Inactive cleanup balances storage cost against later retrieval

What it is

After archive, objects still cost storage. Further on, a product may truly delete long-unopened content, or move it to storage that barely opens. A cleanup policy has to write both sides: what is saved (capacity, money), and whether people can still get it, how long they wait, whether they can export. This is destruction or deep cold, not moving active items to archive. Auto-archive notice answers why it left the main list; cleanup answers whether it can come back. One silent countdown for both makes dormancy and death the same event.

Why it happens

Storage costs; retrieval has a long tail: tax, disputes, nostalgia years later. Optimizing only cost cuts the tail, with no evidence when something breaks. Never cleaning bloats search and backups, and daily locate suffers. The balance is visible layering: archive first, then cold storage, delete last; each hop has a different window and reversibility. People decide from the window number and “can I export before delete,” not from “we value your data.” Cleanup by last-opened also treats read-only evidence as idle. Cleanup by last-edited kills published manuals that no longer change. The policy must name which inactivity definition it uses, and allow an object to be marked Keep.

Studying it

Give material tagged “might need this in a few years,” and show different cleanup windows with or without export. Watch external backups, Keep marks, and whether the window is accepted.

Independent variables: window length, export-before-delete, inactivity definition (opened / edited / referenced), whether a Keep mark exists. Dependent variables: external backup rate, irreplaceable items purged, ability to restate the window.

A lab cannot wait real years; use a scenario: “a tax audit in three years wants this file.” Do not treat unsearchable archive as cleanup—the object is still there. When a plan upgrade changes the window, measure whether the new rule is restated.

Where it stops holding

Legal retention forbids cleanup inside the required period. A short window on a free personal plan is acceptable if stated; people can choose not to keep the only copy here. Content referenced by live objects (embedded images, templates still in use) will break actives if cleaned; exempt them or break references first. User-initiated delete is not this policy. Cold storage that takes hours to open must say the wait up front, or it looks corrupt.

Applying it

• Write the layers: how long after archive until cold storage, how long until delete; whether each hop undoes, whether export exists.
• Name how inactivity is computed. Offer Keep, do not auto-clean.
• Before delete, offer export and one last notice. Referenced objects are exempt by default.
• Verify: ask whether a contract unopened for three years is still there, and how to get it. No window, or a belief that archive means instantly openable storage, means the human side of the balance failed. Mark one item Keep and confirm it stays out of the cleanup queue.

Related

• Within the group: H8.14.1 Lifecycle stages from create to archive need explicit names · H8.14.2 Archive is retrievable recovery, not deletion · H8.14.3 Auto-archive rules must be announced before items vanish
• Adjacent: H8.06 Version History · H3.08 Soft Delete and Trash · H6.14 Account Deletion and Data Erasure
• Search terms: inactive purge · cold storage · retention versus retrieval