How to rebuild trust in a relationship — it depends what broke

Two entirely different injuries hide behind that one phrase. They need different first moves, and on one point they need opposite ones. Every page that treats trust as a single subject has to stay vague to cover both.

Here are two people describing the same thing on the same forum.

I don’t believe anyone who comes out with the ‘we are stronger than ever’ bullshit. How can you be when the person you are supposed to trust implicitly destroys that trust forever?

Mumsnet, on surviving infidelity

Our relationship is better now than it was before.

Mumsnet, on trust after cheating — the same poster says it took a few years

Both of them are telling the truth. That is the honest starting position, and it is roughly what the only long-term study of this found as well — though the study has a detail attached that nobody quotes, and we will come to it.

The advice you will be given is the same eight steps everywhere. Acknowledge it. Apologize. Communicate openly. Be consistent. Validate their feelings. Set boundaries. Be patient. Get help if you need it. None of that is wrong, and none of it can be failed or checked, which is why it is possible to follow all eight and get nowhere.

It has to stay that vague, because one list is being asked to cover an affair and a decade of broken promises at the same time. Those are not the same injury and they do not take the same first move.

The number nobody publishes

In 2014 three researchers followed couples for five years after couple therapy. Some had come in with an infidelity the therapist knew about, some with one that stayed secret through the whole course of treatment, and some with none.

The half that never gets quoted is the better half. The couples who did stay married after an affair were, by the end of the five years, statistically indistinguishable from the couples who had never had one — on stability and on satisfaction both. And everyone who stayed married got steadily happier over time, affair or no affair. There are two roads out of this, not one, and the study could not tell in advance which couple was on which.

Which of the two things broke

It is worth having names for them, because the advice diverges immediately.

A rupture is an event with a date. An affair, a hidden debt, one enormous lie. You know when it was. There is a set of facts you do not have.

An erosion has no date. Promises made and dropped, small lies, money quietly moved, four separate occasions of “I’ll stop”. There is no event, so there is nothing to disclose and no apology shaped like the injury.

People living through an erosion routinely wonder whether they are allowed to be this hurt — he didn’t cheat, he just never does what he says — and are often told by the other person that they are overreacting. They are not. When the Gottman Institute lists the kinds of betrayal, sexual infidelity is one of about ten; breaking promises, coldness, absenteeism, unfairness and selfishness are on the same list.

Read the last row twice. It is the only place in this whole subject where the two cases get opposite advice, and it is the advice every ranking page gives undifferentiated.

Why monitoring usually makes it worse

Transparency-as-access is the single most common prescription in this category. Share locations, hand over the phone, open the accounts. The Gottman Institute’s own post on affair recovery recommends credit card records, messages, daily calendars and phone location.

That same post is written by someone who went through it, and she supplies the cost in one word. She describes initially refusing on the grounds that it was an invasion of privacy, eventually accepting it, and then feeling “imprisoned” by the constant accountability.

The research is not on the side of the prescription either. A study of 136 young couples, with both partners reporting, found that doing the checking was associated with lower trust in your partner, more jealousy, and more negative behaviour towards them — an effect on the person doing the checking, not the one being checked. A survey of 329 married people found phone snooping sitting on the road from distrust to breakdown. And a 2024 study of over five hundred couples found that across twenty-four different sharing and monitoring arrangements, every single one had passionate defenders and passionate objectors — and both sides justified their position by appeal to trust. The one thing people agreed on was that whatever the arrangement is, it should be mutual and freely chosen.

All of it is cross-sectional, so none of it proves direction. But put together, the defensible statement is this: there is no evidence that being monitored builds trust, and consistent evidence that doing the monitoring corrodes the person doing it. What settles it is who asked. Access offered by the person who broke trust is a cost they are choosing to pay. Access demanded of them is a symptom being managed.

Asking the question that helps

After a rupture the injured partner has an enormous number of questions, and Esther Perel divides them into two kinds. Detective questions — where, when, how many times, what exactly did you do — produce images that cannot afterwards be unseen. Investigative questions — what did this mean to you, what were you looking for, who were you when you were with them, did you think about me — are the ones her clinical experience says make a recovery possible.

This is not an argument against the truth; the number above is an argument for the truth. It is an argument against mistaking a timeline for an explanation. Perel’s line for couples in the aftermath is that the first relationship is over, and the question is whether they would like to build a second one with each other.

Three things that are true and almost never said

Asking again is a symptom, not nagging

The eleventh time the same question comes round, both people tend to read it as a character flaw. It is closer to a trauma response: intrusive replay and hypervigilance are documented after discovery of an affair, and discovery itself carries a measurable signal for depression independent of how the marriage was doing. Treating the repetition as a fault is the commonest way repair breaks down — the person who caused it gets frustrated that it is not over yet, and the frustration re-breaks the thing.

The one who broke it is allowed to ask for something too

Janis Abrahms Spring’s framework has a piece nobody else carries: both partners make requests. The hurt one asks for what would make them feel safe. The one who caused the injury asks for something that tells them the effort is registering. Without that second channel, repair quietly dies around month eight, which is a real and very common ending. It is not a demand to be forgiven on a schedule. It is a request for a signal.

The old trust is not what you are rebuilding

The trust you had before was the kind you never had to think about, and that particular thing does not come back. What is reachable is smaller and more useful: I can predict you. Naming the smaller target is not a consolation prize. It is the difference between a goal you can make progress against and one you will fail at forever.

When the answer is no

Not everything here is repairable, and the places where it is not are reasonably clear.

• The betrayal has not actually stopped. The affair is ongoing, the account is still hidden, the drinking has not stopped. Nothing is built on a live secret, and the 80% above is what that looks like at five years.
• There is coercive control, intimidation or abuse. This is not a trust problem and joint work is the wrong setting — what you say in the room can be answered for afterwards.
• An untreated addiction or compulsion is driving the pattern. That gets treated first; the relationship work will not hold on top of it.
• There is no accountability at all. The practical test is the cheap things. Someone who will not send a text when plans change is not going to do the expensive things later.
• The injured person is not coping. Intrusive replay, sleep gone, any thought of self-harm — that is a clinician, and saying so is the right answer to the question rather than a disclaimer on it.

And one more, which is the sharpest sentence found anywhere in the research for this page, from a therapist rather than a researcher. There is “a difference between giving someone time to change and giving up your standards so you can tolerate the fact that they haven’t.”

Where inbetween fits, and where it does not

Everything above converges on one bottleneck. The injured person needs to ask the same question many times without being punished for it. The other person needs to stay present for pain they cannot fix, without getting defensive. Both are hard for the same reason: they happen in a live conversation where each of you is reading the other’s face in real time and adjusting.

inbetween takes that one pressure off. You write to it, they write to it, and it is the only one who hears both sides — so the eleventh version of the question can be asked without being watched, and the defensive first answer can exist without landing on anybody. Nothing crosses unless you allow it.

The limit, said plainly because you are right to be suspicious of it: this is not a treatment for betrayal and we are not going to imply that it is. The best evidence that structured help works outside a therapist’s office comes from trials that all included calls with a human coach, and none of them recruited couples in the acute aftermath of an affair. What we can honestly say is that a place to say the true thing helps two distressed people talk. What we cannot say is that it resolves this.

If you are earlier than all of this, the apology page covers the piece that comes first, and what a first therapy session is like is a plainer description than most.