot tell you who you will click with, because that information does not exist yet at the moment of sorting.
Pew asked the public about this and found appropriate scepticism, which is unusual. Only 21% of American adults think dating site programs could determine whether two people will eventually fall in love. Around 35% say they could not and 43% are unsure.
Which reframes what a good app is. Not one with a clever algorithm, because there is no such thing. One with enough people in your city and a low enough friction to a meeting.
## What the experience is actually like, and why yours differs
This is the part missing from almost every article on this subject, and it matters more to a man in his twenties than anything above.
Among people who had used a dating app in the past year, the reported experience splits sharply.
**64% of men said they had felt insecure because of the lack of messages they received.** Among women, that figure was four in ten.
Running the other way: **54% of women said they had felt overwhelmed by the number of messages they received**, against a quarter of men.
Overall experience is more mixed than the discourse suggests: **53% of users describe their personal experiences as positive against 46% negative**, with men more positive than women, at 57% against 48%.
On motive, **44% of recent users say a major reason was to meet a long-term partner** and 40% say casual dating. Casual sex was a major reason for 31% of men against 13% of women.
Those two numbers, 64% of men insecure at the silence and 54% of women overwhelmed by the noise, are the same market described from opposite sides. This is worth internalising properly.
The scarcity you experience is a property of the pool, not a verdict on you. Nobody is choosing to ignore you personally. The person on the other side is triaging a volume of contact that would be unmanageable for anyone, and the mechanism that produces your silence also produces their exhaustion.
Two things follow practically. The first is that optimising your profile has a ceiling that is much lower than the advice industry implies, because you are not losing to a better bio, you are competing for finite attention. The second is that the emotional cost of app use, for men, arrives mostly as accumulated non-response, and that is a specific thing that can be managed by using apps in short bursts rather than continuously.
It is also worth naming that the same survey records substantial harassment running the other way. Among women under 50 who have used apps, 56% have received an unsolicited sexually explicit message or image, and 43% have had someone keep contacting them after being told no. If you have wondered why an early message gets a wary reply, that is the context it lands in.
## The same picture, from a different country
Almost everything written about apps in English is about the American market. [A German study](https://journal-bmp.de/wp-content/uploads/05_Aretz_final.pdf) gives a look at the same behaviour in a different population, and the shape holds.
An online survey of **N = 817 participants** examined who uses Tinder in Germany and how. The user base skewed young: mean age **24.06**, with **69% of users between 20 and 26.**
Usage was lighter than the discourse suggests: participants used the app for **around seven minutes a day on average**, and had been using it for about three months.
Aretz found that Tinder serves several different needs at once, some internal, such as passing time, amusement and self-affirmation, and some interactional, such as communication, flirting and sex. Which needs dominate **differs by gender and by relationship status**, and the paper analyses the over-choice effect as a central mechanism.
Seven minutes a day is the number worth keeping. The mental model most people carry is of endless swiping, and the measured behaviour of a large sample was a short daily check. If yours is much longer than that, the app has become something other than a tool for meeting people, which is a separate problem with a simpler solution.
## The one rule the research actually supports
If you take one behavioural instruction from all of this, it is about timing, and it is the opposite of what most people do.
The review found that [a short period of messaging before meeting can help, and that longer periods can actively hurt](https://faculty.wcas.northwestern.edu/eli-finkel/documents/2012_FinkelEastwickKarneyReisSprecher_PSPI.pdf). The mechanism is that people overinterpret the limited social cues available in text. Absent tone, face and body, we fill the gaps generously, build a version of the person out of the fill, and then meet somebody else. The review calls the result an unpleasant expectancy violation.
In practice: exchange enough to establish that a person is real, pleasant and interested, then propose something. A few days of messages, not three weeks. This also solves the more mundane problem that long text threads run out of energy on their own, and a thread that dies before a meeting produces nothing at all.
The corollary is that the messaging is not the relationship and cannot be made into one. What you are optimising for is arriving at the point covered in [the first date guide](/social/first-dates/), where the only measured lever is asking follow-up questions in person.
## The fatigue is not only in your head
If the apps feel worse than they did, that impression is [shared widely enough to be visible in the industry's own accounts](https://news.medill.northwestern.edu/chicago/the-bumble-burnout-as-dating-app-enthusiasm-dims-industry-scrambles-to-prove-digital-matching-can-still-deliver-meaningful-connections/).
Reporting from Northwestern's journalism school documents dating app fatigue **showing up in the stock market rather than only in user complaints.**
Bumble's share price fell **91% from its 2021 peak**, and in June the company **laid off 30% of its workforce**, a cut expected to save around $40 million a year.
The piece frames the open question as whether the apps can still deliver on the original promise of meaningful connection, at a point where users are burned out and paywalls have grown.
This is not evidence that apps do not work, and it should not be read that way. It is evidence that a lot of people found the experience wearing enough to stop paying, which is a stronger signal than a survey because it costs something.
The practical reading is about dosage rather than abstinence. A tool that gives access has a use. A tool you check compulsively while feeling progressively worse about yourself has stopped being that tool.
## Treating it as a data problem, honestly
Since the apps invite an engineering mindset, it is worth looking at [the most rigorous documented attempt to apply one](https://www.ted.com/talks/amy_webb_how_i_hacked_online_dating), including the part where it did not go as planned.
