The only long series that exists
One American survey has followed the question since the 1940s by asking couples how they met: How Couples Meet and Stay Together, analysed by Michael Rosenfeld and colleagues. It is the only source that shows a curve rather than a snapshot, and its conclusion is unambiguous.
| Before 1995 | Meeting online is at zero. Friends dominate, with about a third of couples; then bars and restaurants, work, study, family, the neighbourhood and church. |
|---|---|
| Around 2009 | The internet reaches roughly 22% of heterosexual couples formed that year. |
| About 2013 | The internet overtakes friends and becomes the leading channel. |
| 2017 | Close to four heterosexual couples in ten met online. Among same-sex couples the share is far higher, and has been the majority route for longer, for want of a large enough local pool. |
The title of the 2019 paper sums up the mechanism: disintermediating your friends. The internet did not merely add a channel, it took away the friends' role as intermediaries. The share of couples formed through friends fell from about a third in the mid-1990s to roughly one in five twenty years later.
What the European surveys say
There is no European equivalent of that long series; what exists is national snapshots, and they are more cautious.
- France. The Épic survey by Ined and Insee, covering 7,825 people aged 26 to 65 in 2013-2014, measured the use of dating sites for the first time: 12% of women and 16% of men had signed up, but their part in actually forming couples was still modest at that date. More recent polls put the internet and apps at around 16% of meetings, behind friends and work — so the gap with the United States is real.
- German-speaking countries. The surveys contradict each other on first place: the ElitePartner study has digital first contact ahead of the circle of friends, others still put friends first. The detail is on Partnersuche im deutschsprachigen Raum.
Why the figures do not line up
Three reasons, worth knowing before quoting any percentage.
- The question differs. "Where did you meet your current partner?" and "how did this year's couples form?" do not give the same answer: the first includes couples formed thirty years ago and mechanically understates the recent channels.
- Who paid for the survey. A good share of the figures in circulation come from dating platforms. Not necessarily false, but interested, and the sample is often their own members.
- Mixed cases are counted badly. You cross paths at a party, find each other on an app, then talk again through a mutual friend: the respondent has to pick one answer, and picks differently depending on age.
What has not changed
The channel changed; almost nothing else did.
- Homogamy holds. People still partner with someone of similar education, background and origin, online as much as offline. Apps widened the geographical pool far more than the social one.
- Proximity still decides. Even on an app, the default search radius does the sorting. "Chance" has always had a radius of a few kilometres.
- The declining channels were declining long before the internet. Family, the neighbourhood and the parish have fallen steadily since the war. The internet did not start that movement, it accelerated it.
- What follows the meeting is identical. No study has established that a couple formed online lasts less or gets on worse than any other. The channel decides the meeting, not the relationship.
The practical point
If you are looking for someone today, all of this comes down to one sentence: the majority channel has changed, but no channel has disappeared. The people who meet the most use both — see making online dating work and meeting in real life.