What Rayyan is
Rayyan is a web platform for the screening stage of a systematic review. A team uploads a deduplicated set of records, each reviewer votes include or exclude independently, and Rayyan withholds everyone’s decisions from everyone else until the blind is lifted — at which point disagreements surface as a conflict list to be resolved.
It came out of the Qatar Computing Research Institute and spread through the systematic-review community for one straightforward reason: the collaborative screening workflow that other tools charged for was free and worked. That origin still shapes it. Rayyan is built around the methodological requirement for two independent screeners, not around being a general-purpose literature tool, and it does that one job with an audit trail that maps directly onto what PRISMA reporting asks you to state.
Why researchers use it
- Blind mode done properly — reviewers cannot see each other’s votes, which is the entire methodological point and something a shared spreadsheet cannot enforce.
- Automatic conflict detection — disagreements are collected into a list to work through together, rather than discovered by accident at the end.
- Keyword highlighting — mark inclusion and exclusion terms and Rayyan colours them across every abstract, which measurably speeds up the fifth hour of screening.
- Exclusion reasons as structured labels — the data your PRISMA flow diagram needs, captured while you screen instead of reconstructed afterwards.
- It works on a phone — genuinely useful for burning through a few hundred obvious excludes on a commute.
Where it fits in a research workflow
Rayyan sits at exactly one stage: title-and-abstract screening with more than one screener. Upstream is search and deduplication, done in EndNote, Zotero, or a dedicated deduplication pass — and this matters, because Rayyan’s own deduplication is the weakest part of the product and should not be your only line of defence.
Downstream is full-text assessment and data extraction, which Rayyan does not do. That is where Covidence takes over, or a structured spreadsheet if the review is small. If you want machine assistance in the ordering of the pile rather than in the collaboration around it, ASReview covers the other half of the problem — and the two are not mutually exclusive, though running both means moving records between them by hand.
Getting started
An hour to set up a review properly, and the order of the first three steps matters.
- Deduplicate before uploading, in your reference manager. Rayyan will catch some duplicates; it will not catch enough, and a duplicate that survives into screening becomes a discrepancy in your PRISMA counts.
- Create the review, upload the RIS or CSV, invite your co-reviewers, and switch blind mode on before anyone screens a single record. Turning it on later does not retroactively blind decisions already made.
- Agree the exclusion-reason labels with your team up front and enter them as a fixed list. Free-text reasons produce a flow diagram nobody can reconcile.
- The step people skip: export a full backup of your decisions once a week. It takes a minute and it is your only defence against a platform problem mid-review.
Rayyan vs the alternatives
| Alternative | Does it better | Pick it if |
|---|---|---|
| Covidence | Full-text stage, risk of bias, extraction, automatic PRISMA numbers | You need one audited system from import to extraction table |
| ASReview | Peer-reviewed active learning that reorders the pile; runs locally | Your pile is huge and you can defend a stopping rule |
| EndNote | Deduplication at scale before anything reaches screening | You are merging four database exports and need them clean |
| A shared spreadsheet | Nothing, methodologically | Your review has under 200 records and one screener |
Pick Rayyan when the constraint is people rather than volume: two or more screeners, a real protocol, and no budget. Pick Covidence when the review is funded and you need the stages after screening too.
Cost, licensing, and your data
Freemium, and the free tier has narrowed since the early years — features that were once unrestricted now sit behind Rayyan Pro or a team plan, and the number of records or reviews you can run free is capped in ways that were not originally the case. Check the current limits against your review’s actual size before committing a team to it, because migrating a half-screened review is miserable.
Your records and decisions live on Rayyan’s servers. For most published-literature screening that is unremarkable — you are uploading bibliographic metadata for papers that are already public. The case to think through is a review whose search strategy is itself sensitive: a grant-competitive topic, an unpublished protocol, or an industry-funded review under confidentiality. If that describes you, ASReview runs locally and never uploads anything.
The honest review
Strengths. Rayyan made proper blind dual screening free, and in doing so raised the methodological floor for an entire generation of reviews that would otherwise have been screened in a shared spreadsheet with everyone’s votes visible. The conflict-resolution workflow is well judged: it surfaces disagreement at the right moment, in a form a team can actually work through in a meeting.
Limitations. Deduplication is weak enough that treating it as sufficient is a real risk to your PRISMA counts. Performance degrades noticeably on very large record sets — well past ten thousand records the interface becomes something you endure rather than use. The AI relevance suggestions are far less validated than ASReview’s published approach and should be treated as loose ordering hints, never as evidence in a methods section. And the steady narrowing of the free tier means the tool you evaluated may not be the tool you get eighteen months into a review.
Verdict. Adopt it for any multi-reviewer screening stage without a budget — it remains the best free option and the workflow is sound. Skip it if you are screening alone, if your review needs the full-text and extraction stages managed too, or if your record count runs into five figures. The condition that flips the answer is funding: with a Covidence licence available, Covidence covers more of the review with fewer handovers.
Common questions
Is Rayyan still free?
There is still a free tier, but it is more limited than it once was — record and review caps, and several features moved to paid Rayyan Pro and team plans. Check the current limits against the size of your review before you start, since moving a partly screened review is painful.
Does Rayyan support blind double screening?
Yes, and this is its core function. Enable blind mode before anyone begins; reviewers then cannot see each other’s include/exclude decisions until the blind is lifted, at which point conflicts are listed for resolution.
Rayyan or Covidence — which should I use?
Rayyan if you need free collaborative screening and will handle full text and extraction elsewhere. Covidence if the review is funded or institutionally licensed and you want screening, full-text assessment, risk of bias, extraction, and PRISMA numbers in one audited system.