By Matthew Firth, founder of NexSpark · Jul 20, 2026 · 5 min read
Why a Circle is four to six people, not two or fifteen
Research on natural conversation groups, most associated with anthropologist Robin Dunbar's work on the cognitive limits of social bonds, has found that free-flowing conversation tends to max out at around four active speakers before it naturally splits into smaller threads. A Circle of four to six sits right at that seam: big enough that no single person has to carry the whole conversation, small enough that it stays one real conversation, or two that merge and split on their own, instead of becoming a crowd nobody actually talks to.
The hours it actually takes to make a friend
Communication researcher Jeffrey Hall tracked 355 people through real relocations, new jobs, and new schools, and published the results in the Journal of Social and Personal Relationships in 2019. The finding: it takes roughly 50 hours together to move from stranger to casual friend, about 90 hours to become a real friend, and over 200 hours before a friendship counts as close. Crucially, the hours that count are shared activity, doing something together, not passive co-presence like sitting near each other. That single finding shaped the product more than any other: it's why a Circle isn't a one-off dinner but a group with tools built in to keep meeting, an availability poll, venue ideas, one tap to plan the next hangout.
Why groups beat one-off meetups
A one-on-one meetup with a stranger borrows all the pressure of a blind date: performance, silence with nowhere to hide, an unspoken verdict at the end. Friendship rarely starts that way in real life, it starts alongside other people, where you can drift in and out of a conversation and like someone before you're ever alone with them. Small-group research on friendship formation consistently finds mixed groups, some steady presences, at least one natural organizer, outperform groups of near-identical people, especially once a group is meeting regularly rather than once.
Similarity gets you in the door. It's not the whole story.
Half a century of friendship research keeps landing on the same starting point: similarity is where friendship begins, a shared interest, a shared neighborhood, a shared life stage. That's why every Circle is built around one anchor interest everyone actually has. But similarity alone doesn't sustain a group. The same research that shows similarity sparks a first conversation also shows that groups made entirely of similar people, five quiet observers, five talkers, tend to stall. So Circles are composed on purpose: one shared anchor, then deliberately mixed energy and pace, so the conversation has somewhere to go after the first ten minutes.
Why loneliness is the actual problem
None of this matters in the abstract. The U.S. Surgeon General's 2023 Advisory on loneliness found the health toll of poor social connection is comparable to smoking up to 15 cigarettes a day, roughly half of U.S. adults report feeling lonely, and, notably, the most digitally connected generations report the most loneliness. That's the problem NexSpark Friends is built to help with: not another feed, an actual room with actual people already gathered in it.
Source: U.S. Surgeon General's Advisory, "Our Epidemic of Loneliness and Isolation," 2023.
What we won't claim
No algorithm can guarantee two strangers will click, and we don't pretend otherwise. What research-informed design can do is remove the bad odds: the person with nothing in common, the group that can never find a shared free evening, the mismatched energy that never gets a conversation started. The 200 hours are still yours to put in. We just try to make sure none of them are wasted on the wrong table.
Common questions
Is NexSpark Friends actually based on science?
Yes, the core design choices are grounded in published research: Dunbar's work on natural conversation-group size, Jeffrey Hall's study on how many hours it takes to become friends, and similarity-attraction research on what kicks off a new friendship. We also say plainly where the evidence is thinner, like whether an algorithm alone can guarantee two people will click.
Why are Circles four to six people instead of a one-on-one meetup?
Research on natural conversation groups finds free-flowing conversation tends to max out around four active speakers before it splits. Four to six sits right at that seam.
How long does it actually take to make a friend?
Roughly 50 hours together to become casual friends, about 90 hours for a real friendship, and 200+ hours for a close one, according to Jeffrey Hall's 2019 study, provided the time is spent on shared activity rather than passive co-presence.
Can matching software really guarantee a Circle will click?
No, and we say so directly. It can remove the bad odds. The rest is genuinely up to the people at the table.