Resources & Guides

What MARCReady Can Fix — and What a Librarian Should Still Review

Understand what MARCReady fixes automatically, what Kai suggests for review, and which MARC decisions still need librarian judgement.

May 19, 2026 6 min read

What MARCReady Can Fix — and What a Librarian Should Still Review

MARCReady is designed to help libraries identify and repair common catalogue data problems. It can save time, reduce migration risk, and make messy exports easier to review.

But MARC repair is not only a technical process. Some decisions depend on local cataloguing policy, collection context, and librarian judgement.

This article explains the difference.

How MARCReady works

MARCReady uses two layers of processing:

Rule-based autofix — a deterministic engine that applies known MARC21 specification rules. These fixes are applied automatically because the correct behaviour is unambiguous.

Kai (AI mapping and review) — for structured data (CSV, TSV, Excel, JSON), Kai proposes field-to-MARC mappings for your review. Kai also reviews MARC sources for complex or ambiguous issues and provides per-field suggestions with confidence levels. Kai suggestions are not applied until you accept them.

What the rule-based autofix handles

MARCReady’s rule engine applies these fixes automatically:

Invalid indicators — approximately 100 MARC field tags have their indicators corrected to match the LC MARC21 specification. For example, 245 ind1 is set based on whether a 1XX main entry is present; 245 ind2 is recalculated from the nonfiling article in the title.
ISBN-10 to ISBN-13 conversion — bare ISBN-10 values in 020 $a are converted to ISBN-13; the original ISBN-10 is moved to $z.
ISBN qualifier separation — text qualifiers mixed into 020 $a (such as “paperback” or “hardcover”) are moved to $q.
Empty subfield removal — subfields with empty or whitespace-only values are stripped.
Non-repeatable field deduplication — fields such as 001, 010, 100, 245, 250, and 300 that should appear only once are deduplicated; extra occurrences are removed.
Leader sanitisation — invalid values at key Leader positions (record type, bibliographic level, encoding level) are replaced with safe defaults.
260 to 264 conversion (RDA) — legacy 260 publication fields are converted to 264 with ind2=1.
336/337/338 generation (RDA) — content type, media type, and carrier type fields are generated from the Leader record type if absent.
245 $h [GMD] removal — bracketed general material designators such as [electronic resource] are removed from 245 $h per RDA practice.
650/651/655 ind2 correction — if ind2=7 (source specified in $2) is used but no $2 is present, ind2 is changed to 0 (LCSH).
856 indicator correction — 856 indicators are set to 4 (HTTP) and 0 (online resource).
264 ind2 default — a blank 264 ind2 is set to 1 (publication).

Advanced Repair: replacing a record from Library of Congress

When a record is too damaged or incomplete to repair field by field, Advanced Repair provides a different option. It uses the record’s ISBN or LCCN to query the Library of Congress catalogue directly via SRU. If a match is found, the record’s standard fields are replaced with the LoC authoritative version. Local holdings fields (852, 942, 9XX) are preserved.

If LoC returns no result, MARCReady falls back to Open Library.

Advanced Repair is a per-record action — you trigger it on specific records that need it. It is most useful when:

a record is so garbled that incremental repair is impractical;
the record has a valid ISBN or LCCN that identifies the work;
you want the authoritative LoC cataloguing for a specific title.

This gives you the same bibliographic data as a Z39.50 copy cataloguing lookup, without a separate Z39.50 client.

ISBN validation against external databases

MARCReady can also validate each record’s primary ISBN against Google Books and Open Library, checking title similarity, author match, page count, and publication year. Records where the ISBN does not match external data are flagged — useful for catching ISBNs that were incorrectly assigned or copied from a different edition.

What Kai suggests for review

For CSV, TSV, Excel, and JSON files, Kai proposes a mapping from each column or field to a MARC tag and subfield. These suggestions appear in the review interface before any records are exported.

For MARC sources, Kai may additionally suggest:

mapping ambiguous or incorrectly tagged fields;
updating 300 physical description to RDA terminology (e.g. “p.” to “pages”);
populating 008 date positions from 264 or 260 data;
subject heading quality notes;
records that appear structurally unusual and warrant manual review.

All Kai suggestions show a confidence level (high, medium, or low). You accept or skip each suggestion before export.

What a librarian should still review

Some MARC decisions are not purely mechanical.

A librarian should still review:

subject heading quality;
local cataloguing conventions;
authority heading choices;
material type coding;
whether a record represents the correct edition;
local 9XX fields;
item type mapping;
branch and location mapping;
whether duplicate candidates are truly duplicates;
whether a vendor record is complete enough for local use.

MARCReady can highlight issues and apply rule-based fixes, but it should not replace professional judgement.

Examples of human-review decisions

Are two similar records truly the same edition?

Two records may share an ISBN or title but represent different editions, formats, or manifestations. MARCReady can flag them as potential duplicates, but only a librarian can confirm.

Should a local field be kept?

A 9XX field may look unusual but contain important local migration data. Do not delete it without understanding its purpose.

Is the material type correct?

A record may need review if the Leader and 008 values do not match the physical item.

Is item data mapped correctly?

MARCReady can help identify item data, but the destination Koha branch, item type, and location codes must match your actual Koha configuration.

Recommended workflow

Upload a sample file.
Review the autofix results.
Accept, adjust, or skip Kai suggestions.
Export a small test file.
Stage it in Koha.
Check staff and OPAC display.
Process the full file once the workflow is trusted.

Try the free preview

Use the free preview to see what MARCReady finds in your own records. Review up to 3 records per upload at no cost (up to 15 records per month). It is the best way to judge whether the repairs match your catalogue data and migration goals.