Skip to content
Copper Sun

When gratitude backfires in a marriage

November 18, 2026 · 3 min read

A married couple in their fifties in a warm kitchen, one bringing the other a cup of coffee with a hand resting on their shoulder

Say thank you more. It's the one piece of relationship advice almost no one argues with. A newer line of research adds a catch: appreciation helps when it's real.

This is a spoke off what the research says about a strong marriage; the sources are on the gratitude and appreciation index. It pairs with does gratitude really help a marriage?.

Gratitude usually helps

Start with the upside, because it's well supported. Across observational and experimental samples, one partner's expressed gratitude for the other's efforts to change was linked to greater success at making that change, consistently explained by more self-determined motivation (Sisson, 2025). The experiments support a causal reading, though the change was self-reported.

Behavior shifts too. In a five-week randomized field experiment, couples assigned to a gratitude-expression plan spent more time physically together in daily life than a control group (Chang, 2022). As a randomized design it points to cause, though it measured time together rather than relationship quality directly.

The cost of overstating it

The catch arrives when the thanks outruns the feeling. Across four dyadic studies of 639 couples, expressing or perceiving more gratitude than was genuinely felt — the authors call it "gratitude amplification" — was linked to personal and relationship costs, explained by lower authenticity and responsiveness (Shimshock, 2026). The designs are correlational and short-term, so they suggest rather than settle.

Sincerity does the work

What a partner reads as sincere seems to matter more than the literal truth of it. In 214 couples studied through daily reports, a lab session, and a three-month follow-up, both expressing and perceiving sincerity about gratitude predicted greater well-being, while accurately reading a partner's sincerity did not (Le, 2026). Perception outweighed accuracy. The findings are correlational.

Felt beats said

There's also a case that being appreciated matters more than being thanked. In 316 couples followed over about 16 months, feeling appreciated buffered the link between stressors and worse relationship quality, whereas expressed gratitude did not — and the protection held only for ineffective arguing, not financial strain (Barton, 2023). It is one specific population, observed over time, so it may not generalize.

The practical read

The takeaway isn't to say thank you less. It's that appreciation works when it's genuine and specific, and that making a partner actually feel appreciated counts for more than a reflex "thanks." Noticing real things worth appreciating, and remembering them, is the part that takes effort.

Husband Field Guide is a private field guide for iPhone — a memory and connection tool, not counseling or therapy — that helps you hold on to the particulars worth appreciating, so the thanks you give has something true behind it. It reports research like this and recommends nothing.

Common questions

Does gratitude help a relationship? Often. A partner's expressed gratitude for change efforts was linked to more successful change (Sisson, 2025), and a gratitude-expression plan increased couples' time together in a randomized experiment (Chang, 2022).

Can saying thank you too much backfire? Yes, when it's overstated. Expressing or perceiving more gratitude than is genuinely felt was linked to personal and relationship costs through lower authenticity (Shimshock, 2026).

Does it matter whether gratitude is sincere? It appears to. Perceived sincerity about gratitude predicted greater well-being, more than whether a partner's sincerity was read accurately (Le, 2026). These are mostly correlational findings.

Is feeling appreciated more important than being told? In one study, feeling appreciated buffered stressors on relationship quality while expressed thanks did not (Barton, 2023). It was a single population observed over time.