Can one person change a relationship alone?

Reviewed for evidence accuracy · July 2026

You can. Partly. And the part matters more than the word "partly" suggests.

If you're the one reading the books, booking the therapy, learning the vocabulary while your partner does none of it, here is the honest shape of what you're working with. You can reliably change your own reactivity and your own half of whatever loop the two of you keep running. That changes the interaction, in real time, every time, because a loop needs both halves to keep executing. You can also, over months, shift the odds on what your partner does, which is a real thing and a documented one. What you can't do is guarantee the relationship gets happier, and you can't do it by understanding your relationship better. Understanding doesn't transmit. Behavior does.

That sentence isn't a slogan. It's the difference between two studies.

The cleanest test anyone has run on this question

Most relationship research can't answer "does solo work work?" because it never tried the same program both ways. One group did.

Brian Doss and his colleagues built OurRelationship, an online program derived from Integrative Behavioral Couple Therapy, and they built it in two versions. The couple version puts both partners through roughly eight hours of content. The individual version puts one partner through the identical eight hours, alone.

The couple version works. In a nationally representative randomized trial, couples improved on relationship satisfaction (d = 0.69) and on confidence in the relationship (d = 0.47). A later trial with around two thousand lower-income couples replicated the satisfaction gain (d = 0.53) and found a large drop in conflict (d = −0.78). When both people are in it, the relationship moves.

Now the individual version. Same content. One person. Compared against a waitlist, those participants improved on quality of life (d = 0.69), work functioning (d = 0.44), and perceived health (d = 0.49). Real effects, and not small ones.

Look at what isn't on that list. The thing the couple version delivers, the satisfaction gain, doesn't show up. Same material, same framework, same eight hours, delivered to one person instead of two, and the relationship outcome largely evaporates while the personal outcome survives intact.

Which means the bleak reading is available, and honest people have to sit with it for a second: you can feel better, function better, and understand your relationship better without the relationship getting better. That's what the data show. Any product, book, or program that sells solo work as relationship repair has to answer for it, including ours.

But that's not the end of the evidence, and the rest of it changes the picture.

The counterweight comes from addiction treatment, of all places

There's a program called CRAFT, for Community Reinforcement and Family Training, developed by Robert Meyers, Jane Ellen Smith, and William Miller. It was built for the hardest version of the problem you're living: your partner has a serious problem, they refuse treatment, and you're the only one who wants anything to change.

CRAFT coaches one person. The partner never shows up, never consents to the program, often doesn't know it exists. And in head-to-head trials it engaged 64 to 67 percent of treatment-refusing partners into treatment, against 13 percent for Al-Anon-style facilitation and 30 percent for the confrontational "intervention" everyone has seen on television. It also improved the concerned person's own wellbeing, reduced the partner's substance use, and improved the relationship's quality. In a self-directed version, with no clinician in the room at all, it still engaged 40 percent.

Its founding move is a rejection of the premise most solo advice is built on. Where Al-Anon teaches that the concerned person is powerless and should practice loving detachment, CRAFT says the person has an active role in the other person's behavior change, and then teaches them how to use it.

So the field contains a null and a strong positive on the same question. The interesting work is in why they differ, because that's where your instructions are.

The difference is what the program asked the person to do

OR-I taught one person to understand their relationship. It walked them through their differences, their sensitivities, the outside stress on them, and the pattern they get stuck in. That understanding improved the person who held it, and then stopped at the edge of their own skin, because insight held by one person doesn't change what happens between two people. Nobody else can feel your reframe.

CRAFT taught one person to do things differently. What they reinforce. What they stop rewarding. What they say, when they say it, and what they allow to happen as a natural consequence. Those changes are visible from the outside. They land in the other person's actual experience of being in the room with you, and the other person's behavior responds, because a relationship is a feedback loop and you control half the inputs.

That's the line. The question is whether the thing you changed is something your partner can actually encounter. A realization can't be. A different response to the same trigger can.

What one person can move, reliably

This is the ledger. It's not everything, and it's more than most people think they have.

