Reset has to say what will be lost, or users will not touch it

What it is

The button says “Reset.” Beside it, not one sentence that playback will go, that tuned classes will go, that you will return to cold start. People read reset as irreversible self-destruction and would rather live with a wrong profile. Reset cost disclosure means saying, before the press, which abilities and data will be lost, which will remain, and whether the step can be undone — or the control exists and is not used.

Seeing the object answers “change which part.” Cost disclosure answers “what do I pay.” If people dare not touch it, the control is as good as absent.

Why it happens

Loss aversion makes people more sensitive to a vague large loss than to a concrete small one. An unexplained reset is filed as a vague large loss: maybe the account dies, maybe paid benefits die, maybe “the system will never know me again.” Daily discomfort from bad features is more concrete and bearable. The rational strategy is not to press.

“This will clear personalization data” is still vague. People need a checkable list: playback, wishlist, explicit ticks, behavioural counts, inferred tags — each gone or kept. They also need time structure: immediate, undoable within twenty-four hours. Vagueness prices every item at worst case; a list aims loss aversion at what will actually go.

Studying it

The same reset, three copy conditions: no explanation, a vague “clear data,” an itemised list plus whether undo exists. Task: decide whether to press when a wrong profile is already perceptible. Dependent variables: press rate, hesitation before press, regret from pressing wrong (clearing what should have stayed). Independent variables: grain of the list, whether irrecoverable items are marked.

Do not count “the button was seen” as use. Daring to press is the endpoint of this question. A contrast should also include whether what is actually lost after the press matches the list — a mismatch teaches the next person to be even less willing.

Where it stops holding

An action that truly destroys the account must not be named “reset personalization”; send it through delete-account so the words stay unpolluted. When there is nothing to lose (there was no profile), disclosure is empty and the button should not appear. This entry is “if the cost cannot be said, no one will use it.” It does not restate “must see first,” and it does not treat the editorial character of the post-off default ranking.

Applying it

• In the confirm dialog, list what will be lost and what will be kept, in names people recognise (“in-progress episodes stay / taste tags go”), not “data.”
• If it can be undone, write the window; if it cannot, mark that separately. Do not mash both into one “irreversible.”
• Check: make a wrong profile obvious, and count how many people press reset before and after disclosure. If almost no one presses after, inspect whether the list makes kept items look as if they will go too. If actual loss after press disagrees with the list, the next cohort will be even less willing.

Related

• Same group: L6.10.1 Being able to see the profile is a prerequisite of being able to reset it · L6.10.3 The default ranking after personalization is off is still an editorial choice and must be intelligible · L6.10.4 Deleting one record and resetting everything solve different problems and both must exist · L6.10.5 Still inferring from the current context after personalization is off is undeclared personalization
• Nearby: L6.05 Turning Personalization Off · L6.03 Cold Start · L1.06 Graceful Degradation of AI Failure
• Search terms: reset cost disclosure · what you lose · loss aversion and reset