What actually makes relationships last - and why we keep choosing with the wrong instincts

1. Why we lead with emotion (and why that's not stupid)

Leading with feeling isn't a modern failing - it's biology. When researchers scanned the brains of people who were intensely in love, looking at a photo of their partner activated dopamine-rich regions of the brain's reward and motivation system, the same circuitry involved in wanting and pursuit^[6]. Early-stage love behaves less like a judgement and more like a drive.

It even looks a little like obsession. In one study, people who had fallen in love within the previous six months showed the same reduced density of a serotonin transporter as unmedicated patients with obsessive-compulsive disorder - both significantly lower than controls^[7]. That “I can't stop thinking about them” feeling is not a metaphor.

And instinct has real benefits. Seeing a partner a little more generously than they see themselves - what researchers call positive illusions - is linked to greater satisfaction in close relationships^[8]. Romantic bonds also behave like attachment bonds, the same system that bonds infants to caregivers ^[9]. Emotion is how humans bond.

2. What we screen for - and why it doesn't predict much

Modern dating adds a second instinct on top of chemistry: the checklist. Height, job, personality type, hobbies, “must love travel”. It feels rational. The evidence says it mostly isn't measuring what matters.

Chemistry can't be predicted from a profile

In two speed-dating studies, single people answered more than 100 questions about their traits and preferences - the kind of information dating apps collect - then went on four-minute dates with each other. Machine-learning models could partly predict who tends to desire people in general, and who tends to be desired. But they could not predict who would be attracted to whom, using any combination of the 100+ measures ^[2].

Personality matching explains almost nothing

Using nationally representative samples of married couples in Australia (5,278), the UK (6,554) and Germany (11,418), researchers separated three things: your own personality, your partner's personality, and how similar the two of you are. Similarity - the thing most matching quizzes sell - consistently explained less than half a percent of relationship satisfaction ^[4].

Notice what does show up: your partner being agreeable, conscientious and emotionally stable helps a little. Being alike does not.

What we say we want vs. what moves us

Men and women reliably say they weigh attractiveness and earning prospects differently. But a meta-analysis of 97 samples of real attraction and real relationships found that, in practice, attractiveness predicted romantic evaluations about equally for both sexes, earning prospects only weakly for both, and the sex difference was tiny and non-significant ^[3]. Our stated ideals are a poor guide to what actually draws us in.

Similarity matters on paper, not in practice

A meta-analysis of 313 studies found that actual similarity drives attraction before people have really interacted - but its effect shrinks with interaction and was not significant in existing relationships. What kept predicting attraction in real relationships was perceived similarity: feeling like you're alike ^[5].

3. What actually predicts relationship quality

The most comprehensive answer so far comes from a project that pooled 43 longitudinal datasets from 29 labs - 11,196 couples - and used machine learning to test hundreds of variables against each other^[1]. Two kinds of predictor were compared: relationship-specific (how you experience this relationship) and individual differences (your personality, mood and history, whoever you're with).

Three findings stand out:

• How the relationship feels from the inside beats everything else. Once someone's own relationship-specific perceptions were known, their personality, wellbeing, and even their partner's reports added no further predictive power.
• Your own view predicted two to four times more than your partner's view of the same things.
• Nobody could predict change. Whose relationship would get better or worse over time was largely unpredictable from any combination of self-reports. Relationships are made as they go, not found fully formed.

Rankings as reported in Joel, Eastwick et al. (2020), in order.

Notice what's not on either list: shared hobbies, matching personality types, similar music taste, star signs. Notice also what is there: your own wellbeing and attachment patterns. A healthy relationship isn't only about choosing the right person - it's also about the state you bring into it.

4. Conflict, warmth, and the 14-year study

John Gottman and Robert Levenson spent decades filming couples talking - about their day, and about disagreements - and following them for years. In one study of 73 couples, those whose interactions were not balanced toward the positive showed more negative and less positive emotional expression, more stubbornness and withdrawal, and greater defensiveness - and a higher risk of separating^[10]. (Gottman later popularised this as a rule of thumb: stable couples keep far more positive than negative moments, even while they argue.)

