# Americans did not stop wanting partners. They stopped believing the search works.

In 2019, 44 percent of American adults were living without a spouse or partner. By 2023 that number was 42 percent.

Two points does not sound like much. What makes it worth stopping on is the direction. That share had been climbing for about twenty years, through every version of the story about how romance was ending, and then it turned around. It is the first decline in two decades, and it happened during the exact stretch when the collapse narrative got loudest.

## Start with the number hardest to argue with

The figure comes from Pew Research Center's analysis of Census Bureau data, drawn from the [American Community Survey](https://www.pewresearch.org/short-reads/2025/01/08/share-of-us-adults-living-without-a-romantic-partner-has-ticked-down-in-recent-years/), the largest household survey in the United States, with a sample of more than three million addresses. That is a very different thing from a panel of a thousand people who answered an online questionnaire.

The rest of the picture moves the same way. Between 2019 and 2023 the married share of adults went from 50 to 51 percent, and the cohabiting share from 6 to 7 percent. The decline in unpartnered adults showed up across every age group.

Be careful about what this measures, because the precision matters and almost every write-up skips it. The ACS is counting **living arrangements**. Someone in a committed relationship who has not moved in with their partner is counted as unpartnered here. So this is not a measure of romance. It is a measure of how many people are sharing a home with someone, which is a high bar and a slow-moving one, and that is why the reversal is interesting rather than noise.

Pew is also explicit about two limits, and they are worth carrying rather than burying. Census and ACS data slightly overestimate the size of the unpartnered population. And the ACS only identifies cohabiting partnerships that include the household head, so it does not capture every cohabiting couple. Both push in the same direction, which means the level is a bit too high, though neither gives a reason to doubt the change over time.

So the careful version of the headline is this. On the best long-run measure available, the share of American adults not living with a spouse or partner stopped rising around 2019 and has come down slightly since.

## So what did decline?

Two things, and they are both about the search rather than the outcome.

The first is participation. Here Pew switches instruments: a [survey fielded in July 2022](https://www.pewresearch.org/short-reads/2023/02/08/for-valentines-day-5-facts-about-single-americans/), using a broader definition of single than the ACS uses. Not married, not living with a partner, and not in a committed relationship. About three in ten US adults were single by that definition, the same share as in 2019. What changed is what they were doing about it: the share saying they were looking for a committed relationship or dates fell from 49 percent in 2019 to 42 percent in 2022.

That decline is almost entirely men. Among single men it went from 61 percent to 50 percent. Among single women it barely moved, 38 percent to 35 percent, from a level that was already much lower. Any account of what is happening has to fit that shape, and most of them do not.

The second is confidence, and this comes from a third source with its own methods. The Survey Center on American Life, in research by Daniel Cox drawn from two national surveys of more than 5,000 respondents, found that 55 percent of single Americans are pessimistic that they will ever find a long-term partner. Among single men it was 57 percent and among single women 54 percent, so pessimism, unlike searching, is not lopsided by sex. College-educated singles were the most pessimistic of all, at 69 percent.

Cox also found the number that sits closest to the mechanism. Only 41 percent of Americans see dating apps as a safe way to meet people.

:::pullquote
Fewer people searching, most of them pessimistic it will work, and a minority who trust the main venue. That is not a culture losing interest in love. That is a queue at a broken machine.
:::

## Wanting and looking are not the same measurement

It would be easy to read "only 42 percent are looking" as proof that people have stopped caring. That reading confuses two different questions, and the difference is the whole point.

"Are you currently looking for a relationship or dates" is a question about behavior this month. "Do you want a partner" is a question about what you hope your life contains. Cox's finding on the second is that most single Americans who have never married still aspire to find a partner, while confidence and hopefulness are, in his phrase, in short supply.

Put the two together and one reading suggests itself. It is what people do when they still want an outcome and have stopped believing the available method delivers it. They step back from the method. Nobody keeps queueing at a machine that has taken their last four coins.

Label that honestly, because we are about to build on it. That is an inference and not a finding, and it is ours. These numbers come from three different organizations, in different years, using different definitions of who counts as single. Nobody asked the people who stopped looking why they stopped, and nobody has shown that falling confidence caused falling participation. The two could share a cause, or run the other way, and the fact that the participation drop is concentrated among men while the pessimism is not is a genuine loose end in our version of the story.

What can be said without inference is narrower and still worth having: partnership stopped declining, most single people still say they want a partner, fewer of them are actively looking, and confidence in the main venue is low. The causal thread between those is our reading of it.

That reading is a very different diagnosis from the one in circulation, and it carries a very different implication. A culture that no longer values partnership is not something a product can fix, and honestly not something anyone should want a product to fix. A search process people have lost faith in is a different kind of problem. Those get fixed all the time.

## Why the search lost people

We have written about most of this already, and it is worth naming rather than gesturing at. Each claim below is argued with its sources in the linked post rather than re-argued here, and the through-line connecting them to the survey numbers above is Bloom's analysis rather than anything a study has demonstrated.

The dominant format asks you to evaluate dozens of strangers from a few photos, which measurably drives your acceptance rate down the longer you do it, and which leaves people feeling processed rather than met. That is the subject of [why a feed of strangers makes you feel like inventory](/blog/why-a-feed-of-strangers-makes-you-feel-like-inventory). The exhaustion it produces has been surveyed directly, and we went through those numbers in [swipe fatigue](/blog/swipe-fatigue-the-case-for-slower-dating).

Underneath the format is an incentive. When the product is a monthly subscription sold to single people, the best customer is the one who stays single and keeps paying. That is not a conspiracy theory, it is an accounting identity, and its consequences have started showing up in earnings reports and legal settlements rather than only in group chats. We laid that out in [the dating app giants are shrinking](/blog/dating-app-giants-are-shrinking).

And the thing the search was supposed to be optimizing turns out not to be optimizable. When researchers collected more than a hundred traits and preferences from speed daters and asked a machine learning model who would want whom, the model could not answer. We covered that in [machine learning cannot predict chemistry](/blog/can-an-algorithm-predict-chemistry). The category has spent fifteen years selling a prediction that the best available evidence says nobody can make.

None of that is a failure of the people using the apps. It is the predictable output of a system built to sell searching.

## The hopeful reading, which is also the accurate one

Here is what the data supports, stated as plainly as we can.

More Americans are living with a partner than were five years ago. Most single people still want that for themselves. What has fallen away is the belief that the process on offer will get them there, plus the willingness to keep grinding at a process they have stopped trusting.

That is a good position to be in, strange as it sounds. The hard part of building anything is demand, and the demand is intact. What has to change is the thing standing between the wanting and the having.

The specific bet Bloom makes is that the fix is not a cleverer sorting algorithm at the front, because the front is where the evidence runs out. It is the rest of the arc: fewer, more considered introductions, and then an app that keeps going once two people find each other rather than treating the match as the finish line. That is why Bloom spans dating and the relationship itself, and why our business does better when a relationship works rather than when a search continues. The [pricing page](/pricing) sets out how that actually works, without the part where seeing who liked you is the paywall.

## The recession that is not one

Recession is the wrong word for what the numbers show, and we would not have chosen it. A recession implies the demand went away.

The demand did not go away. What contracted was participation in one particular marketplace, by people who mostly still want exactly what that marketplace promised to deliver.

Fixing the marketplace is a smaller and much more tractable job than fixing a culture, and it is the one actually in front of us.

Bloom is pre-launch, starting in Utah along the Wasatch Front. The only thing to do today is [join the waitlist](/#waitlist) and grow with us.