Direct answer
An online similarity plagiarism checker works by fingerprinting your document into overlapping text strings, matching those strings against a repository of student papers, web pages and publications, and then reporting every matched passage with its source and a total similarity percentage [1]. That percentage is a measure of matched text, not a verdict on intent — a 12% score can be entirely legitimate citations, while a 4% score can contain one copied paragraph that matters [1]. The practical value of a checker therefore lies in the detail behind the number: which sources matched, how long each match runs, and whether quoted material was excluded before the score was calculated [1].
How does an online similarity plagiarism checker actually match your text against sources?
A similarity checker does not read your argument or judge your ideas; it compares strings. Turnitin's system breaks a submission into small overlapping segments and looks for those segments inside its comparison repositories, then groups consecutive matches into the passages you see highlighted in the report [1]. The repositories behind that comparison are the part students rarely inspect, and they are the part that decides whether your score is meaningful: Turnitin compares against its current and past student paper database, a large index of internet content, and a set of academic publications and journals [1]. A checker with a thin or web-only index can return a low score simply because it never held the paper your text came from [2].
The matching mechanics also explain why the report looks the way it does. Every matched string is attributed to a specific source, ordered by how much text it contributed, and colour-coded so a reader can see at a glance whether the overlap is a short phrase or a multi-sentence block [1]. Because the engine works on strings rather than meaning, close paraphrase can still register as a match when the wording and structure stay near the original, while a genuinely independent rewrite with new sentence structure usually does not [2]. That is why a similarity report is best read as a map of textual overlap, and why the number alone cannot tell you whether a passage was cited properly [1][2].
Institutional practice matters here too. Turnitin's similarity checking is not a single global comparison: whether a submission is checked against the institution's own paper repository, and whether that paper is then retained for future comparisons, depends on settings the institution controls [2]. A draft checked outside that environment is compared against the same public corpora but not against your classmates' unpublished work — a limitation worth knowing before you treat a low score as final clearance [2].
Why does the similarity percentage differ between checkers for the same document?
The same essay can produce very different percentages on two checkers, and almost none of that difference comes from one tool being "more honest" than the other. The largest driver is corpus coverage: a checker whose index excludes student papers cannot match a paragraph lifted from an essay posted to a course-sharing site, and a checker with a smaller journal set will miss publication overlap that Turnitin's index catches [2]. Turnitin's own guidance treats the similarity score as dependent on the repository set and on the exclusions applied, which is why the same file can score differently across settings [1].
The second driver is exclusion configuration. Institutions routinely switch off small matches below a word threshold, exclude bibliography and quotes, or exclude a source that has already been cited, and each of those choices moves the headline number without changing a word of the text [1]. Turnitin's report documentation makes the point explicitly: the percentage is a total of matched text after the configured filters, so two readers looking at the same submission under different filters are looking at two different scores [1]. If you compare a third-party checker's output with a university report, compare the match list and the exclusion settings, not just the percentage [2].
The third driver is timing and retention. Because student-paper repositories grow as submissions are added, a document checked in week two may match more sources in week ten, and a draft that was never added to a repository will not appear in anyone else's future matches [2]. Turnitin's help documentation also notes that similarity results are re-generated against the repository as it stands at the time of the check, so a stored report is a snapshot rather than a permanent property of the text [1]. Treat any single percentage as a reading taken at a moment, under a specific set of settings, against a specific corpus [1][2].
What should you check in a similarity report before you submit your draft?
Read the match list before you read the score. Turnitin's report orders sources by the volume of matched text and lets you open each one to see the exact highlighted passage, which is the only reliable way to tell a legitimate quotation from an unattributed lift [1]. Work down the list from the largest contributor: a single source accounting for most of your percentage usually means a block of text was reproduced closely, while a spread of many 1–2% matches across reference works is the normal signature of a properly cited literature review [1]. Fixing the top source typically moves the number more than editing a dozen small matches [2].
Then check whether your citations are doing the work they appear to do. A quote that is correctly attributed but not enclosed in quotation marks still shows as a match, and a paraphrase that keeps the original clause order will match even with synonyms swapped in [2]. Turnitin's interpretation guidance recommends reviewing matched passages in context and confirming that quoted material is marked and referenced, because the report flags overlap rather than misconduct [1]. If a match is a genuine quotation, the fix is formatting and attribution; if it is close paraphrase, the fix is rewriting the sentence structure rather than swapping words [2].
Finally, verify the conditions of the check itself. Confirm which repository the draft was compared against, whether bibliography and quotes were excluded, and whether small matches below the institution's threshold were suppressed — because those settings determine how your number will compare with the official one [1]. Turnitin's documentation also stresses that the report is a teaching and review instrument, and that interpretation belongs to the instructor applying the institution's policy [1][2]. A report you can read line by line, with sources named and passages highlighted, is the only kind that lets you make a decision before the deadline rather than after it [1].
Reading a similarity report properly is exactly what turnitin0 was built for: students upload a .docx, .pdf or .txt draft and receive two PDFs in one checkout — a Turnitin AI detection report and a similarity report showing the matched sources, the highlighted passages and the score summary, in the same format instructors see. Files are checked without being added to Turnitin's student paper database, most orders finish in 5–15 minutes, and 100,000+ reports have been delivered to 20,000+ students. Run the check while you can still edit the draft.
※ Turnitin0.com - Actual Turnitin AI Report Cover, Score, Flag And Similarity Summary
FAQ
Is a similarity percentage the same as a plagiarism score?
No. Turnitin's documentation describes the similarity score as the total volume of matched text after configured exclusions, and states that a high score is not automatically plagiarism and a low score is not automatically original [1]. Interpretation depends on which passages matched and whether they were cited.
Can an online checker see the same sources my university's Turnitin sees?
Only partly. The public internet and publication corpora are broadly similar, but comparison against an institution's own student paper repository depends on that institution's settings [2]. That is the main reason a third-party score can sit lower than the official one.
What similarity percentage is acceptable?
There is no universal threshold. Turnitin's guidance ties acceptable overlap to the assignment, the discipline and the institution's policy rather than to a fixed number [1]. Review the match list and your citations instead of chasing a target figure.
Does checking my draft add it to a student paper database?
It depends on the service. Turnitin0 checks files without adding them to Turnitin's student paper database, and files can be deleted from the account. Institutional submissions, by contrast, may be retained in the institution's repository depending on its settings [2].
Why do small matches disappear from my report?
Institutions can suppress matches below a word threshold, and Turnitin's report settings include exclusions for small matches, quotes and bibliography [1]. If a passage vanished, check the exclusion settings before assuming the text is unique.