Direct answer
Turnitin's plagiarism detector works by comparing a submitted document's text strings against three repositories — its student paper database, archived web pages, and academic publications — and then reporting every overlapping passage as a numbered, highlighted match inside a Similarity Report [1]. The result is a similarity index, not a plagiarism verdict: it tells you how much of your text also appears somewhere else, and leaves the judgement about whether that overlap is properly cited to the person marking your work [1]. Understanding that distinction is the difference between panicking over a 34% score and fixing the two paragraphs that actually need a citation.
How Does Turnitin's Plagiarism Detector Actually Match Your Text?
The matching engine is a text-string comparison, not a semantic reading of your argument. Turnitin looks for continuous sequences of words in your submission that also appear in a source it has indexed, then flags the longest and most distinctive overlaps first [2]. It does not know whether you meant to paraphrase, whether you were quoting, or whether the idea is common knowledge — it only knows that the same string of words exists in two places [2].
Because matching is string-based, the detector is sensitive to surface form rather than meaning. Replacing a few words inside a copied sentence often still leaves a detectable match, while a completely rewritten passage with a fresh sentence structure usually drops out of the report even if the underlying idea came from a source [2]. That is why tutors keep saying that good paraphrasing is a structural rewrite, not a synonym swap.
Sources inside the report are ranked by how much text they match. A single source accounting for 18% of your document is far more actionable than twelve sources each contributing 1%, because the first case points to one passage you can quote, cite, or rewrite, while the second is often just a reference list and a handful of standard phrases [2]. Turnitin also lets instructors exclude quotes, bibliographies, and matches below a chosen word count, which is why two students can submit near-identical essays and see very different headline numbers [1].
What Does the Similarity Percentage in a Turnitin Plagiarism Report Really Mean?
The headline percentage is the proportion of your document that matched something in Turnitin's repositories — nothing more, and nothing less [3]. A 40% score on a literature review packed with correctly quoted and cited sources is not evidence of misconduct, and a 4% score on a fully paraphrased essay that lifts arguments without attribution is not evidence of integrity [3]. The number measures overlap, not intent, and Turnitin itself tells instructors to read the report rather than judge the score [3].
Context decides whether a match matters. Direct quotations inside quotation marks with a citation are expected overlap; the same sentence without quotation marks is unattributed borrowing, even though the detector highlights both identically [3]. That is why the useful question is never "what is my score?" but "which specific passages matched, and does each one have a citation attached?" [3].
Two further details regularly confuse students reading a report for the first time. First, the index counts a passage once even if it appears in several indexed sources, so the same paragraph can show up under multiple sources without inflating the total [1]. Second, institutional settings vary — some departments exclude the bibliography by default, others do not — so comparing your percentage with a friend's from a different module is close to meaningless [1]. Read the match list, not the peer comparison.
Can You Check a Draft for Plagiarism With a Real Turnitin Report Before You Submit?
In most institutions, students cannot run an official Turnitin check themselves, because only an instructor account can create an assignment and generate a report [4]. Some tutors deliberately enable a draft folder or a practice assignment so you can see your own similarity report once before the graded submission, but that access depends entirely on how the module is set up [4]. If your course has no such folder, the institutional route is closed to you.
That gap is exactly why independent pre-submission checks exist. A non-repository check runs your draft through the same detection process and returns the same report format — cover page, similarity summary, highlighted matches by source — without adding your file to the student paper database, so your work does not become a future match for anyone else and your own submission is not compromised [4]. You upload a .docx, .pdf, or .txt file, and the report comes back with the numbered sources you need to act on.
The practical workflow is to run the check early enough to revise. Fix the uncited paraphrase, add quotation marks and a page number to the direct quote, and trim the block quotation that is only there to pad the word count. Then re-check the revised draft so you can confirm the matches you fixed actually disappeared before the graded submission [4]. Students who do this once, a day or two before the deadline, rarely meet a similarity report they were not expecting.
If you would rather see your own similarity report before your tutor does, turnitin0 gives you the same report layout — cover page, overall similarity index, and every highlighted match listed by source — so you can revise with real evidence instead of guesswork.
※ Turnitin0.com - Actual [Turnitin AI Report](https://www.turnitin0.com/guides/us/originality-ai-alternative) Cover, Score, Flag And Similarity Summary
FAQ
Does a high similarity percentage mean I plagiarised?
No. The percentage measures how much of your text matched indexed sources, and correctly quoted, cited material counts toward it [3]. Read the match list and check each highlighted passage for a citation before drawing any conclusion.
Why does my friend's essay show a lower score than mine?
Instructors choose which exclusions to apply, including quotes, bibliographies, and small matches, so settings differ between modules [1]. Two reports generated under different settings are not comparable numbers.
Can Turnitin detect plagiarism in paraphrased text?
It detects reused word strings, so light paraphrasing that keeps the original sentence structure can still match [2]. A genuine structural rewrite with a new sentence pattern and a citation is what removes the match.
Does a pre-submission check add my paper to the student database?
A non-repository check runs the same detection process without storing your file in the student paper database, so it does not create a future match for other students [4]. Always confirm the repository setting before you upload.
How early should I check a draft?
Early enough to revise and re-check. Running a check a day or two before the deadline gives you time to fix uncited paraphrase, correct quotations, and confirm the matches are gone [4].