Guides · 6 min read

PubScope vs Citely: Which Citation Checker Should You Use?

Both tools do the same core job: you paste a reference list, and they tell you which entries do not check out. They differ most in what else they look for and in what each company can show you about its own accuracy. This page lays both out. Every Citely detail below was read off citely.ai on 29 July 2026; if something has changed since, tell us and we will correct it.

Where Citely is the better choice

Two real advantages, and they may well be decisive for you:

Side by side

PubScopeCitely
PriceCurrently free — every feature is open while we are in beta. Listed plan when we switch it on: $15/month or $99/year, halved for 14 lower-income countries.$9 trial (15 credits) · $19/month · $14/month billed yearly · $347 once (5,000 credits, 3 years). No free tier stated.
Databases searchedCrossref, OpenAlex, DataCite, Semantic Scholar, arXiv — plus a local copy of the Retraction Watch database and our own journal database.States “CrossRef, PubMed, arXiv, Google Scholar”; the page also shows logos for OpenAlex and Semantic Scholar.
Real DOI, fabricated titleReported as a DOI mismatch. 150/150 on our sealed holdout, in four consecutive runs.Not separately stated on the page.
Retracted papersYes — 70,742 Retraction Watch records held locally. Also reports the compound case: the DOI belongs to a different paper and that paper is retracted.Not stated.
Venue trust signalsFlags references to venues our database marks as DOAJ-withdrawn (1,208 journals) or very low trust (52). Catalogue of 46,797 journals.Not stated.
PDF uploadNot in the reference checker — it is paste-only. Upload a manuscript PDF in Paper Review instead and the same engine checks its bibliography.Not stated on the page.
Reference managersNone. BibTeX and RIS paste only.Zotero, Mendeley, EndNote.
Published accuracy evidenceA 14-case public set you can paste in yourself, plus a 1,320-reference sealed holdout with Wilson confidence intervals, the seal hash, our measured blind spots, and the bugs the set found in us.“over 95% accuracy in detecting authentic vs. fabricated citations”. No test set, sample size, date or method is given on the page.
Beyond citationsJournal finder, journal Trust Score, comparison, pre-submission review, advisor, an MCP server with 4 tools.Citation checking and source finding.

“Not stated” means the feature does not appear on their public pages as of 29 July 2026. It is not a claim that the product cannot do it.

The accuracy claim is the real difference

Citely's page says the tool achieves “over 95% accuracy in detecting authentic vs. fabricated citations”. We are not saying that figure is wrong. We are saying that as published it cannot be checked: there is no test set, no sample size, no date, and no description of how a citation was judged authentic. A percentage without those four things is not a measurement you can act on — it is a number.

Our answer to the same question is deliberately shaped so you can attack it. The validation page publishes all 14 labelled references so you can paste them in and reproduce the run. It reports a 1,320-reference holdout that is sealed with a SHA-256 hash before anything is measured. It gives Wilson confidence intervals rather than bare percentages. And it publishes the parts we lose on — a one-word semantic flip in a title is caught between 25 and 38 times out of 60, unstable across runs, and we print the range rather than pick the flattering end.

It also records that building that set exposed two bugs in us, including one that told authors correctly citing Attention Is All You Need that their reference could not be found. We would rather you read that than discover it yourself.

What we do not claim

Our numbers are not an independent audit — we built the set, ran it and reported it. The sample is drawn from Crossref, OpenAlex and Retraction Watch, so it is overwhelmingly English-language and well-indexed; obscure, non-English and very recent references are harder for us. And because our results are also not measured on the same set as anyone else's, they cannot be compared score-to-score with another tool's published figure. Different sets, different constructions, different definitions of a “catch”.

See it for any journal on PubScope

PubScope brings the public signals together for tens of thousands of journals: Web of Science / Scopus / DOAJ indexing, SJR quartile, APC, a 0–100 Trust Score and predatory-risk flags — each linking out so you can confirm it at the source.

Check a journal Find journals for my paper

How to choose

Sources