Direct answer
You check an article for plagiarism by running it through a text-matching tool that compares every sentence against a database of web pages, academic publications, and previously submitted student papers, then reading the resulting similarity report rather than just the headline percentage. A similarity score is not a verdict — it tells you how much text matched something else, and a human still has to decide whether those matches are properly quoted, cited, and paraphrased [1]. The most reliable workflow is: confirm what counts as plagiarism, run the check, then interpret each flagged match individually.
What do plagiarism checkers actually compare an article against?
A plagiarism checker does not "know" whether you cheated. It performs string and phrase matching between your article and an index of documents, and it reports where the wording overlaps. Turnitin's similarity report, for example, compares a submission against three broad pools: its repository of previously submitted student papers, the live internet, and a collection of academic publications and journals [2]. That three-pool structure is why the same paragraph can match one source at 4% and a completely different source at 11% — each match is scored separately, per source, not as one lump sum [2].
The practical consequence is that the checker can only find what is in its index. If you copied from a printed book, a paywalled database the tool does not license, a private group chat, or a translated article, the match may never surface. This is why "I ran it and it came back clean" is weak evidence on its own — and why academic integrity guidance stresses that plagiarism includes taking ideas without credit, not just lifting sentences [1]. A checker catches text; it cannot catch an uncredited argument.
Finally, understand what a checker is not doing. It is not reading your article for comprehension, it is not judging whether your paraphrase genuinely restructured the idea, and it is not deciding your grade. It produces evidence — highlighted passages, source links, and percentages — that a human then evaluates [2]. Treat the report as a diagnostic instrument, not a judge.
What does a similarity percentage actually mean — and when is it a problem?
The similarity percentage is the proportion of your article's text that matched something in the tool's index. It is a matching metric, not a misconduct metric, and Turnitin states plainly that there is no universal cut-off above which a document is plagiarised [3]. A literature review that quotes heavily and cites correctly can legitimately sit at 30%, while a short essay with one unattributed copied paragraph might sit at 8% and still be a serious integrity problem. The number alone tells you almost nothing until you open it.
What matters is the composition of the matches. Ask three questions of every flagged passage: Is it quoted and attributed? Is the citation present and correctly formatted? Is the overlap a common phrase ("the results of this study show") rather than a distinctive idea? Small matches against widely used phrases are usually noise, and most institutional setups let the reviewer exclude quotes and bibliography so the remaining figure reflects actual unoriginal prose [3]. If you have never been shown how to read the report, the honest answer is that the percentage is uninterpretable to you yet.
A second layer worth knowing about is AI-writing detection, which is reported separately from similarity. Turnitin's AI writing report estimates how much of a document was likely generated by a large language model, and it is displayed alongside — not merged into — the similarity score [4]. So a draft can be 0% similar and still carry a high AI indicator, or vice versa. When you check an article, check both dimensions, because universities increasingly look at both.
Can I check my own article against Turnitin before I submit it?
Usually not directly. In most institutional setups, a student can only see a similarity report when the instructor has enabled the student view for that assignment, and whether a draft is added to the paper repository depends on settings the student does not control [4]. That means the "submit it to Turnitin first and see what happens" strategy is unreliable — you may burn a submission slot, you may not see the report at all, and in some configurations your draft enters the repository and then matches against itself on the real submission.
The workaround students actually use is a pre-submission preview service that returns the same two documents an instructor sees: a similarity report and an AI writing report. On turnitin0, a check accepts .docx, .pdf, or .txt in English between 300 and 30,000 words, returns both PDFs in one order, and runs without adding the file to Turnitin's student paper database — so the draft you check does not become a future match for anyone, including you. Turnaround is under 15 minutes in 98% of cases. Pricing is pay-per-use at $3.80 for a single check, or $2.75 per check in a 10-check pack, with no subscription.
Before you spend anything, though, do the free pass first: read your own article against its source list and confirm every borrowed idea has a citation, then check the flagged passages yourself. The paid check is for the cases where you genuinely cannot tell whether your paraphrase is too close — which, in practice, is most of them [4].
If you want to see the exact report your instructor will open — the flagged passages, the per-source percentages, and the AI writing indicator side by side — turnitin0 lets you run that check on your own draft before it counts.
※ Turnitin0.com - Actual Turnitin AI Report Cover, Score, Flag And Similarity Summary
FAQ
Is a 20% similarity score bad?
Not necessarily. Turnitin does not publish a threshold above which a document is plagiarised, and a heavily quoted, properly cited review can sit well above 20% legitimately [3]. Open the report and inspect what matched before drawing any conclusion.
Do free plagiarism checkers use the same database as Turnitin?
No. Free tools typically search the open web only, while Turnitin also matches against its repository of previously submitted student papers and licensed publications [2]. A free check can miss a match that an institutional check will find.
Can a checker detect plagiarism from a translated or paraphrased source?
Paraphrased text is often caught when the wording stays close to the original; translated text is frequently missed because the index holds the source in another language [1]. Neither outcome proves the writing is original — citation is still your responsibility.
Will checking my article first get me flagged for self-plagiarism?
Only if the checked draft is stored in a repository and later compared against your real submission. A non-repository check avoids that loop, which is why the storage setting matters as much as the score [4].
Does the AI writing report replace the similarity report?
No — they are separate reports measuring different things. Similarity measures text overlap with sources; the AI writing report estimates likelihood of machine generation [4]. Read both.