Direct answer
Similarity detection compares a student's paper against a large database of web pages, academic publications, and previously submitted student papers, then reports the percentage of text that matches those sources. That percentage is a similarity score, not a plagiarism verdict — it tells you how much of your wording overlaps with existing text, and it is your instructor who decides what that overlap means [1]. For students, the practical value of similarity detection is diagnostic: it shows which sentences, quotes, and reference entries are being matched, so you can fix citation problems before a grade is affected. Understanding how the score is built, and why honest work can still score high, is the difference between panicking over a number and using it properly.
What Does Similarity Detection Actually Measure in a Student Paper?
Similarity detection measures textual overlap, not originality of thought. The system breaks your document into segments, searches those segments against its indexed sources, and flags any segment that closely matches wording found elsewhere. What it reports is the proportion of your paper's text that matched something in the index — nothing about whether you understood the material, cited it correctly, or wrote it yourself [2].
The sources are not one undifferentiated pool. Reports typically separate matches into distinct categories: internet sources, academic publications and journals, and other students' papers held in the institution's repository [2]. This categorisation matters because it changes your response. A high internet match usually points to copied phrasing or an unquoted definition; a high publications match often signals paraphrasing that stayed too close to the original article; a student-paper match can mean you and a classmate worked from the same template or shared notes.
Reports also apply filters that change the displayed number without changing your text. Institutions can choose to exclude quoted material and exclude the bibliography, which removes the two most common legitimate sources of matching text from the calculation [2]. A paper showing 28% with quotes included may show 11% once quotations and references are excluded. If you are reading your own report and the number looks alarming, check which filters were applied before drawing any conclusion.
Finally, similarity detection is deliberately built as an interpretation tool rather than a scoring machine. Turnitin's own guidance tells instructors that the score is a starting point for reading the highlighted passages, and that small matches can be filtered out by word count so trivial phrase overlaps do not dominate the report [2]. The report's real content is the highlighted text and the numbered source list beside it — that is where the actionable information lives.
Why Can a Similarity Score Be High Even When a Student Wrote the Work Themselves?
A high score on genuinely original work is common, and the causes are structural rather than suspicious. Academic writing is full of conventional phrasing: standard definitions, methodological boilerplate, transitional sentences, and discipline-specific terminology all appear in thousands of papers, so they match by default [3]. A literature review that correctly names and defines six theories may accumulate matches on every definition simply because the standard wording is the standard wording.
Reference lists and quoted evidence are the second major contributor. A bibliography is, by design, a list of strings that already exist elsewhere — author names, article titles, journal names, years, and DOIs — so it will match almost completely [3]. Direct quotations match by definition, which is exactly why institutions offer the exclude-quotes filter. Neither of these is misconduct; both inflate the raw percentage.
There is also a counting behaviour that surprises students: if the same source is matched repeatedly throughout your paper, it is generally counted once as a source but its matched text still contributes across the document [3]. One heavily cited textbook can therefore account for a substantial share of the score while representing perfectly proper scholarship.
Most importantly, there is no universal "safe" threshold. Institutions and individual instructors set their own expectations, and the same score can be unremarkable in one department and worth a conversation in another [3]. This is why the number alone cannot tell you whether you have a problem — only the pattern of matches, read against your assignment's citation conventions, can.
Can Students Check Their Own Similarity Score Before Submitting?
Access to a pre-submission check depends entirely on your institution's configuration. Turnitin's student guidance makes clear that whether you can see a similarity report for a draft is controlled by your instructor — some set up draft or self-check assignments that let you review a report and revise, while others do not enable that option at all [4]. If no such assignment exists in your course, there is no institutional route for you to generate a report yourself.
Where draft submissions are enabled, there is a detail worth knowing before you upload: depending on the assignment settings, a draft you submit may be stored in the institution's paper repository, which can affect later submissions of the same work [4]. Students should confirm with their instructor what the draft assignment does with the file, and whether resubmission is permitted, rather than assuming a practice run is invisible.
Because availability varies so widely, the reliable first step is to ask. Turnitin's guidance for students recommends confirming with your instructor or department what checking is permitted before you use any external service, since policies on pre-submission checking differ between institutions and courses [4]. Getting that answer in writing removes the risk of breaching an academic integrity rule while trying to comply with one.
For students whose course does not provide a draft check, or who want to see the full report — similarity and AI writing together — before the real deadline, an independent preview service is the practical alternative. The key is that the report should show what your instructor's system will show, with the same highlighted matches and source list, so the preview is genuinely informative rather than a different metric wearing a familiar name.
Reading the highlights is only useful if you can act on them before the deadline closes. That is exactly what turnitin0 is built for: upload your .docx, .pdf, or .txt draft and receive the same two PDFs your instructor sees — a similarity report with numbered, highlighted sources and an AI writing report — usually within 5 to 15 minutes, without your file being added to Turnitin's student paper database.
※ Turnitin0.com - Actual [Turnitin AI Report](https://www.turnitin0.com/) Cover, Score, Flag And Similarity Summary
FAQ
Is a similarity score the same as a plagiarism score?
No. It is the percentage of text matching indexed sources, and Turnitin explicitly frames it as something to interpret rather than a judgement [1]. Plagiarism is a determination your instructor makes after reading the matches in context.
What similarity percentage is considered safe?
There is no universal threshold. Institutions and instructors set their own expectations, and the same figure can be routine in one department and flagged in another [3]. Read the highlighted matches rather than the headline number.
Why did my score drop when quotes and references were excluded?
Because quoted material and bibliographies match existing text by design. Reports can exclude quotes and the bibliography, which removes those legitimate matches from the calculation [2]. A large drop is normal and not evidence of a problem.
Does checking my draft add it to the student paper database?
It depends on the assignment settings your institution uses, and a draft submission may be stored in the repository [4]. Confirm with your instructor what the draft assignment does with your file before you upload.
Can I see the report my instructor will see?
Only if your course enables a draft or self-check assignment, which is the instructor's decision [4]. Otherwise, an independent preview service that returns the same similarity and AI writing reports is the practical way to check before the deadline.