This guide is for cataloging librarians, heads of technical services, and systems librarians who want to maximize Koha’s copy cataloging capabilities, improve workflow efficiency, and calculate ROI.
Key outcomes for your library:
Original cataloging (manually creating MARC records from scratch) is expensive:
| Library Size | Annual Acquisitions | Cataloging Staff | Avg Time/Item | Cost Pressure |
|---|---|---|---|---|
| Small (50K titles) | 3,000 items/yr | 0.5 FTE | 45 min | Moderate staffing load |
| Medium (200K titles) | 12,000 items/yr | 2.5 FTE | 45 min | High staffing load |
| Large (1M+ titles) | 40,000 items/yr | 8–10 FTE | 45 min | Very high staffing load |
For a library adding 12,000 items/year, original cataloging can consume a substantial amount of cataloger time before you account for:
Real cataloging cost depends on local salaries, staffing mix, match rates, and workflow quality control.
With copy cataloging (importing MARC records from Z39.50, OCLC, or vendors):
| Step | Traditional | Copy Cataloging | Time Saved |
|---|---|---|---|
| Keyword search in catalog | 2–3 min | 2–3 min | 0 min |
| Find matching record | 3–5 min | 1–2 min | 2–4 min ⬇ |
| Verify accuracy | 5–10 min | 1–2 min | 3–8 min ⬇ |
| Edit MARC record | 20–30 min | 2–3 min | 18–27 min ⬇ |
| Add local holdings (item data) | 5–10 min | 5–10 min | 0 min |
| TOTAL | 35–60 min | 8–12 min | 27–48 min saved |
Copy cataloging can significantly reduce cataloging time for many mainstream published materials, but results depend on match rates, record quality, and local editing practice.
First, set up Z39.50 servers in Koha. See our guide: How to Add Z39.50 Servers in Koha.
Essential servers for copy cataloging:
| Server | Database | Coverage | Authority Records? |
|---|---|---|---|
| Library of Congress | LC | All English-language items | Yes (rich) |
| OCLC WorldCat | OLUCWorldCat | 600M+ bibliographic records worldwide | Yes |
| British Library | BLPC | UK & International publications | Yes |
| Your library consortium | Varies | Regional holdings | Maybe |
Pre-copy-cataloging preparation:
1. Export incoming order records from your ILS/acquisitions system
2. Extract ISBN, title, author, publication date
3. Import batch of 10–50 ISBNs into Koha Z39.50 search
(or use Acquisitions → create purchase order → search Z39.50)
4. Match each item to a Z39.50 result
5. Import matches in bulk
6. Manually original-catalog non-matches (typically 5–15% of items)
In Koha’s Cataloging → Import MARC Records:
Step 1: Search Z39.50 Server
└─ Server: "Library of Congress"
└─ Database: "LC"
└─ Query: ISBN = 978-0-201-63361-0
└─ Hit search → Results displayed
Step 2: Review Match
└─ Title: "Design Patterns: Elements of Reusable Object-Oriented Software"
└─ Author: Gamma, Erich (verified as LC Authority record)
└─ Publication: 1994 (vs. new item 2024 reprint — OK)
└─ MARC fields: 050 (LC call number), 082 (Dewey), 100 (author), etc.
Step 3: Import to Catalog
└─ Click "Import this record"
└─ Koha generates new biblio_id in your catalog
└─ Record now searchable by your patrons
└─ You can now add item records (barcode, location, etc.)
For large acquisitions (50+ items), batch import is faster than one-by-one Z39.50 search:
Workflow:
1. Vendor (Amazon, library jobber, etc.) provides order records
└─ File format: MARC, MARCXML, or CSV with ISBN
2. Load vendor MARC file into Koha
└─ Cataloging → Import MARC Records → Upload file
└─ Koha matches ISBN/title against LC/OCLC Z39.50 automatically (if enabled)
└─ Or manually review 20–50 records for accuracy
3. Check for import warnings
└─ Missing 245 (title field)? Record skipped.
