Can saying thank you strengthen a relationship?

Possibly. The evidence is modest and context-dependent, and it does not say that more gratitude automatically produces a better couple.

Naturally felt gratitude tracked small later changes in relationship ratings among satisfied couples, although the timing differed between self and partner analyses. A randomized month of gratitude conversations produced only two nominally significant average effects across seven models compared with a structured ordinary-event conversation, with no significant change in global relationship satisfaction. Better outcomes clustered when the person receiving the thanks experienced the partner as understanding, validating and caring.

That distinction turns gratitude from a positivity slogan into a relationship signal. A thank-you can say, “I saw what you did, and I understand something about you.” It can also feel compulsory, self-focused or inattentive. The words are not the mechanism by themselves.

Gratitude is not one behavior

Search advice often combines four different things:

1. feeling grateful after a partner does something thoughtful;
2. telling that partner about the gratitude;
3. receiving and interpreting the expression;
4. completing a private exercise such as a journal or gratitude list.

Those events can influence one another, but they are not interchangeable. A private list does not test what happens when a partner hears a thank-you. A warm interaction does not prove that assigning the interaction every week will improve a relationship. And an observer cannot know whether the recipient felt seen merely because the expresser used grateful language.

The strongest relationship studies become useful when these stages stay apart.

Everyday gratitude tracked later relationship ratings

A 2010 daily-diary study followed 67 heterosexual cohabiting couples for 14 nights. Both partners reported thoughtful actions, their emotional responses and how connected and satisfied they felt. Together, they submitted 1,768 of 1,876 possible daily reports.

Thoughtful acts reported by a partner and acts perceived by the recipient both predicted same-day gratitude. Gratitude then predicted changes in relationship quality. Previous-day gratitude predicted men’s next-day satisfaction and was only marginally associated with women’s. It predicted men’s next-day connection but not women’s; neither gender difference was significant. For the partner, a person’s same-day gratitude was associated with the partner’s same-day connection and satisfaction after accounting for the partner’s prior-day rating.

The models adjusted for the previous day’s corresponding relationship rating. The researchers also tested the reverse temporal path and did not find that a better relationship day predicted an increase in gratitude the next day.

That sequence is stronger than a one-time survey, but it remains observational. The researchers did not randomly create gratitude. A good interaction, reduced stress or another unmeasured event could influence both emotion and relationship ratings. The couples were also already doing well: average daily satisfaction was 7.1 on a 9-point scale.

Three recruited same-gender couples were excluded because the model used gender to distinguish the two members of each dyad. The result therefore says nothing about whether the pattern generalizes to those couples.

The diary supports a plausible everyday process. It does not prove that an assigned gratitude routine will reproduce it.

The randomized test used a meaningful comparison

A later experiment asked the causal question more directly. Fifty-three heterosexual cohabiting couples were randomly assigned to one of two structured conversation routines for a month.

In the gratitude condition, partners thanked each other for specific kind acts. In the active-control condition, one person described ordinary parts of the day that the partner had not witnessed, and the partner could respond as much or as little as normal. Couples held initial lab conversations, were prompted for four to six conversations at home and completed up to 28 nightly reports.

The comparison matters. This was not gratitude versus silence. Responding to a partner’s self-disclosure can itself create intimacy. The trial asked whether gratitude added something beyond another potentially responsive exchange; it did not assign attentive behavior itself.

Six couples reported completing no at-home conversation and were excluded from the manipulation analysis after randomization—five from gratitude and one from control. No intention-to-treat analysis was reported. That left 47 couples: 24 in gratitude and 23 in the active control. Participants completed 2,069 of 2,576 assigned nightly reports, or 80.3%. One more couple supplied no nightly data and another missed the second lab visit, so relevant analyses used about 92 people; the acute lab table used 90.

The average effect was modest

After the at-home conversations, people in the gratitude condition rated the exchange as more responsive and reported more positive relational consequences. Across the month, however, the wider result was mixed.

Of five nightly outcomes, gratitude produced nominally significant higher average positive emotion and adaptation or resilience. Relationship evaluation was marginally higher. Daily negative emotion and life satisfaction did not differ. The paper reported no correction for testing the seven main models. At one month, the conditions did not significantly differ in change in global relationship satisfaction or global life satisfaction.

The authors summarized the result across seven main models—five nightly outcomes and two global satisfaction outcomes. Two were nominally significant; a third approached the conventional threshold. These were nominal tests rather than a multiplicity-adjusted family. The couples started at an average 6.33 out of 7 for relationship satisfaction, so the study tested whether a new routine could improve relationships with little room left on the scale.

Responsiveness changed the pattern

Immediately after the first lab conversation, recipients rated whether their partner’s behavior felt understanding, validating and caring. This perceived partner responsiveness became the most informative boundary in the study.

Within the gratitude condition, higher initial responsiveness was associated with better average daily relationship evaluation, positive emotion, adaptation and life satisfaction, and with lower negative emotion. The interactions for one-month changes in global relationship and life satisfaction were only marginal, so they are signals for further work rather than confirmed endpoints.

At low responsiveness, people in the gratitude condition showed more negative emotion than similarly low-responsiveness people in the active control. The trial had no waitlist group, so this does not establish that gratitude harmed people compared with ordinary life. It does show why “just express more gratitude” is too simple. An artificial positive exercise can miss if the recipient does not experience the behavior as responsive.

Other-praising content tracked recipient response

A separate research program analyzed 370 video-recorded gratitude conversations across independent samples. Coders distinguished between two kinds of content.

Self-benefit language centered what the act did for the grateful person. Other-praising language recognized the benefactor’s effort, actions or qualities. Expressions with more other-praising behavior were associated with the recipient perceiving the expresser as more responsive, feeling better and feeling more love.

The content was observed rather than randomly assigned, so this does not prove that adding praise to a script causes a durable benefit. The association is consistent with attention to the person, not merely happiness about the outcome, but it does not establish a formula.

What the studies suggest

The intervention asked partners to discuss a specific act, but it did not compare specific with generic thanks. The conversation study found that more other-praising content was associated with better immediate recipient responses, but that content was not randomly assigned. Together, those studies suggest a hypothesis rather than a proven script: notice the person’s effort or quality, not only your own gain.

The trial did not identify a beneficial frequency. It deliberately avoided daily compulsory conversations because repetition could become uncomfortable or routine. Quality, timing and recipient experience remain open empirical questions.

A prudent boundary

These studies did not test active conflict, harmful behavior or couples therapy. Not using gratitude to erase conflict, excuse harm or demand positivity is therefore editorial safety guidance, not an experimental result. Appreciation can coexist with a difficult conversation; it is not evidence that the difficult conversation is unnecessary.

Primary sources

• Algoe, Gable and Maisel, everyday gratitude in romantic relationships
• Algoe and Zhaoyang, randomized gratitude expression in ongoing relationships
• Algoe, Kurtz and Hilaire, other-praising behavior in expressed gratitude

Independent editorial summary. The authors are not affiliated with LifeScore.