How often do you need to see someone to become friends?

"How often should you see friends" is really two questions wearing one search query. If you've just met someone promising, you're asking a building question: how often do we need to hang out before this becomes a friendship? If you're worried about the friends you already have (the group chat that's gone quiet, the best friend two time zones away), you're asking a keeping question: how often is enough to stay close?

They have different answers, because building and keeping run on different clocks. Most of the frustration in adult social life comes from running both on the same vague setting ("sometime"), which is the wrong speed for each. Here's what the research actually supports, for both.

How often should you see someone you want to become friends with?

About once a week, and the reason is more interesting than the number. Frequency in early friendship isn't linear, where monthly delivers a quarter of weekly's results. It's closer to categorical: above a certain rhythm the relationship compounds, and below it, each meetup spends part of itself re-establishing what the last one built. Warmth, ease, in-jokes, the thread of each other's lives: all of it decays between meetings. Meet again soon and you resume; wait too long and you partially restart, covering old ground with someone who has become slightly a stranger again.

We've written about this threshold at the level of a whole room: the recognition mechanics of being a regular, where below a certain rhythm a group's memory of you resets between visits. The same mechanism runs inside a single budding friendship. Two people who meet weekly are continuing a conversation. Two people who meet every six or eight weeks are having a series of pleasant first meetings, indefinitely. That's why "we grabbed dinner twice this year and it was great!" never seems to graduate into friendship: it isn't accumulating, it's repeating.

Why weekly beats monthly by more than 4x

Frequency also decides how fast you can bank the raw material friendship is made of: shared hours. The University of Kansas research we've covered in depth found it takes roughly 50 hours of time together to reach casual friendship, ~90 to friend, and 200+ to close friend. Once you know the thresholds, cadence stops being a vibe and becomes a rate problem. A monthly two-hour dinner banks about 24 hours a year: two years to a casual friend, most of a decade to a close one. A weekly three-hour activity banks around 150: casual friends in about four months, close friendship inside a year and a half. Same person, same chemistry, wildly different arrival times.

And the bars actually understate the gap, because of the decay problem above: the monthly hours don't just accumulate slowly, they accumulate leakily.

One more honest note on timing: the early weeks matter disproportionately. This isn't a precise lab finding (no study has clocked a magic window), but everything in the mechanics points the same way, which is why our standing advice is to protect the streak for the first six weeks. Early on there's no history to coast on, so momentum is doing all the work; a skipped month at week three usually ends the experiment quietly, while the same gap at month eight is a blip. Front-load the rhythm when the friendship has the least to fall back on.

How often do you need to see friends you already have?

Far less often, and this is where the research shifts from Hall's stopwatch to Robin Dunbar's decay studies. Dunbar's group has shown that friendships, unlike family relationships, are maintenance-hungry: in an 18-month study tracking people's full personal networks across a major life transition (Roberts & Dunbar, 2015), friendships that didn't get active upkeep measurably declined in emotional closeness, while family ties largely held their rank without it. Friends drift down the rungs; kin mostly don't. The same work found a wry gender split in what counts as upkeep: for women, decline was best prevented by more talking; for men, by more doing things together.

How much contact keeps a friendship on its rung? Dunbar's layer research offers the most cited answer, and it's worth stating carefully: across his studies, the innermost circle of about five closest friends is typically in touch around once a week, and the next circle of about fifteen around once a month. Those are observed averages (how often people who stay in each layer actually make contact), not experimentally tested prescriptions. But as a planning floor they're the best-supported numbers we have: meaningful contact with your closest people about weekly-to-monthly, and if a close friend has quietly gone six months without any, the research says that friendship isn't paused. It's demoting.

The good news is that established friendships fail much more gracefully than new ones, because the hours you banked act like a battery. A 200-hour friendship holds a deep charge: it drains slowly through a gap and refills fast on contact. One long evening and you're "back like no time passed." A 10-hour almost-friendship has no reserve at all, which is why the same three-month gap that a close friendship shrugs off will kill a new one outright. This is also why rekindling an old friendship is usually a better bet than its long silence suggests: the charge is low, not gone.

And the further out the rung, the less the clock asks. Dunbar's layers get bigger and lower-maintenance as they go (the circle of ~50, the wider 150), and those outer rungs genuinely run on ambient contact: the gym nod, the birthday message, the twice-a-year catch-up. That's not neglect; it's the correct cadence for that tier. You don't owe every relationship a weekly slot, and trying to give one to everyone is how people burn out and then under-maintain the five who actually needed it. Cadence is a budget. Spend it steeply where it's steep.

Does it have to be in person, and does it have to be big?

Two clarifications that make the maintenance clock much easier to hit.

First, small and regular beats big and rare. Maintenance is about keeping your model of each other's life current (who's stressed about what, what changed since last time), and a 15-minute weekly walk does that better than a lavish quarterly production that spends half its runtime on catch-up. If seeing a close friend has started to require an occasion, the friendship is being maintained expensively and badly at the same time. Downgrade the grandeur, upgrade the rhythm.

Second, calls count, but asymmetrically. For an established friendship, a phone call is genuine maintenance: the knowledge-updating that keeps closeness alive works fine over audio, which is how good long-distance friendships actually survive. For a new friendship, calls are weak currency: the same research tradition that gave us the hours thresholds found that shared, chosen, in-person leisure time is what moves an acquaintance toward friend, the same way work hours barely count either. Calls protect what in-person time built. They rarely build it.

The two-clock model

Put both halves together and the whole question resolves into one usable rule. Different rungs need different cadences:

The deepest mistake the model exposes is using the keeping clock to do building work. Monthly-ish contact is a perfectly good preservative and a hopeless construction schedule: it maintains friendships that exist and quietly fails to create ones that don't. If you have plenty of almost-friends and no new real ones, this is almost certainly why: everyone involved is maintaining a friendship that hasn't been built yet.

It's also, plainly, why AroundMe's groups meet weekly. A small group of 4 to 6 people, a real meeting spot, the same faces on a standing rhythm: that's the building clock, installed by design, so the cadence the research keeps pointing to is the default rather than a scheduling achievement. The keeping clock you can run yourself: it only asks for a monthly walk. The building clock is the one nobody's calendar produces on its own anymore. So we built one.

Sources: Jeffrey A. Hall, "How many hours does it take to make a friend?", Journal of Social and Personal Relationships (2018); Sam G. B. Roberts & Robin I. M. Dunbar, "Managing Relationship Decay: Network, Gender, and Contextual Effects", Human Nature (2015); Alistair Sutcliffe, Robin Dunbar, Jens Binder & Holly Arrow, "Relationships and the social brain: integrating psychological and evolutionary perspectives", British Journal of Psychology (2012).