└─ Encoding errors? Log entry created.
4. Approve import
└─ Koha inserts all valid records in ~5–10 seconds
5. Add item records
└─ Use batch tool to add items to all imported records at once
└─ Barcode range, location, loan type, etc.
Time savings: Batch import saves ~30 min per 50 items vs. one-by-one entry.
Before importing, use Koha’s MARC Modification Templates to auto-correct common issues:
Example template: Enforce consistent subject headings
Condition: 650 (subject heading) exists AND not tagged with ind2 = "0" (LCSH)
Action: Modify 650 → set ind2 = "0" (force LCSH standard)
Result: All imported subjects now standard Library of Congress Subject Headings
Another example: Auto-populate local item type
Condition: 245 (title) contains "DVD"
Action: Insert 949 $t = "DVD" (Koha item type field)
Result: Items auto-categorize as DVD without manual editing
When you import a MARC record via Z39.50, author/subject headings often link to external authority records automatically. Linking improves:
Koha’s authority linker (Cataloging → Link authorities):
Before: 100 field shows "Shakespeare, William, 1564-1616"
└─ No link to authority record
After: Run authority linking
└─ Koha queries LC authority file (Z39.50)
└─ Finds matching authority record
└─ Biblio 100 field now points to authority
└─ Result: "Shakespeare, William, 1564-1616" is now
a clickable link to all works by Shakespeare
Not every Z39.50 record is import-ready. Librarians must assess quality:
| Factor | Import | Create Original | Hybrid |
|---|---|---|---|
| Match quality | 90%+ accurate | N/A | 70–89% accurate |
| Time available | Yes (routine acquisitions) | No (rush acquisitions) | Yes, needs refinement |
| Authority coverage | LC/OCLC has strong record | Niche/local item | Some authority data exists |
| Language | English, major European langs | Obscure language | Mix of eng + other |
| Publication | Major publishers, recent | Self-pub, very recent | Academic press, older |
| Expected match rate | High for well-covered mainstream materials | Low for unique or local materials | Moderate when records need local refinement |
Decision: When to import vs. create:
IF (ISBN exists AND LC/OCLC search returns exact match AND titles match)
→ IMPORT (copy cataloging)
ELSE IF (Item is rush, patron waiting, no record found)
→ CREATE ORIGINAL (fast minimal record)
ELSE IF (Item is ethnic collection, recent small press, non-English)
→ CHECK multiple Z39.50 servers, or CREATE ORIGINAL
Checklist for each Z39.50 import:
☐ Exact title match? (title, author, edition all correct?)
☐ Publication date reasonable? (reprint vs. new — does it matter for your collection?)
☐ Language correct? (Hebrew book should be in Hebrew MARC, not English translation)
☐ Edition matches your order? (3rd edition vs. 2nd edition?)
☐ ISBN/ISSN matches? (most reliable match field)
☐ Call number reasonable? (LC or Dewey, not junk data?)
☐ Subfields populated? (245 $b, $c, $n properly filled?)
☐ 856 field OK? (if URL included, is it active/relevant to your library?)
If 5+ checks fail → consider creating original instead of importing.
Not all items can be copy-cataloged. Estimate your library’s copy-catalog rate (percentage of acquisitions available for copy cataloging):
| Library Type | Copy-Catalog Rate | Items/Year | Cataloging FTE Needed |
|---|---|---|---|
| University research library | 40–60% (heavy special collections) | 30,000 | 8–10 FTE |
| University general library | 75–85% (mostly mainstream academic) | 15,000 | 2.5–3.5 FTE |
| Public library, large | Often high for mainstream popular materials | 12,000 | 2–2.5 FTE |
| Public library, small | 70–80% (mix of popular + niche) | 3,000 | 0.5–1 FTE |
| Special library | 50–70% (specialized collections) | 2,000 | 1–1.5 FTE |
Example calculation: Medium Public Library
Annual acquisitions: 12,000 items
Copy-catalog rate: 80% = 9,600 items ✓ (via Z39.50)
Original-catalog rate: 20% = 2,400 items ✓ (manual creation)
Time per copy-cataloged item: 10 min = 1,600 hours/year
Time per original item: 45 min = 1,800 hours/year
Total: 3,400 hours/year ÷ 2,000 hrs/FTE = 1.7 FTE
Without copy cataloging (all original):
12,000 × 45 min = 9,000 hours ÷ 2,000 = 4.5 FTE
Capacity difference: Copy cataloging materially reduces the staffing load needed to keep up with acquisitions.
