Gratitude for Complicated Relationships: Can You Be Grateful for Someone Who Also Drives You Mad?

By Matt Hughes · 2026-08-02 · 8 min read

Kansha's AI insights tag some people as 'mixed' — genuine joy and genuine challenge, at once. Here's why that's more common than you'd think, and what the research on ambivalent relationships says about it.

Gratitude for complicated relationships is one of the things people don't expect a journalling app to surface, and it caught me off guard the first time I saw it in Kansha's own AI insights. Someone I'd been writing about for weeks got tagged with the sentiment "mixed" — not joy, not challenge, not the steady grounding label reserved for the reliable people in your life. Mixed. Both at once. And once I sat with it, I realised that's probably the most honest label most of my closest relationships deserve.

We built the People insight in Kansha to read back through your entries and notice who actually shows up — how often, in what mood, and with what kind of charge. Four sentiment categories came out of testing it against real entries: joy (they're a source of good moments), challenge (they're tied up with difficulty), grounding (steady, quietly present, not dramatic either way), and mixed — both joy and challenge, genuinely, in the same person. Mixed turned out to be far more common than any of us expected when we designed the feature.

Why do so many relationships come out "mixed"?

Because that's what most real relationships are like. The parent who's endlessly generous and endlessly critical in the same conversation. The colleague who makes you laugh harder than anyone else at work and also undermines you in meetings you're not in. The old friend who shows up exactly when you need them and cancels on you every third time. If you've been journalling honestly for more than a few weeks, at least one person in your entries is going to earn that label, because you're not writing about a character — you're writing about someone with their own contradictions.

What I hadn't expected, before I read the research properly, is that psychologists have a name for this and have been studying it for two decades: ambivalent relationships. Bert Uchino and Julianne Holt-Lunstad, who between them have done most of the foundational work here, define an ambivalent tie as one that mixes real positivity with real negativity, rather than sitting cleanly on one side or the other. Their examples read like a description of half my own address book — the mother you love deeply who's also overbearing, the good friend who's brilliant company and quietly competitive with you.

Is a mixed relationship actually worse for you than a difficult one?

This is the part that genuinely surprised me. You'd assume a clearly negative relationship — someone who's mostly just hard work — would be the more damaging one to carry. The research says the opposite, at least on some measures. Ambivalent ties are associated with higher ambulatory blood pressure in daily life, greater cardiovascular reactivity when people are asked to discuss a stressful event with an ambivalent friend compared with a supportive one, and higher rates of depression. One study even found shorter telomere length — a cellular marker linked to disease risk and mortality — associated with having more ambivalent people in your social network. Holt-Lunstad and Uchino's later work estimates that as much as half of a typical adult's social network falls into the ambivalent category, which makes this less a niche finding and more a description of ordinary life that most of us haven't put a name to.

The theory for why mixed relationships take a bigger physiological toll than purely negative ones is fairly intuitive once you hear it: with someone who's just difficult, you know where you stand and you can manage your expectations accordingly. With someone who's both a source of real support and real stress, you can't relax into either mode. You're braced for the good and the bad in the same interaction, and that low-grade vigilance is expensive to run, day after day, for years.

So does writing gratitude entries about a difficult person actually help?

Here's where I want to be careful, because the honest answer is "sometimes, and not always in the way you'd hope." There's decent evidence that gratitude helps couples weather ordinary relationship friction — appreciation from a partner is linked to higher satisfaction and more resilience to the everyday stress of arguments and money worries. That's the mechanism most gratitude-for-relationships advice leans on, and it's real.

But a mixed relationship isn't the same problem as ordinary friction, and the research on gratitude and power imbalance is more cautious than the general advice suggests. A 2024 study in Scientific Reports found that in relationships with a clear power imbalance, the usual link between feeling grateful and feeling that your partner is responsive to you weakens considerably for the lower-power partner — gratitude doesn't do the same protective work when the ground isn't level. Writers on unhealthy relationship dynamics go further and warn that gratitude can become "blinding" in genuinely imbalanced situations: you fix so hard on the crumbs of kindness that you stop registering the pattern of harm around them.

I want to be direct about what that means in practice. If the "mixed" person in your journal is mixed because they're inconsistent, occasionally thoughtless, or hard work in the ordinary way people are hard work — writing honestly about both sides of them is healthy, and probably clarifying. If the difficulty is closer to control, fear, or feeling smaller around them than you used to — that's not a journalling exercise, and no amount of noticing the good moments should be asked to carry that weight. If that's what you're describing when you write, please talk to someone — a GP, a friend outside the relationship, or in the UK the National Domestic Abuse Helpline on 0808 2000 247 (freephone, 24 hours) is a good place to start.

How do you actually write about someone who's genuinely both?

Assuming you're in the far more common situation — a relationship that's just properly, humanly complicated — here's what's worked for me, and what I see working when Kansha's own AI insights land well rather than feeling glib.

Write the specific moment, not the verdict on the person. "Grateful Dad drove two hours to help me move a sofa" is a true, useful entry even in a week where the same relationship also produced a conversation that left you flat for a day. You're not being asked to conclude "he's a good dad" or "he's a difficult dad" — you're recording one true thing that happened, and the entry can hold that without resolving the whole relationship.

Let the "and" stand instead of reaching for "but." "She remembered my birthday when nobody else did, and she also said something about my job that I'm still turning over" is more accurate — and easier to write honestly — than forcing it into "she remembered my birthday but she can be thoughtless," which quietly asks you to pick a side.

Notice the pattern over time rather than judging any single entry. This is really what the People insight is doing under the hood — not scoring one interaction, but reading across months of entries to see whether someone trends toward more joy, more challenge, or genuinely sits in the middle. A single entry about a hard conversation with someone you mostly write warmly about doesn't need to reframe the whole relationship, and a single good day with someone who's mostly been difficult doesn't erase that either.

Separate "I'm grateful this happened" from "I'm grateful for this relationship overall." Those are different claims, and conflating them is where gratitude practice can genuinely mislead you about a relationship you should be reassessing. You can be honestly, specifically grateful for a single act of kindness from someone whose overall presence in your life still needs a harder look.

What I've actually changed since seeing my own "mixed" tags

I run a consultancy, which means a fair number of the relationships in my own working life are exactly this shape — people I've learned enormous amounts from and also found genuinely difficult to work alongside, sometimes in the same project. Seeing that reflected back as "mixed" rather than forcing it into a tidier label was oddly relieving. I'd been quietly editing my own journal entries towards consistency — writing the grateful version on good days and just not writing at all on the harder ones — because I think I assumed a "real" gratitude practice should point in one direction about a person. It doesn't need to. The research suggests it probably shouldn't.

The point of noticing a relationship is mixed isn't to force a decision about it. It's to stop pretending it's simpler than it is, so that when a decision does need making — about a friendship that's drifted more towards challenge than joy, or a professional relationship that's costing more than it's giving — you're working from an honest record instead of whichever mood you were in on the day you happened to write.