Koha implementation guide: phases, timeline and what to prepare

What a Koha implementation covers

Koha is an open-source integrated library system that handles cataloging in MARC21 or UNIMARC, circulation, acquisitions, serials, patron management, reports and a public catalog (OPAC). If you are still evaluating the platform itself, start with what Koha is. This guide assumes you have chosen Koha and need to put it into service.

A complete implementation has eight workstreams. Not every library needs all of them at the same depth, but each one should appear in the scope, even if only to record that it is excluded:

• Assessment of the collection, the current data and the library's workflows.
• Installation of Koha on your own servers or on managed cloud hosting.
• Configuration of parameters, patron categories, item types and circulation rules.
• Data migration of bibliographic records, items, patrons and active circulation.
• OPAC customization with your institution's visual identity.
• Integrations with authentication, self-check machines and campus systems.
• Training of staff by role.
• Go-live and support, with a planned cutover and a support contract afterward.

The Koha implementation process, phase by phase

1. Assessment

The assessment documents your starting point: the exact name and version of your current system, how many bibliographic records, items and active patrons you have, the description format (MARC21, UNIMARC or a proprietary structure), the character set, the number of branches and every system Koha must talk to. It ends with a written scope that states timeline and cost. Most budget surprises later in a project come from questions nobody asked at this stage.

2. Installation

Koha runs on Linux with MariaDB or MySQL and a search engine, Zebra by default or Elasticsearch for large catalogs, so there are no operating-system or database licenses to buy. You can deploy on your own infrastructure or on managed hosting; the software is identical and you can move between models later. The public catalog should live on a subdomain of your institution, served over HTTPS. Our guide to Koha hosting covers server sizing and the hosting decision in detail.

3. Configuration

Configuration is rebuilt, not migrated. It includes MARC frameworks, item types, collection codes, authorized values, branches, holiday calendars, patron categories, circulation and fine rules, staff permissions and the notices Koha sends by email or SMS. In an institution with three branches, rebuilding these parameters can take as long as loading the data. Libraries with several branches can run them all in one instance, with shared or separate rules.

4. Data migration

Migration moves five data sets: bibliographic records, items, authorities, patrons and circulation. Koha imports bibliographic and authority records in MARC (ISO 2709 or MARCXML) and patrons through a CSV import. Every serious migration runs a pilot load, adjusts the field mapping and validates counts before the final load. It is the phase that varies most in duration and cost, so it has its own guide: how to migrate to Koha.

5. OPAC customization and integrations

The OPAC is responsive and can carry your institution's branding. Applying your visual identity is quick; redesigning the search experience is a development project, best left to a second phase. On integrations, Koha exposes Z39.50 and SRU for copy cataloging and federated search, SIP2 for self-check machines and security gates, OAI-PMH for metadata harvesting, and ILS-DI plus a REST API for custom integrations. Patron accounts can be synchronized with LDAP, Active Directory, CAS, SAML or OpenID Connect, or loaded on a schedule from your student information system.

6. Training by role

One general session does not work. Effective training separates technical services (cataloging, authorities, subject analysis), circulation and public services, acquisitions and serials, and system administration for IT. It happens on your own instance, with your own catalog, and leaves manuals and recordings behind for staff who join later.

7. Go-live and support

The switchover follows a fixed sequence: freeze cataloging and circulation in the old system at an agreed date and time, load the records captured since the last export, point your domain to Koha and open the OPAC. Keep the old system available in read-only mode for at least a month; it costs little and settles questions. The first weeks deserve close monitoring, after which the library moves to annual support with planned upgrades. See what Koha support includes.

How long a Koha implementation takes

A typical academic library takes three to five months from assessment to go-live. Without a historical catalog to migrate, installation and configuration usually take six to eight weeks, and a small library with clean data can be operational in a few weeks. When there is a migration, the timeline depends far more on data quality and on the number of branches than on the number of records:

These ranges assume an export in a standard format and library staff taking part in validation. Each additional branch adds its own configuration, circulation rules, patrons and validation.

Decisions to make before you start

Projects move fastest when the library arrives with these decisions made, or at least assigned to someone:

1. Hosting model: your own servers, managed cloud, or your own server administered remotely by the provider.
2. Circulation policies: loan periods, renewals, holds, fines and patron categories, written down.
3. Bibliographic format: MARC21 or UNIMARC, and what happens to records that arrive in the other format.
4. Search engine: Zebra for most catalogs; Elasticsearch for large collections or heavy concurrent use.
5. Circulation history: migrating active loans and holds is standard; the full history rarely justifies its conversion cost.
6. A responsible counterpart: one person in the library who validates test loads without delay.
7. Phasing: starting with one branch and the standard OPAC, then adding branches, integrations and redesigns.
8. Deliverables at closing: manuals, a migration report with verified counts and administrator credentials.

Patron privacy and multilingual catalogs

Patron data stays on the infrastructure your institution chooses, under its own data policy. Koha can anonymize circulation history automatically after a set period, give each staff profile granular permissions and keep an audit log of operations. Combined with institutional policies, those features help libraries meet data-protection laws such as the GDPR or their national equivalent, which matters most for reading history.

Koha is translated into dozens of languages, and the staff interface and the OPAC are configured separately. Several languages can be enabled at once in the public catalog so each user picks one, which is common in universities with international programs or many international students.

Common Koha implementation mistakes

• Choosing on license price alone. Check that your data can leave any system in a standard format. Records trapped in a proprietary format turn a future change of provider into a forced project.
• Underestimating training. A team that does not master reports and circulation rules uses a fraction of what Koha offers.
• Ignoring upgrades. Ask who applies new versions and whether they cost extra; some providers charge for every upgrade, which over time can exceed the initial project.
• Incomplete configuration. Patron categories or loan rules that were never replicated tend to surface on the first day of the semester.
• Broken characters. An undeclared character set (MARC-8, ISO 8859-1 or UTF-8) breaks accented letters; catch it in the pilot load.
• Not freezing the old system, which leaves two diverging catalogs for weeks.

AI-assisted cataloging from day one

Cataloging is usually the slowest process in a library, and a new Koha instance is the right moment to change that. Every Koha implementation by BiblioLabTech includes BiblioScan AI at no extra cost. The cataloger photographs the cover, the title page and the table of contents; BiblioScan AI reads them and fills in a draft MARC21 record inside Koha, which the cataloger reviews, corrects and saves. Copy cataloging over Z39.50 remains the first option when a good record exists elsewhere, and AI covers the material nobody else has cataloged, such as theses, local publications and donations. See how AI cataloging works.

Implementing Koha with BiblioLabTech

BiblioLabTech delivers Koha implementations end to end, from the first assessment to the last upgrade, so the library deals with one team throughout. We migrate from commercial and open-source systems, including Aleph, Symphony, PMB and ABCD, and deploy on your servers or on managed hosting. The Escuela Colombiana de Ingeniería Julio Garavito runs its Koha catalog with BiblioScan AI cataloging on a dedicated server administered by our team; read the case study.

We work remotely with libraries in any country, from a UTC−5 base that overlaps the full working day across the Americas and the afternoon in Europe and Africa. For budget ranges, see Koha implementation cost; to scope your project, visit our Koha service or tell us about your library.

Koha is free software developed and maintained by its international community at koha-community.org.