Tier 1: Fast/routine (80% of copy-cataloging) — 8–12 min per item
Tier 2: Complex (15% of copy-cataloging) — 15–25 min per item
Tier 3: Original cataloging (5% of acquisitions) — 45–60 min per item
Before Koha Z39.50 (Traditional copy cataloging from OCLC terminal or vendor):
Setup: OCLC terminal software, training (1–2 weeks)
Annual workflow:
12,000 items × 25 min/item (search OCLC, download, import to ILS)
= 5,000 hours/year
= 2.5 FTE
Other costs to evaluate locally:
- OCLC subscription terms
- Per-transaction or vendor support fees
- Ongoing training and workflow maintenance
After Koha with Z39.50 (copy cataloging through configured Z39.50 sources):
Setup: Koha Z39.50 configuration and staff workflow setup
Annual workflow:
12,000 items × 10 min/item (Z39.50 search in Koha, import, add items)
= 2,000 hours/year
= 1 FTE
Other costs to evaluate locally:
- Hosting and support arrangements
- Whether you still rely on paid external record sources
- Training and quality-control time
Practical takeaway: If your match rate is high and your workflows are disciplined, copy cataloging in Koha can materially reduce staff time per item compared with heavier manual workflows.
If you want a cost-per-item model, build it from your own local inputs:
The time-per-item improvement can be significant, but the dollar outcome should be modeled from your own staffing and vendor assumptions rather than from a generic public estimate.
Scenario: Cataloging team with 20% budget cuts
Before Koha:
Team size: 2.5 FTE catalogers
Can process: 12,000 items/year
If we cut to 2.0 FTE: Can only process 9,600 items/year
Result: 2,400 item backlog, community unhappy
After Koha:
Team size: 1.0 FTE cataloger (+ new hire for other duties)
Can process: 12,000 items/year (10 min each)
Additional capacity: Can now handle 7,200 additional items/year
Result: No backlog, increased service level, same budget
ROI Insight: With copy cataloging, Koha can increase staff capacity without requiring the same level of manual cataloging effort.
For a multi-year ROI model, use your own local assumptions for:
That approach produces a defensible internal business case without relying on generic public savings figures.
Plus intangible benefits:
When the Library of Congress doesn’t have a record, search your backup servers automatically:
Koha’s Z39.50 multi-search:
1. Query Library of Congress LC database
2. If no match (0 results):
├─ Search OCLC WorldCat (larger, may catch international items)
├─ Search British Library (for UK/European items)
└─ If still no match, create original record
Result: 92–97% copy-catalog rate (vs. 70–80% with single server)
DVDs, audiobooks, maps, electronic resources: Most available on Z39.50, but require specific search techniques.
DVD Example:
Standard search: ISBN + title → Often fails (DVDs have ISBNs but inconsistently cataloged)
Better search: Title + "DVD" + year → Better match
Best search: Use OCLC WorldCat, search for "DVD" specifically in format field
E-book Example:
Standard search: ISBN → May retrieve print version, not e-book
Better search: Title + "[electronic resource]" (standard MARC phrase)
Search server: OCLC WorldCat has best e-resource coverage
Sometimes the Z39.50 record is 70–80% good, not perfect. Hybrid cataloging blends import + editing:
Z39.50 import result:
245 10 |a Design patterns : |b elements of reusable
object-oriented software / |c Erich Gamma ... [et al.].