Your own regulation. In Gottman's physiological work, when heart rate climbs past roughly 100 beats a minute, the parts of the brain that listen and take another perspective start going offline. Above that line neither of you can hear anything. You can only control one of the two bodies in the room, and controlling one is often enough to stop the cascade, because the escalation needs both of you feeding it.

Your half of the loop. The most documented destructive pattern in couples research is pursue–withdraw: one person presses to talk, the other pulls back to keep the peace, the pulling back reads as abandonment, which produces more pressing. Both people are usually trying to protect the relationship. The pattern is what's hurting them. Here's the mechanical fact that gets buried under all the advice: the loop requires both moves. When you stop making yours, it cannot run the way it did. It might run some other way, and we'll get to that, but the specific thing that has been eating your Sunday evenings for two years can't execute unchanged.

Your differentiation. This is the family-systems term for staying yourself while staying connected: speaking from your own position instead of prosecuting theirs, not getting recruited into every triangle, self-soothing rather than requiring your partner to regulate you. A scoping review across 295 studies found ample support for differentiation as a predictor of higher marital quality and lower conflict. That's correlational evidence, not proof that raising yours will raise your marriage, and it's worth being honest about the difference. But the clinical tradition here has always worked through one person by design, and it's a plausible mechanism for why solo work sometimes moves the room.

(A related belief in that tradition failed the test, and you should know it because it protects you from a bad inference: the assumption that people pair up with partners at a similar level of differentiation was not supported. Your own reactivity tells you nothing reliable about theirs.)

What you reinforce. The CRAFT engine, in ordinary form. What you reward with attention and warmth, what you decline to fight about, what you stop rescuing them from. It works slowly and it works from the outside in. It has to be authentic, not a technique you're running on someone, and I'll say plainly why in the safety section below.

Your own appraisal. Eli Finkel's group tested a small individual writing exercise: a few times a year, participants wrote about a recent conflict from the perspective of a neutral third party who wants the best for everyone. Marital quality in that group held steady over a year while the control group's declined. It's the one clean case of a low-touch, one-person intervention moving a relationship measure, and it's worth noticing what it targeted. Not the relationship in general. One specific reappraisal of one specific fight.

The part you can only influence

Your partner's behavior, over time. CRAFT proves that's real, and the effect isn't marginal. But 64 percent is not 100 percent, and probability is not a lever. You can change the odds on someone. You cannot operate them, and any framework that promises you can is either lying or selling something.

The practical consequence is that you have to hold your effort and their response as two separate things. Otherwise every week they don't change becomes evidence that your work failed, and it wasn't your work that failed.

Four things you can't do alone

Four things, and they're worth knowing before you spend a year on them.

You can't manufacture safety the other person won't reciprocate. You can offer it. Attachment repair, the kind Emotionally Focused Therapy does, needs both nervous systems present and willing. Yours alone can't complete the circuit.

You can't solve a problem that is genuinely structural. Incompatible core goals, an untreated illness, a decision one of you has already made and hasn't said out loud. Getting better at your half of the conversation doesn't dissolve any of those, and it can waste years pretending to.

You can't out-change a system that's pushing back. This one nobody warns you about. When one person in a stable pattern changes, the pattern often exerts counter-pressure to restore itself, and the counter-pressure can look like your partner getting worse right when you're finally doing it right. Family-systems clinicians have described this for fifty years, though the evidence for it is clinical rather than measured, so hold it as a lens and not a law. Practically: don't read the first bad week as proof it isn't working.

And you can't move things through insight alone. That's the OR-I result, and it deserves to be said twice.

When this is the wrong question entirely

Everything above assumes a relationship where two people, on their better days, want each other's good.

If what's happening at home is coercive control, intimidation, or violence, none of this applies, and the "change your half, influence the rest" frame becomes actively dangerous. Abuse is not a conflict pattern with a stubborn participant. It's a different phenomenon, and the responses that help conflict are the responses that endanger someone in an abusive relationship. Being coached to improve your side of a dynamic, to reinforce better behavior, to stay regulated in the face of someone else's escalation, is a script that puts the responsibility for another person's cruelty onto the person absorbing it. If you've been reading this and quietly translating it into "so I need to get better at not provoking him," stop reading. That thought is the tell, and this piece isn't the help you deserve.

