The default that decided

Changing a setting's default from off to on moved about seventy per cent of users, and the setting was one that almost everyone would have described, if asked, as a personal choice.

Defaults are powerful for several reasons that stack. Changing one costs attention, and attention is the scarcest thing a user brings. A default also carries information: a reasonable person assumes the people who built the product chose the sensible option, and that assumption is usually correct. And a default establishes a reference point, so the choice is no longer between two options but between staying and acting.

This makes the selection of a default an exercise of power, which is uncomfortable for designers who would prefer to be neutral. There is no neutral position. A setting must ship with some value, and 'off' is as much a decision as 'on'. The honest response is not to pretend otherwise but to adopt a rule, and the rule most widely used is simple: the default should be whatever a well-informed user would choose for themselves, not whatever the organisation prefers.

That rule is testable. Describe both options to a sample of users in plain language with the consequences spelled out, ask what they would pick, and set the default to the majority answer. Where the majority is thin, make the choice explicit at first use instead of defaulting at all.

The test is also a diagnostic. If a team is unwilling to run it, the likely reason is that the answer is known and inconvenient, and in that case the default has already been chosen for the organisation rather than for the user.