Couples

How to Write a Relationship Agreement (With Examples)

Servanda · 9 min read · Aug 12, 2026

You've had this fight eleven times. It always ends with something that sounds like a resolution: "Okay, I'll say something before I make plans." Three weeks later plans get made, nothing gets said, and now you're arguing about two things instead of one, because there's a broken promise sitting on top of the original problem.

Verbal agreements don't fail because someone lied. They fail because they were never specific enough to break. "I'll be more considerate about plans" has no edge on it. Nobody can tell, on a Tuesday, whether it was honored. So the question stops being what to promise and becomes how to write a relationship agreement that survives contact with a normal week: one page, written by both of you, kept somewhere you can actually find it.

This is not a legal document. Nothing in it is enforceable, and it shouldn't try to be. Its entire value is that it's specific, it's outside your heads, and it stops the two of you from renegotiating the same terms from scratch every six weeks while tired.

Here are the five ways couples wreck this, worst first.

Mistake 1: Writing the relationship agreement while you're still in the fight

Don't draft anything during an argument. An agreement written inside a fight records which of you was more exhausted, not what you both actually want, and it will be resented by morning.

Here's how it goes. It's 11:40 at night, the argument has been circling for an hour, and one of you says, "Fine. Let's write it down then, since apparently I don't listen." One person dictates, the other types with their jaw set. What ends up on the page is a clause like "I will not bring up my mother's visits after 9pm," which addresses the last four minutes of the conversation and nothing about the actual recurring problem. Signed at midnight, dead by Thursday.

The pull toward doing it right then is real, and it isn't stupid. You finally have the other person's attention, which is rare. There's a fear that if you don't lock something in now, the whole thing evaporates back into the fog it came from. And typing something feels like progress after an hour of going nowhere.

But you're both physiologically incapable of good drafting at that moment. John Gottman's research on conflict describes flooding, the state where heart rate climbs past roughly 100 beats per minute and problem-solving capacity drops away; the Gottman Institute's standard recommendation is a break of at least 20 minutes before returning. Twenty minutes is enough to resume a conversation. It isn't enough to write policy.

Instead: split it into two sittings, and put the second one in the calendar.

Sitting one happens within a day of the fight, alone, separately. Each of you writes three sentences and nothing more: what happened, what I actually wanted, what I'd do differently. No clauses, no negotiating, no showing each other yet. This takes four minutes and it captures the detail you'll lose by the weekend.

Sitting two happens between two days and a week later, at a time you'd never naturally choose for a hard conversation. Sunday at 10am with coffee. Set 40 minutes. The strangeness of talking about the dishes when nobody is angry about the dishes is the point.

And write one rule into the agreement itself: nothing gets added to this document during the argument it came from. If a new problem shows up on a Wednesday, it goes on a list. Lists are for Wednesdays. The document is for Sundays.

Mistake 2: "Be more considerate" is not a clause. Write things you could film.

A line belongs in your agreement only if a stranger with a camera could tell whether it happened. Attitudes, effort and intentions aren't checkable, so every one of them produces a second argument about whether the first one was honored.

This is the most common failure in written couple's agreements, and it's the one that makes people conclude the whole exercise is useless. You write "be more present in the evenings." Two weeks later one of you says it hasn't changed and the other says it absolutely has, and there is no fact in the world that settles it. You've written a mood, not an agreement.

Compare:

Instead of this Write this
Be more present in the evenings Phones go in the drawer by the door from 7:00 to 8:30. Either of us can call it earlier if something's on fire.
Don't be so critical about the house Kitchen counters clear before bed. Whoever cooked doesn't clean.
Communicate better about money Any non-grocery purchase over $150 from the joint account gets a text first. Not permission. A heads-up.
Stop shutting down when we argue Either of us can say "I need twenty minutes." Whoever calls it names the time they'll be back, and comes back at that time.
Share the mental load Toilet paper, dish soap, coffee: Dev's list. If we run out, he orders it that day and nobody has to mention it twice.

Attitude clauses are tempting because they feel generous. Writing "clear the counters before bed" feels like nagging with extra steps, and there's a real fear of sounding petty in writing. The deeper reason is that the thing you actually want is emotional. You want to feel like a priority. Reducing that to a drawer and a time slot feels like it misses the point entirely.

So write both. Put the feeling in as a heading and the behavior underneath it as the clause. The heading explains why the clause exists, which is what stops it from feeling like a chore chart:

What we're after: neither of us feels ambushed by the calendar.

  • The shared calendar is the truth. If it isn't in there, it isn't happening.
  • Anything that takes up a weekend day gets floated by Wednesday.
  • Thursday nights, neither of us schedules anything, including work.

Numbers, days of the week, dollar amounts, names. If a clause contains none of those, it isn't finished.

Mistake 3: Trying to cover the whole relationship instead of the three fights you actually have

A relationship agreement covering 30 topics is a document you will open exactly once. Write down only the arguments you've had more than twice, and leave everything else alone.

The overbuilt version is easy to spot: a four-page shared doc with headings for Finances, In-Laws, Chores, Sex, Communication, Future Kids, Pets, Holidays. Most of those sections regulate things that have never once caused a problem, which is what makes the whole document feel like surveillance rather than relief. Nobody wants a written policy on how they load a dishwasher they've never been criticized for loading.

Comprehensiveness is tempting because it feels like commitment. It's also easier. Drafting careful language about a hypothetical disagreement over future holidays is far more comfortable than writing the one clause about the thing that has genuinely been eroding you since February.

