How many couples meet online?

The current-partner answer starts with partnered adults

Pew Research Center asked U.S. adults in July 2022 how they met their current partner. Among adults who were married, living with a partner, or in a committed relationship, 10% said they met that person through a dating site or app. The denominator is currently partnered adults, not every adult and not every couple that began in 2022. Each respondent reports one current partnership. The estimate includes relationships that began long before the survey, including years before modern dating apps existed. That history matters when someone expects a larger share based on how people meet today.

The 10% also has a narrow definition of online meeting: a dating site or app. Someone who first connected through social media, a game, a work forum, or another online space may not fall into that category. A study that counts all online introductions can therefore report a different result without contradicting Pew's dating-platform figure. Before comparing two percentages, check both the relationship group and the kind of online connection each study counted. The ordinary phrase 'met online' can hide both differences.

Pew reported that 69% of surveyed U.S. adults were partnered under its definition. The 10% meeting-route estimate is calculated within that partnered group. It should not be introduced as 10% of all adults, because the survey first filtered to people with a current partner. Nor should it be called 10% of all relationships ever formed; relationships that ended before the survey are outside the current-partner question. This filter is the quiet part of the statistic that makes the headline number intelligible.

Younger and LGB partnered adults had higher shares

Within Pew's 2022 partnered-adult group, 20% of adults under 30 said they met their current partner on a dating site or app. Among partnered lesbian, gay, or bisexual adults, the share was 24%. Each is a within-group rate. These people are also included in the all-partnered 10%, so the three figures should not be added. The LGB and under-30 groups overlap with each other as well. The comparisons say that the reported meeting route was more common in these named groups, not that any route works better for a particular person.

A younger current-partner group may include more relationships that began after dating apps became widely used, while the all-age group also includes older relationships. That timing is a plausible reason the denominators matter; Pew's 10% and 20% alone do not isolate a causal explanation. A person in a long marriage still contributes to the all-partnered measure even if the service did not exist when they met. For a question about newly formed relationships, a survey of all current partners is an imperfect lens. Its strength is describing how the people partnered at the survey date said they met.

Current-partner samples change as relationships begin and end. A couple who met years ago stays in a current-partner survey while they remain together, so older routes to meeting can remain visible for a long time. A new-couple survey instead focuses on recent starts and gives older relationships no place in its denominator. Neither design is automatically more truthful; they answer different timing questions. The Pew figure is particularly useful when someone asks about the partners adults have now. It cannot by itself map every year's new meeting routes, and a change in the all-partnered share would not reveal the cause without more evidence.

Why another headline says 50%

SSRS asked a different crowd in January 2026: U.S. adults who had ever tried online dating. Half of those past users said they had at some time been in a committed relationship with someone met through a site or app. The 50% includes former relationships and people who may not currently be partnered. It excludes adults who never tried online dating. Pew's 10% starts with all partnered adults and asks about one present partner. Those figures differ sharply because the questions and denominators differ sharply. Neither is a direct success rate for a new user.

Imagine keeping two lists, without treating either as a prediction. Pew's list contains current partners and marks how they met; SSRS's list contains past online-dating users and marks whether any committed relationship ever followed. A person may appear in both survey concepts, but the published totals do not tell us how many. You cannot divide 50 by 10 to calculate a better chance, and you cannot average them into one share of couples meeting online. Choose Pew for the current-partner question and SSRS for the history-of-relationships question.

Beliefs about relationship quality are another measure

SSRS also asked all U.S. adults in 2026 how successful they believed relationships that began on dating sites or apps were compared with relationships that began in person. Fifty-eight percent thought they were just as successful, 37% thought less successful, and 5% more successful. These are beliefs about relationships in general. SSRS did not follow those relationships over time for this question or compare their breakups. A reader should not replace the 10% meeting-route share or 50% ever-relationship share with the 58% opinion figure. Each percentage is attached to a separate population and question.

The gap between an opinion and an observed outcome is especially easy to lose in a headline that uses the word 'success.' A survey respondent can believe online-started relationships work well without having met a partner online. A past user can report a committed relationship without claiming it lasted or met any particular goal. If a claim says apps make couples more or less likely to stay together, it needs a defined outcome and follow-up period. The SSRS opinion item, by itself, does not provide that test.

What research that follows couples adds

Stanford University's How Couples Meet and Stay Together project offers a different research design. Its 2017 panel began with 3,510 U.S. adults, and the project followed respondents in 2020 and 2022. The public documentation records 2,107 follow-up respondents in 2020 and 1,722 in 2022. Researchers collected meeting stories and coded how people connected. A panel can study change among respondents over time, but its published sample counts are not themselves percentages of couples who met online. Do not turn the 3,510 into a meeting-route estimate.

The Stanford data also require choices about which relationships, waves, and definitions to analyze. A newly formed relationship in a follow-up wave is a different group from all people partnered in 2017 or from Pew's all-partnered adults in 2022. Follow-up loss matters too: fewer people answered later waves, so analyses need their documented weights and selection rules. The dataset is valuable for a researcher who wants to examine meeting routes and relationship histories, while the Pew 10% remains the clear published answer to the current-partner question used here.

Keep the meeting route and denominator with a quote

When you repeat a couples statistic, put four pieces beside it: the survey year, who counted as partnered, whether the question asked about the current partner or any past relationship, and what counted as online. For Pew's 10%, that reads: current partnered U.S. adults in July 2022, current partner, dating site or app. For SSRS's 50%, it reads: past online-dating users in January 2026, any committed relationship at any time with someone met through a site or app. The wording may seem longer than a headline, but it prevents a large and common mistake.

If you are wondering which app produces the most couples, none of the sources cited here compares relationships formed per user across individual apps. If you are wondering whether a relationship that began online is likely to last, the current-partner meeting share cannot answer that either. The evidence table lays out the related measures so you can select the one that fits your question. The number is most useful when it stays with the people and relationship it actually counted.