After her own online dating went badly, Webb built a system: **72 data points** describing what she wanted, sorted into tiers and weighted, with a **scoring threshold of 700** below which she would not go on a date.
She then built test profiles to research how the most successful users of the sites presented themselves, and rewrote her own on what she found. She documented the entire method publicly, in the talk and in a subsequent book.
Her own account of the outcome is the useful part, and she gives it plainly: her conclusion was that the algorithms were not failing, they were doing exactly what they were built to do, which is match user-generated information with other user-generated information.
That is practice rather than research, and it is one person. What makes it worth including is the conclusion, which is the same one the peer-reviewed review reached from the other direction. The system is not broken. It is matching descriptions to descriptions, and descriptions are not where compatibility lives.
The transferable half of Webb's method is the scoring, not the profile engineering. Writing down what actually matters to you, in advance, in a form specific enough to check, is a defence against the assessment mindset the review warned about. Without it you drift into judging people against whoever you saw last.
## What a profile can and cannot do
The advice industry treats the profile as the lever, because it is the only part of this that can be sold as a service. The review's finding about what a profile fundamentally is puts a ceiling on that.
A profile is a two-dimensional display of information standing in for a three-dimensional person, and the things people actually decide on are the experiential ones: how somebody talks, what they find funny, whether being near them is easy. None of that is representable in the format. This is not a criticism of anyone's photos. It is a property of the medium.
So the honest division is between the things that plainly cost you and the things that are guesswork dressed as science.
**What plainly costs you** is a profile that fails at its one job, which is establishing that you are a real, specific person worth twenty minutes. Photos where nobody can tell which one you are. A blank bio, which reads as either low effort or a placeholder. Six group shots. Nothing that indicates what you actually do with your time.
**What is guesswork** is nearly everything else. The optimal number of photos, the psychology of the first line, whether to smile, which colour shirt. You will find confident numbers for all of it and none of them trace back to anything you can check. I have looked, and if a good study exists on photo selection I did not find one I could open and verify.
The useful reframing is that a profile is not a pitch, it is a filter. Its job is to be specific enough that people who would like you can recognise it, which also means being specific enough that people who would not can pass. A profile engineered for maximum appeal is one that has removed everything a person could recognise, which is why generically optimised profiles produce matches that go nowhere.
Say what you actually do, name something real rather than a category, and put in one thing that would be slightly embarrassing to defend at a party. The last one does more work than the other two.
## The first meeting, and an asymmetry worth understanding
Getting to the meeting quickly is the instruction the evidence supports. How you propose it runs into something the Pew data documents clearly, and which is easy to be oblivious to from the male side.
Among **women under 50 who have used dating apps, 56% have been sent a sexually explicit message or image they did not ask for.** About **43% have had someone keep contacting them after they said they were not interested**, 37% have been called an offensive name, and **11% have received threats of physical harm.**
Across all users of any gender the figures are lower but not small: 38% unsolicited explicit content, 30% continued unwanted contact, 24% an offensive name, 6% a threat of physical harm.
The share of American adults who say online dating is a generally safe way to meet people has **slipped from 53% to 48% since 2019**, and around **six in ten say companies should require background checks** before a profile can be created.
Reading that changes how an ordinary message looks from the other end. A stranger proposing to meet is not, from where they are sitting, a neutral event, and a wary or slow reply is usually not about you.
The practical consequences are small and they cost you nothing. Propose somewhere public and ordinary at a normal hour. Make the plan easy to leave, which as [the first date guide](/social/first-dates/) argues is better for you too. Do not push when a suggestion is deflected. Do not ask for a phone number as a condition of meeting. If somebody wants to tell a friend where they are going and share the venue, that is sensible behaviour and not an insult.
The same applies in your own direction, incidentally. Meeting a stranger from the internet carries a non-zero risk for anyone, telling somebody where you are going costs one message, and men are the group least likely to do it.
## What to actually do
Everything above collapses into a short and slightly boring set of instructions.
**Use them for access and nothing else.** They are a directory of people who are available. That is the product. Every other promise is unsupported.
**Set the criteria before you open the app**, not while scrolling. Four or five things that genuinely matter, written down. The pool will otherwise set them for you.
**Move to a meeting within days.** The evidence points the same way every time: the useful information starts when you are in a room.
**Run it in bursts.** A week on, deliberately, with the intention of arranging two meetings, beats a permanent low-level presence that produces mostly accumulated silence.
**Stop reading the non-response as data about you.** Two thirds of men on these platforms feel what you feel. That is a ratio, and ratios are not personal.
**Keep at least one offline route open.** The Stanford data shows friends declining as a route, not disappearing, and [meeting people as an adult](/social/how-to-make-friends-as-an-adult/) is the same skill set that produces introductions.
## What this does not fix
Apps solve the access problem and leave every other problem exactly where it was.
They do not make you better at the meeting itself, and the meeting is where things are actually decided.
They do not shorten the process. The base rate of pleasant, inconclusive evenings is a property of meeting strangers, not of the medium that introduced you.
And they do not help with the harder questions further along, like whether a thing that has started is worth continuing, which no amount of matching can answer in advance and which gets its own guide.
The apps are a good answer to one question and no answer at all to the rest. Used that way they are genuinely useful. Used as a place to spend time, they are the assessment mindset with a subscription attached.