Instead: three to five clauses. One page. Under 500 words.

Select them with a single question, asked separately and then compared: which arguments have we had at least three times? Not the worst fight you've ever had. The repetitive one. Gottman's research, summarized in The Seven Principles for Making Marriage Work, puts the share of marital conflicts that are perpetual rather than solvable at around 69%. You're not writing this page to solve those. You're writing a procedure so that the same disagreement costs you twenty minutes instead of a weekend.

Here's what a finished one looks like. Say it belongs to a hypothetical couple, Maya and Dev, both working, no kids yet, and the recurring fights are money, weekends, and the fact that Maya raises problems at 11pm.

Our agreement. Written 12 March. Next review 12 June.

Money (so neither of us gets surprised by the balance) - Purchases over $150 from the joint account: a text before, not after. Either of us can say "can it wait till the 1st?" - Personal accounts are nobody else's business, no questions, no minimum. - We look at the joint account together on the first Sunday of the month. Fifteen minutes. Dev opens the laptop.

Weekends (so we stop bidding against each other) - Anything claiming a weekend day goes in the shared calendar by Wednesday night. - One weekend a month, nothing social gets booked. Marked in the calendar in advance.

Hard conversations (so they stop happening at midnight) - Nothing heavy starts after 10pm. Maya can say "I'm putting this on Sunday" and Dev has to accept that as a real answer, not a brush-off. - Sunday 10am is the slot. Twenty minutes, coffee, at the table, not in bed. - If Sunday gets skipped, whoever skipped it names the replacement time within a day.

That's it. Three sections, ten lines, and it addresses everything they actually fight about.

The hardest part of this isn't the writing, it's making sure both of you propose clauses rather than one person authoring the document and the other approving it, which is the single most reliable way these agreements curdle. If you want structure for that, Servanda walks both of you through proposing terms separately and then reconciling them into one page, so the finished agreement doesn't read in one person's voice.

Mistake 4: No clause for what happens when someone breaks it

The agreement will get broken, usually inside the first three weeks. The clause that determines whether the document survives that is the one describing what happens next, and almost nobody writes it.

Picture it. The agreement says phones in the drawer at 7:00. On day nine, one of you is standing in the kitchen at 7:40 dealing with a work thing that genuinely couldn't wait. The other now has three options and all of them are bad: say nothing and file it away, cite the document like a parking inspector ("we literally wrote this down"), or decide the whole thing was unrealistic and quietly stop referring to it. Read a clause aloud in the wrong tone and you've converted a shared page into a piece of evidence.

Writing a breach clause feels like planning to fail. It also feels, if you're honest, like pre-forgiving something you're not sure you want to forgive. That's why it gets skipped.

Instead: write the repair, not the penalty. Three parts, and you need all three.

  1. How it gets named. Agree on a short phrase that carries no argument in it. "That's the phone thing." Five words, no case-building, no history. The whole purpose is to make flagging it cheap enough that it happens immediately instead of accumulating.
  2. What the legal responses are. Two of them: "You're right, one minute" and "I need a pass tonight, work call." Both count. If only agreement counts as a valid response, the named person will start defending instead of adjusting, and Gottman's work on repair attempts is blunt about this: what separates stable couples is not that they avoid the rupture, it's that the repair gets accepted rather than swatted away.
  3. A budget for failure. Two passes each per month, one sentence of explanation, no interrogation. Passes are what make the agreement liveable rather than a purity test.

Add one line distinguishing a lapse from a pattern: the same clause missed three times in a month goes onto the review list, not into a fight. That single sentence removes the incentive to escalate, because there's now a scheduled place for the complaint to land.

And write down one thing that's out of bounds: quoting the agreement during an unrelated argument. If the document can be used as ammunition, it stops being safe to sign.

Mistake 5: No review date, so the relationship agreement becomes a rule instead of a decision

An agreement with no review date turns into something your past selves imposed on your present selves, and people resent rules they didn't get to revise. Put a date on the page before you put anything else on it.

Two things go wrong without one. First, clauses go stale and nobody updates them. The Thursday no-plans night made sense in March; in September one of you has a class on Thursdays, so that line is now fiction, and a page that's 40% fiction teaches you both that the rest is optional too. Second, the document stops being a working tool and becomes an archive. Somebody opens it in October to prove a point about something agreed in March. That's the death of it.

Reopening a calm agreement feels like poking a bear. Things are fine, nobody's fighting about the counters, why invite it. But the review is exactly what keeps it from needing to be enforced.

Instead: 90 days, 20 minutes, three options per clause. Book the next four now: first Sunday of January, April, July, October. Sit down with the page and go line by line asking only keep, change, or delete. Anything either of you wanted to add during the quarter comes off the running list and gets discussed here, not in the moment it came up.

Save the old versions with dates on them. Reading the March version next to the September one is the clearest evidence you'll get that any of this is working, and it's worth more than either of you remembering accurately, which neither of you will.

Deleting clauses is the part people don't expect. When the counters have been clear every night for four months without anyone mentioning it, that line comes off the page. It's not a rule anymore, it's just how you live. Expect two of the clauses you write this month to be gone within a year because they did their job, and expect the page to get shorter every review rather than longer. That's what a working agreement looks like: three lines, a review date, and a fight you can barely remember having.

Having this argument yourself?

Describe what's actually happening. A neutral mediator takes your side of it first, then brings the other person in.