What nobody knows yet, including us

Here's the seam, and you may as well see it.

Every study above tested a planned, deliberate, multi-session intervention. Eight hours of program. Structured behavioral training. A writing exercise done a few times a year in a calm room. None of them tested whether these effects survive compression into something you reach for at 11pm with your hands shaking, mid-fight, in the bathroom with the door locked. That question is genuinely open.

And CRAFT's evidence base is substance-use-specific. Extending its logic to ordinary romantic conflict is an extrapolation, a reasonable one, but not a tested finding, and you should treat anyone who cites it at you (including us, including this essay) as making an argument rather than reporting a result.

We build a tool that coaches one person during a conflict. So we're wagering on exactly the unproven part, and we'd rather tell you that than let you find out. What we can honestly say the work supports: regulating yourself, breaking your half of a pattern, saying the harder thing in a way that can actually be heard, influencing what your partner does over time, and getting clear about what you're in. What no one can honestly promise you, from a phone, alone: that your relationship will be happier, or that your partner will change.

Where this leaves you

Change your half. Influence the rest. Get clear either way.

That third clause is not a consolation prize. Some of the people who do this work well end up with a relationship that softened, and some end up knowing, without bitterness and without a case file, that they're leaving. Both of those are the work succeeding. What isn't succeeding is another year of understanding the pattern in exquisite detail while doing exactly what you did last Tuesday.

So the question worth taking into the next fight isn't whether you can fix this by yourself. It's smaller and harder than that: when they say the thing they always say, what will you do instead?

---

Where these claims come from

• OurRelationship couple version (OR-C). Nationally representative randomized trial: relationship satisfaction d = 0.69, confidence d = 0.47. Replication in a trial of ~2,000 lower-income couples: satisfaction d = 0.53, conflict d = −0.78. Research-backed.
• OurRelationship individual version (OR-I). Same content delivered to one partner: quality of life d = 0.69, work functioning d = 0.44, perceived health d = 0.49, against a waitlist control. No relationship-satisfaction gains of the kind the couple version reliably produces. Research-backed. This is the strongest single piece of evidence against solo relationship claims, and we've cited it against ourselves on purpose.
• CRAFT (Meyers, Smith, Miller). Engaged 64–67% of treatment-refusing partners versus 13% (Al-Anon-style facilitation) and 30% (confrontational intervention); self-directed version ~40%. Improved the concerned person's wellbeing, reduced the partner's substance use, improved relationship quality. Research-backed within substance-use treatment. Its application to general romantic conflict is our extrapolation, not a tested finding.
• Conflict reappraisal (Finkel and colleagues). A brief individual writing exercise, done periodically, preserved marital quality over a year where the control group declined. Research-backed, and narrow: it targeted the appraisal of specific conflicts, not relationship education in general.
• Differentiation of self (Bowen tradition). Scoping review across 295 studies: ample support as a predictor of higher marital quality and lower conflict; the assumption that partners pair at similar differentiation levels was not supported. Research-backed as a correlate; clinically grounded as an intervention.
• Flooding. Sympathetic arousal past roughly 100 BPM narrows cognition and degrades listening, empathy, and problem-solving (Gottman's physiological work). Research-backed. Note the threshold isn't universal across all couples and bodies; treat it as a strong default, not a law.
• Pursue–withdraw. The most documented destructive cycle in the couples literature. Research-backed as a description. The claim that removing your move breaks the loop is mechanical reasoning from that description, and it's the honest weight to give it.
• System counter-pressure (homeostasis). Long-standing family-systems observation. Clinically grounded, not measured. We flag it because it's useful and because you'd otherwise misread it as failure.

✦Keep reading

• Three things worth practicing before the hard conversation
• The science we build on
• What happens in your body mid-fight