The cost of leaving includes the loss of social ties

What it is

In social products, leaving exits far more than a service — it exits a web of ties. The leaver pays a cost of presence: friends' updates, group chats, the conversations happening here. That cost does not track product quality; it tracks who else has not left. This is social lock-in, distinct from the general lock-in effect in what it holds people with: not data or re-learning costs, but people.

Why it happens

The core structure is a coordination problem: each person's best choice depends on others' choices, and the value of staying equals the share of one's network that is present. When friends are scattered and nobody moves first, everyone could prefer to leave and nobody does — the first mover bears the disconnection cost alone. So even as quality declines, the user base holds: people stay not out of satisfaction but because leaving alone is not viable. A second layer is the non-portability of identity and history: data export carries away photos and archives, not your position in the relationship — who knows you, what you are called, what interactions you had with whom. Half of that lives in other people's accounts and is physically non-exportable. Relational data is bidirectional; a one-sided migration cannot carry a two-way connection.

Where it stops holding

The cost varies with migration channels: where both parties share an alternative venue (offline contact, mutual follows elsewhere), leaving becomes cheap — the strength of social lock-in is a function of relational exclusivity. For content subscriptions and tools, this cost approaches zero; there, exit resistance comes from data and habit. Nor should every instance of staying be attributed to lock-in: genuine use value exists, and only the social component of exit cost is at issue here.

Applying it

Provide portable forms of the relationship itself: full export of follows and friend lists, prompts for alternative channels between mutual ties (exchanging contact information, finding each other across platforms), and announcements or migration help when leaving a group. At the metric level, never read "users did not leave" as satisfaction: under social lock-in, retention systematically overstates perceived value — separate genuine stickiness from locked-in stickiness with willingness-to-recommend items and hypothetical questions about substitute availability. Verification: measure the intent-behavior gap of "wanted to leave but stayed," and track the relationship between export-tool usage and subsequent retention — tools being used while people stay indicates lock-in at work.

Related

• Same group: P3.10.2 The count and wording of retention steps decide whether they inform or obstruct · P3.10.3 Deactivation and permanent deletion are two different exits · P3.10.4 Data processing that continues after exit makes the exit incomplete
• Adjacent: P3.05.3 Lock-in effects weaken freedom of choice · P3.05.2 Data portability is a precondition of exit
• Search terms: social lock-in · network effect · coordination problem