Their 14-year follow-up showed that when a marriage ends depends on what goes wrong^[11]:

Warmth on ordinary days matters too. Research by Shelly Gable and colleagues found that how a partner responds when you share good news - enthusiastically and with interest, rather than flatly or dismissively - is tied to how good people feel about the relationship^[12].

5. Passion vs. obsession: does the spark have to fade?

If early love is part-obsession, does long-term love have to mean losing the spark? Not quite. A review and meta-analysis of 25 studies separated romantic love (intensity, engagement, sexual interest) from obsession (anxious preoccupation) ^[13]:

This is the clearest picture of why emotion-first dating misleads us. The anxious, all-consuming feeling that makes a new relationship feel “right” is the very component that turns negative in the long run. The kind of love that lasts is calmer, and it's still romantic.

6. Sliding vs. deciding

Emotion-first relationships often progress without a decision ever being made: a few nights a week becomes moving in, becomes “well, we're already living together.” Researchers Scott Stanley, Galena Rhoades and Howard Markman describe how transitions that are slid into rather than decided can create inertia - some couples end up staying together, or marrying, who otherwise would not have, regardless of how well they fit ^[16].

The long view of marriage research points the same way. A review of 115 longitudinal studies covering over 45,000 marriages argued that how relationships change over time depends on what each person brings, the stresses the couple faces, and how they handle them together ^[17] - not on how strong the start was.

7. What to look at instead

None of this means ignoring attraction - you need it to start, and the research says it persists in good long-term relationships. It means not mistaking it for evidence. Here's what the research suggests paying attention to, stage by stage.

8. How dating apps “create” partners - and why it doesn't add up

Most dating apps follow the same recipe: a feed of photos, a few seconds per person, a swipe, a match, a chat, and - maybe - a date. The app decides who you see, you decide in seconds based mostly on how they look, and a relationship is supposed to emerge from whatever is left. Put next to the research above, the recipe has the order backwards.

1. 1It judges on the weakest signal. A photo and a few lines of bio are excellent at predicting who you'll click on, and close to useless at predicting who you'll click with - nobody can predict chemistry from a profile ^[2], and what we say we want barely matches what draws us in^[3].
2. 2It promises compatibility it can't measure. A major review of online dating by five relationship scientists concluded that the matching algorithms dating sites advertised weren't backed by convincing evidence, and argued that browsing endless profiles can push people to evaluate partners more like products than people^[14]. Personality matching - the most common “science” on offer - explains under half a percent of satisfaction ^[4].
3. 3Endless choice makes us worse at choosing. Across three studies, people browsing potential partners online started rejecting more of them almost immediately - their chance of accepting someone fell by an average of 27% from the first option to the last. Satisfaction with the photos and their sense of dating success both declined, and for women, matches became less likely ^[15]. More options produced a more pessimistic, more rejecting mindset, not better choices.
4. 4The incentives point the wrong way. This one isn't a study, it's arithmetic: an app paid by monthly subscription earns most while you're still searching. We can't speak for any company's intentions, but designs built around endless feeds and likes behind a paywall fit that incentive far better than the goal of helping you meet one person and leave.

9. Reimagining how people meet

We started Oh My Person with one question: if you designed meeting someone around what the research says makes relationships last - instead of around what keeps people scrolling - what would it look like? This is our answer so far. It's a work in progress, and we'd rather be honest about that than pretend we've solved love.

Trusting people again

Underneath all of this is a belief about people: that most of us looking for love come with good intentions. We want to be seen properly, and to see someone else properly. Apps built around speed and scarcity bring out the worst in us - snap judgements, ghosting, treating people as replaceable - and then we blame ourselves, or each other, for how dating feels.

We think the opposite is also true. Give people more context, fewer distractions, and a reason to show up as themselves, and something good tends to happen. So we ask one thing of everyone who joins: come with the right intentions. Be honest about who you are and what you want, be curious about the people you meet, and be kind - especially when it isn't a match.

How compatibility works on Oh My Person →