Catalog Cleanup By Norva Editorial Team · 18 July 2026 · 7 min read

The Complete Catalog Cleanup Plan for a Personal Media Library

A safe catalogue cleanup audits the current state, protects recovery, defines rules, pilots difficult cases, reconciles every batch, and measures retrieval afterward.

In short: Clean in seven phases: audit, protect recovery, define scope and rules, triage exceptions, pilot a representative batch, reconcile each change set, and establish maintenance. Do not begin with bulk deletion or renaming. Preserve source, version, language, and ownership context, and verify that essential retrieval tasks still work after every batch.

Catalogue cleanup should reduce confusion, not merely reduce visible rows. A smaller catalogue can be worse if useful versions, source context, or household state disappear.

Phase 1: audit the current catalogue

Record:

in-scope authorised sources and owners;
broad media groups and approximate counts;
category list and item counts;
uncategorised queue;
empty categories;
possible duplicates;
multi-version works;
missing or conflicting metadata;
unavailable items;
active filters and profile-specific state;
known source changes.

Do not correct during the audit. Preserve a baseline with the pre-cleanup audit guide.

Phase 2: protect recovery

Identify supported exports, snapshots, backups, or change histories available for catalogue data and any files the user is authorised to manage. Confirm what each recovery method contains and perform a small restoration or rollback test where safe.

The Library of Congress personal archiving material recommends identifying important content, organising it, and keeping copies in different places. The exact method depends on the source and rights. A catalogue export is not automatically a media backup.

Use the reversible cleanup framework before broad edits.

Phase 3: set cleanup rules

Define:

what counts as a confirmed duplicate;
title and category conventions;
minimum metadata;
language field meanings;
work/version relationships;
active, archive, unavailable, and review states;
actions an executor may perform;
actions requiring approval;
rollback and review triggers.

Version the rules. Do not change them silently halfway through a batch.

Phase 4: triage exceptions

Use four destinations:

ready for rule-based cleanup;
needs metadata evidence;
needs source or rights review;
no change under current scope.

Uncategorised items need a documented triage path, not a miscellaneous category that becomes permanent. Apply the uncategorised media triage method.

Phase 5: pilot difficult cases

Choose a batch containing:

normal items;
duplicate candidates;
several sources;
a version group;
multilingual media;
an unavailable item;
an archive candidate;
empty and low-use categories.

Run the rules on the pilot. Record every transformation and exception. Test search, browse, versions, language evidence, profile state, and return navigation where relevant.

Phase 6: execute and reconcile batches

For each batch, compare:

expected and processed items;
changes completed;
exceptions created;
categories added, merged, or removed;
metadata values transformed;
source and version relationships preserved;
rollback checkpoint;
retrieval tests.

Stop if the ledger does not reconcile. Large batches make errors harder to locate.

Phase 7: maintain the result

Set intake rules, an exception owner, category review, archive review, source-change triggers, and a seasonal success check. Cleanup without maintenance recreates the same backlog.

When empty categories remain after reconciliation, use the empty category removal guide and confirm that no hidden, filtered, or unavailable items depend on them.

Original evidence: seven-phase cleanup ledger

Phase	Entry condition	Evidence produced	Pass condition
Audit	Scope agreed	Baseline	Counts and issues recorded
Protect	Baseline complete	Recovery test	Rollback understood
Rules	Recovery ready	Versioned rulebook	Edge cases defined
Triage	Rules approved	Exception queue	Every item routed
Pilot	Sample selected	Change set	Essential tests pass
Execute	Pilot approved	Batch ledger	Reconciles
Maintain	Cleanup closed	Cadence	Owner and triggers active

The ledger is a planning artefact. It does not claim that a cleanup was performed until the household records actual evidence.

Common mistakes and limitations

Starting with deletion.
Treating matching titles as proven duplicates.
Changing rules during a batch.
Ignoring profiles, versions, and languages.
Calling a catalogue export a media backup without verifying it.
Hiding unresolved items to make counts look clean.
Removing categories before checking filters and source state.
Completing cleanup with no intake cadence.

Norva can organise compatible authorised sources and group variants, but it cannot create missing source metadata or resolve rights and ownership decisions.

Frequently asked questions

How much should be cleaned in one batch?

Use the smallest batch that includes the relevant edge cases and can be inspected, reconciled, and rolled back within one review session.

Should duplicates be deleted immediately?

No. First confirm the same work and comparable version, source, language, rights, and availability. Merge or remove only through an approved reversible rule.

How do I know cleanup is complete?

The agreed scope is processed, exceptions are owned, batches reconcile, essential retrieval tasks pass, and maintenance has a named owner and cadence.