KeptKept
← All research

How many friends do you lose over time?

Half a personal network turns over every seven years

Mollenhorst, Völker & Flap · Social Networks

1,007 adults interviewed, 604 re-interviewed seven years later.

Seven years on, only 48% of the people in someone’s personal network were still in it. For the closest roles — the people you discuss important matters with, the people who help you out — only around 30% still held the same position.

How it was done

A Dutch panel study. 1,007 respondents aged 18–65 were interviewed, and 604 of them were re-interviewed seven years later about the same personal network.

What it means

The striking part is not the loss. It is that network size stayed constant. People did not end up with fewer relationships; they ended up with different ones. Names left and other names arrived, and the total held steady, which is precisely why the turnover is so easy to miss from the inside.

The ties that ended largely did not end in conflict. They ended because the shared context that produced them — the job, the course, the neighbourhood, the weekly thing — stopped existing, and nothing replaced the regular unplanned contact it had been supplying.

That reframes the problem. It is not that people are bad at friendship. It is that friendship in adulthood loses its infrastructure, and almost nobody notices in time to do anything about it.

What it doesn’t show

This is one panel, in one country, over one seven-year window, and attrition between the two waves is itself a source of bias. It also describes a population, not a person: the average says nothing about whether your own network is turning over. Widely circulated coverage attaches a figure of 52.9% to the share of ties that ended for want of chances to meet — we have not confirmed that number against the paper’s own tables, so it is not stated here.

What Kept does about it

This is the finding the whole product sits on, and it is the reason Kept is framed as maintenance rather than repair. If ties mostly lapse rather than break, the useful intervention happens years before anyone would say there is a problem.

Read the source

Mollenhorst, Völker & FlapSocial Networks

Confidence: B — a single strong peer-reviewed study, or a consistent multi-study literature. We don’t publish anything weaker than this.

Related

Kept is what we built from this.

A private voice journal for the people closest to you. Nine seconds of talking after you see someone, and the detail comes back before it matters.