Your library's local practice:
- Add subject: |a Computer programming |x Patterns
- Add 856 field: Publisher's web page URL
- Modify call number from LOC to Dewey (your library uses Dewey)
Hybrid workflow:
1. Import Z39.50 record
2. Add 650 (subjects) for your collection
3. Fix call number (change 050 to 082)
4. Add local 590 note: "Donation from Smith family, 2026"
5. Save and done (15 min total, vs. 45 min for original)
| Metric | Before Koha | After Koha | Target |
|---|---|---|---|
| Avg. time per item | 25 min | 10 min | <12 min |
| Copy-catalog rate | 75% | 85% | >85% |
| Items per cataloger/year | 4,800 | 12,000 | >12,000 |
| Cost model | Higher manual labor load | Lower manual labor load | Improve local efficiency |
| Order-to-shelf time | 14 days | 7 days | <10 days |
| Authority-linked records | 60% | 95% | >90% |
| Patron search accuracy (by click-through) | 65% | 82% | >80% |
Create a simple dashboard for management:
Monthly Cataloging Report (May 2026)
────────────────────────────────────
Items acquired: 850
Copy-cataloged (Z39.50): 721 (84.8%)
Original-cataloged: 129 (15.2%)
Avg. time per copy item: 10.2 min
Avg. time per original item: 44 min
Workflow notes:
Copy-cataloged volume remained high
Original cataloging stayed focused on exceptions
Time per item stayed within target range
On track for annual goal: 10,200 items ✓
Track how copy cataloging affects user experience:
Search success rate:
Before: "I searched but found only 3 results" (low coverage)
After: "I found it!" (95% of items searchable within 48 hrs of acquisition)
Order-to-shelf time (patron perspective):
Before: "I placed a hold 3 weeks ago, still waiting"
After: "Hold arrived in 1 week!"
Browse experience:
Before: Partial authority linking (60%) → confusing name variations
After: Full authority linking (95%) → all Shakespeare works linked
| Problem | Cause | Solution |
|---|---|---|
| Z39.50 search returns 0 results | ISBN not in LOC database (e.g., very new book, self-published) | Try OCLC WorldCat server; or create original record |
| Multiple conflicting records found | Similar ISBN/title; ambiguous query | Refine search: add publication year, author |
| Imported record has encoding errors | MARC source uses non-UTF8 encoding | Use Koha’s character set converter before import |
| Call numbers appear wrong | Record has LC classification, library uses Dewey | Use MARC Modification Template to auto-convert 050→082 |
| Authority records not linking | Koha’s authority linker not enabled | Enable in Administration → System Preferences → Authorities |
| Items not appearing in catalog | Items added but not linked to bib record | Verify item’s biblio_id matches the MARC record ID |
Advanced libraries are using machine learning to auto-catalog:
Koha’s REST API allows programmatic cataloging:
# Pseudo-code: Import 500 ISBNs from vendor file
FOR each ISBN in vendor_file:
result = OCLC_Z39.50_search(ISBN)
IF result.quality_score > 0.9:
import(result.marc_record)
ELSE:
queue_for_manual_review(result)
Result: Routine items auto-cataloged overnight; staff reviews only exceptions.
Once you’ve linked records to authorities, maintain them:
Copy cataloging with Koha’s Z39.50 integration transforms your cataloging department:
✓ Reduces time from 45 min to 8–12 min per item (73–82% faster) ✓ Can reduce per-item processing effort when match quality is strong and staff workflows are consistent ✓ Scales staff to handle 2–3x more items without hiring ✓ Improves quality via authority linking and validated MARC records ✓ Frees time for special collections, rare books, and original research materials
Not on Koha yet? KohaSupport’s Standard plan runs Koha on your own AWS account with Z39.50 pre-configured, automated backups, and the Koha Theme Builder included. Start free →
Next steps: