| Platform | Licensing model | Citation style coverage | Word/LibreOffice plugin | Group/shared libraries | Data portability | Best for |
|---|---|---|---|---|---|---|
| Zotero | Free; paid cloud storage tiers only | Thousands of CSL styles, community-maintained | Yes, Word, LibreOffice and Google Docs | Yes, unlimited free group libraries | Full export (RIS, BibTeX, CSL JSON); no lock-in | A faculty standardising without a central licence budget |
| EndNote | Paid; campus-wide licences common | Very broad, vendor-maintained style list | Yes, deep “Cite While You Write” integration | Yes, via EndNote Online shared libraries | Exportable but proprietary library format adds friction | A large faculty with an existing Clarivate/EndNote site licence |
| Mendeley | Free; paid storage add-ons (Elsevier) | Broad CSL-based style coverage | Yes, Word plugin | Yes, shared groups with size limits on free tier | Exportable (RIS, BibTeX); reasonable portability | A department already inside the Elsevier/ScienceDirect ecosystem |
| RefWorks | Subscription, typically library-licensed institution-wide | Broad, vendor-maintained | Yes, Word plugin (web-based core product) | Yes, shared folders across an institutional account | Exportable; fully cloud-dependent, no offline library | An institution whose library already manages a ProQuest subscription |
None of these four platforms is a bad citation manager on pure functionality; the meaningful institutional differences are licensing cost, data portability and how much central IT and library support each one demands once a faculty tries to standardise on it rather than leaving every student to pick their own.
Why standardise on one reference manager at all?
Where a faculty leaves reference-manager choice entirely to individual students, three costs accumulate quietly: the library and writing centre end up supporting four different tools with four different troubleshooting playbooks, supervisors cannot easily share a group bibliography with a student using a different platform, and a citation-style change — a journal or department switching from APA 7 to Vancouver, for example — has to be explained four different ways rather than once. Standardising does not mean banning alternatives outright; it means the faculty picks a default it actively supports, trains students on, and builds departmental style guidance around, while still allowing an experienced student with a strong reason to use something else.
Zotero: the strongest default where there is no dedicated budget line
Zotero is free, open source, and exports cleanly to RIS, BibTeX and CSL JSON, which means a faculty standardising on it is never locked into a proprietary library format it cannot walk away from later. Its style coverage draws on the Citation Style Language repository, a community-maintained set of thousands of styles that covers the overwhelming majority of journals and institutional formatting requirements a health, business or humanities faculty will encounter. Group libraries — a shared collection a supervisor and their students all draw from — are unlimited on the free tier, which is the feature that matters most for a faculty-wide rollout rather than an individual user decision. The trade-off is support: because there is no vendor account manager or paid institutional licence, first-line troubleshooting has to live with the library or writing centre rather than a vendor helpdesk.

EndNote: the incumbent in institutions with an existing site licence
EndNote, from Clarivate, is the platform most likely to already have a campus-wide licence sitting in a library’s existing budget, particularly at larger or older-established institutions, which makes the marginal cost of standardising on it effectively zero if that licence is already in place and under-utilised. Its “Cite While You Write” Word integration is mature and its style list is broad and vendor-maintained rather than community-maintained, which some institutions prefer for a perceived higher bar of quality control. The friction is on the way out rather than the way in: EndNote’s library format is proprietary, and while export is possible, a faculty that later wants to move away from EndNote faces more migration friction than one standardised on Zotero from the start.
Mendeley and RefWorks: fit is about existing vendor relationships, not raw features
Mendeley, owned by Elsevier, is a reasonable default specifically where a faculty already sits inside the Elsevier ecosystem through ScienceDirect or Scopus subscriptions, since the platforms share account infrastructure and some cross-linking. RefWorks, from ProQuest, is entirely cloud-based with no offline desktop library, which suits an institution whose library already manages a ProQuest subscription and wants one vendor relationship rather than several, but it is a genuine limitation for any student who needs to work with their reference library without a reliable internet connection. Neither platform has a compelling functional advantage over Zotero or EndNote on citation-style coverage or Word integration; the decision for both is really about which vendor relationship the library already has, not which tool cites better.

The recommendation
Zotero is the strongest default for a faculty standardising without an existing site licence to build around: it is free, exports cleanly with no lock-in, supports unlimited group libraries, and covers citation-style needs through the CSL repository as comprehensively as any paid alternative. The named alternative is EndNote, specifically for a faculty or institution that already holds an active, under-utilised campus-wide licence — in that specific case, the marginal cost of standardising is close to zero and the vendor-maintained style list and mature Word integration are genuine advantages worth using rather than letting the existing licence go to waste while students independently drift toward Zotero anyway.
Where does the data actually sit, and does that matter?
All four platforms sync a working library to the cloud by default, which puts reference-manager choice inside the same data-residency conversation that already covers other edtech procurement. Zotero’s free-tier storage and its paid add-on both run on cloud infrastructure the institution does not control the region of by default, though the desktop application can run entirely offline with sync disabled for an institution that wants zero cloud dependency. EndNote and RefWorks are vendor-hosted through Clarivate and ProQuest respectively, with data processing terms set by those vendors’ standard institutional agreements. Mendeley runs on Elsevier infrastructure. For most faculties, a reference library of citation metadata — titles, authors, publication data — is a low-sensitivity data category compared with the full-text student writing covered in where student text is actually processed, but a faculty with an unusually strict data-residency policy should confirm this explicitly with whichever platform it standardises on rather than assuming citation metadata is automatically out of scope.
What standardisation actually requires beyond picking a tool
Choosing a platform is the easy half of this decision. A faculty rollout also needs a training session built into research-methods or induction teaching rather than left to self-service documentation, a departmental style guide specifying which CSL style or output style maps to which assessment type, and a named first-line support contact — usually the library, since IT rarely has the citation-style expertise to troubleshoot a formatting problem. Where the faculty is also standardising its broader thesis formatting expectations, this sits naturally alongside the documentation approach set out in how to write a departmental writing standard for theses, and the citation-accuracy question a reference manager partially but not fully solves is covered in more depth in what academic writing tools actually check — a reference manager formats citations correctly against a chosen style; it does not verify that the underlying source actually says what it is cited for, which remains a human check regardless of which platform a faculty standardises on.
If you would like help mapping a reference-manager standardisation plan against your own faculty’s existing licences and citation-style requirements, request an institutional evaluation and we will work from your current library contracts rather than a generic recommendation.
Rolling out a new default without disrupting current students
A faculty switching its default reference manager mid-cycle should not migrate existing students already deep into a thesis or dissertation, since forcing a library migration late in a project introduces risk with no real benefit to that student. The lower-disruption approach is to set the new default for incoming cohorts only, run both tools in parallel for one or two academic years, and route support requests for the legacy tool to a documented but no-longer-actively-promoted help page rather than withdrawing support abruptly. This is the same phased-rollout discipline set out more generally in how to run a departmental pilot of an AI writing tool, and it applies just as directly to a reference-manager transition as to a new AI writing platform: a cohort-based cutover, not a hard switch, is what keeps a standardisation decision from becoming a support crisis in its first term.
Frequently asked questions
Is Zotero really free for institutional use, with no hidden licensing cost?
Yes, the software itself is free with no per-seat or per-institution licence fee. The only paid component is additional cloud storage beyond the free tier’s allowance, which most individual users and group libraries do not need to exceed for text-only reference libraries.
Does standardising on one reference manager mean banning the others?
No. Standardising means the faculty picks and actively supports one default, trains on it, and builds style guidance around it, while still allowing an experienced student with a good reason to use an alternative.
Which platform is easiest to migrate away from later?
Zotero, because its export formats (RIS, BibTeX, CSL JSON) are open standards rather than a proprietary library format. EndNote is exportable but historically carries more migration friction due to its proprietary library structure.
Why would a faculty choose EndNote over the free Zotero?
Primarily where the institution already holds an active, under-utilised campus-wide EndNote licence, making the marginal standardisation cost effectively zero, or where the institution specifically values a vendor-maintained rather than community-maintained citation style list.
Does a reference manager check whether a citation is actually accurate?
No. All four platforms format a citation correctly against a chosen style once the source is entered, but none verify that the cited source actually supports the claim it is attached to — that remains a manual check.
Is RefWorks usable without an internet connection?
No. RefWorks is entirely cloud-based with no offline desktop library, which is a real limitation for a student working without reliable connectivity, unlike Zotero, EndNote or Mendeley, which all maintain a local library.
Who should provide first-line support once a faculty standardises?
Usually the library, since citation-style troubleshooting is a subject-expertise problem more than an IT problem, and library staff are typically the ones already fluent in the relevant style guides.
Does group library size matter for a faculty rollout?
Yes. Zotero offers unlimited free group libraries, which matters directly for a faculty-wide rollout where every supervisor needs a shared collection with their own students, whereas some competitors cap group size on free tiers.
Should a mid-cycle switch migrate students already writing their thesis?
No. Set the new default for incoming cohorts only and run both tools in parallel for one or two academic years, rather than migrating a student already deep into a project.
Is reference-library metadata a high-sensitivity data category under GDPR?
Generally lower sensitivity than full-text student writing, but a faculty with a strict data-residency policy should confirm hosting terms explicitly with the chosen vendor rather than assume citation metadata is automatically out of scope.
