Before Storage: Writing a Media Archive Keep-or-Discard Policy
Most media archives do not fail because of storage. They fail because nobody decided, in advance, what the archive is for and what belongs in it. Files pile up, duplicates multiply, and eventually a creator or small team faces a drive full of material that no one can confidently describe. A keep-or-discard policy solves this by turning vague intentions into written rules you can apply file by file. This article walks through how to write one, which questions to answer first, and how to test the policy before it governs anything important. The approach is general and applies whether your archive holds photos, video clips, audio, or mixed formats.
Start With the Archive's Purpose
A policy needs a reason to exist. Before deciding what to keep, write one or two sentences describing what the archive is meant to support. Perhaps it exists to let you reuse past material in new projects, to preserve a record of your public work, or to recover original files if a published version is damaged or removed. Each purpose implies different keep decisions. An archive built for reuse favors high-resolution masters and well-labeled clips. An archive built as a record may prioritize completeness and dates over polish.
Write the purpose in plain language and put it at the top of the policy document. When a future question arises, such as whether to keep a dozen near-identical preview images, the purpose gives you a basis for the answer rather than a guess made under deadline pressure.
Define Categories Before Judging Individual Files
Individual judgments are slow and inconsistent. Categories make decisions repeatable. A practical set of categories might include masters (original captures or unedited source files), finals (the versions you published or delivered), working files (edits, layers, drafts, and project files), previews (compressed copies used for review or sharing), and duplicates (exact or near-exact copies). Each category gets a default rule. For example, masters are kept by default, duplicates are discarded unless they carry different metadata, and previews are kept only when no higher-quality version exists.
Keep the category list short. Five to seven categories is usually enough. If you find yourself creating a category for a single file, merge it into an existing one and note the exception in the policy.
Write Explicit Keep and Discard Criteria
Each category needs criteria that someone else could apply and reach the same result. Replace words like good or important with observable tests. For instance: keep any master whose capture date falls within the archive's stated range; keep a final if it appears in a published context you can document; discard a preview when a master of the same frame exists; discard a file that fails to open in two separate viewers and has no recoverable copy elsewhere.
Include a section for uncertain cases. Some files will not fit any rule cleanly. Give them a holding status, such as review-later, and set a date by which the holding status must be resolved. Without this, uncertain files accumulate indefinitely and the archive grows without governance.
A Checklist for Testing the Policy
Before applying the policy to the whole archive, test it on a small, representative sample. Pull roughly fifty files spanning every category and at least two time periods. Apply the rules and record each decision along with the reason. Then ask a reviewer, even if that reviewer is you returning after a break, to apply the same rules to the same sample. Compare results.
Use this checklist during the test: Does each category have a default rule? Can each rule be applied without asking the policy author? Are there files that fall between categories, and does the policy say what to do with them? Does any rule depend on information the archive does not actually record, such as a capture date missing from metadata? Are discard decisions reversible for a defined period, or are they final? Each no or unclear answer is a revision point. Expect to revise the policy two or three times before it is stable.
Questions to Ask Before You Discard Anything
Discarding is the step that cannot be undone, so build a short pause into the process. Before deleting any file, ask whether another copy exists in a location you control, whether the file is referenced by a project or published page you still maintain, whether the file contains metadata or annotations not present elsewhere, and whether the policy explicitly authorizes its removal. If any answer is unclear, move the file to a quarantine folder rather than deleting it.
Set a quarantine period of a few weeks or months, long enough that you will notice if you needed something. Only after that period should permanent removal occur. This habit costs little and prevents the most common archive regret: discovering too late that a discarded file was the only copy of something needed.
Keeping the Policy Current
A policy written once and never revisited slowly drifts away from the archive it governs. Schedule a short review at a fixed interval, perhaps every few months, and check three things: whether the stated purpose still matches how the archive is actually used, whether any category has grown so large that it needs splitting, and whether the quarantine folder contains files that should now be resolved. Record each change with a date so the policy's history is visible.
A keep-or-discard policy is not a one-time cleanup. It is a small document that makes future decisions faster and more consistent. Written clearly, tested on a sample, and reviewed on a schedule, it turns a growing pile of files into an archive you can trust.