Direct answer
A plagiat detector is a tool that compares a piece of writing against large collections of existing text — the live and archived web, academic journals and periodicals, and an institution's own archive of previously submitted papers — and reports which passages overlap with those sources [1]. The number it returns is a similarity percentage: the share of your text that matched something in the database, not a verdict that you plagiarised [2]. That distinction matters more than any single score, because the same paragraph can produce very different numbers depending on which database a tool searches and which filters it applies [3].
What Does a Plagiat Detector Actually Check — And What Does The Similarity Percentage Really Mean?
A plagiarism detector runs your document through a matching engine that compares it, passage by passage, against three broad corpora. Turnitin's similarity check compares a submission against the current and archived internet, a subscription periodicals and journal collection, and the institution's own repository of previously submitted student papers [1]. Each overlap becomes a "match," and the report's Sources panel lists every matching source alongside the percentage of your paper it accounts for, with the overlapping text highlighted inside your document [1]. That panel is the substance of the report; the headline number is only a summary of it.
The percentage itself is routinely misread. Turnitin states plainly that the similarity score is not a measure of plagiarism — a paper can score highly because it quotes extensively and cites correctly, while a paper assembled from unattributed paraphrase can score low if the underlying sources sit outside the indexed databases [2]. A high score therefore answers "how much of this text also appears elsewhere," not "did this student cheat." A low score answers even less, because translated material, lightly reworded passages, and content from unindexed sources can all pass through a matching engine untouched [2].
What actually separates acceptable from problematic overlap is attribution and proportion. Instructors are advised to read the whole report — which sources matched, how much each contributed, and whether the overlap is quoted, cited, or unattributed — rather than reacting to the total [2]. One large match from a single source usually deserves closer scrutiny than a scatter of small matches from a reference list or from unavoidable subject terminology [2]. If your report shows a single 18% block from one journal article, that is a different situation from twelve 1.5% matches spread across your bibliography.
Report settings also move the number without changing a word of your writing. Excluding quotations, excluding the bibliography, or raising the small-match threshold will each lower the displayed percentage, which is why two readers looking at the same paper can legitimately see different figures [1]. Before you panic about a score, check which filters produced it — and remember that your instructor may be looking at a differently configured report than the one in front of you.
Why Do Different Plagiarism Detectors Give Different Scores For The Same Text?
The single biggest reason is that the tools do not search the same database. A checker is only as broad as its index: one service may crawl the open web but have no access to an institution's archive of past student papers, while another may hold journal licences but index fewer archived pages [3]. Because Turnitin's own comparison set includes the institution's paper repository alongside the internet and periodicals [1], a draft can score noticeably higher there than on a free web-only checker that has never seen the essay your classmate submitted last year.
The matching method and the exclusion rules differ too. Some engines look for word-sequence overlap, others attempt semantic comparison, and each applies its own thresholds for what counts as a match worth reporting — plus its own defaults for excluding quotes, references, and short common phrases [3]. Two tools can therefore disagree about the same paragraph without either being broken; they are simply answering slightly different questions. This is also why the underlying definition of plagiarism matters: copying text, paraphrasing without attribution, and submitting someone else's work are all covered, but a matching engine can only ever detect the textual footprint, not the intent behind it [3].
Finally, no detector output is a determination. Turnitin's guidance frames a report as a signal for human review by an instructor, not an automatic finding of misconduct [3]. That framing is worth holding onto when you compare scores across tools: a difference of ten percentage points between two checkers is usually a statement about their databases and settings, not about your integrity. What transfers reliably between tools is the location of the overlap — the specific sentences that need a citation, a quotation mark, or a rewrite.
How Can I Check My Own Draft With A Real Turnitin Report Before I Submit It?
Most institutions do not give students a self-service route into the real Turnitin database. A draft is normally run through Turnitin only when it is submitted to the course's assignment inbox, which means the first report a student sees is often the one their instructor sees too [4]. Where a course does offer a draft-check or originality-check box, use it early rather than the night before the deadline, and keep a clean copy of the version you checked so you can compare it against what you finally submit [4].
When no institutional self-check exists, students commonly turn to a third-party service that returns a report formatted the way the instructor-facing report is formatted, then revise the flagged passages before the real submission [4]. The value here is diagnostic rather than decorative: the report tells you which sentences to paraphrase properly, which to put in quotation marks with a citation, and which to cut. Because the report is read as a whole rather than as a headline figure [2], the useful output is the Sources panel and the highlighted passages, not the percentage alone.
Two practical cautions apply to any pre-submission check. First, confirm the service does not add your file to a student paper repository — a draft that gets indexed can later match itself, and you do not want your own early version showing up as a source in someone else's report. Second, treat the check as the start of revision, not the end of it: rewriting a flagged sentence into a genuinely different structure and adding the citation is what resolves the overlap, whereas swapping a few synonyms leaves the match in place. Used that way, a pre-submission report turns an anxious guess about your score into a concrete editing list.
Reading the report is the easy half; getting an accurate one before the deadline is the other half. That is exactly what turnitin0 was built for: upload your .docx, .pdf, or .txt draft and you receive the same two PDFs your professor would see — a similarity report with every match highlighted, and an AI writing report — typically in under 15 minutes, without your file being added to any student paper database. Students across the US, UK, Canada, Australia, New Zealand, and Ireland have used it for more than 100,000 reports, and you can check a single draft for $3.80 with no subscription.
※ Turnitin0.com - Actual Turnitin AI Report Cover, Score, Flag And Similarity Summary
FAQ
Is a high similarity percentage the same as plagiarism?
No. Turnitin states that the similarity score is not a measure of plagiarism, and a high figure can come from properly quoted and cited material [2]. The report's Sources panel shows whether each match is attributed, which is what actually matters.
What database does a plagiat detector search?
Turnitin compares a paper against the current and archived internet, a periodicals and journal collection, and the institution's repository of previously submitted student papers [1]. Free web-only checkers typically search a much narrower index, which is why their scores differ [3].
Why did my score change between two checks?
Either the databases differed or the report filters did. Excluding quotes, excluding the bibliography, or raising the small-match threshold each lowers the displayed percentage for the same text [1], and tools with different indexes will report different matches [3].
Can I check my draft before submitting it to my university?
Institutional self-check access varies, and a draft is often run only when submitted to the assignment inbox [4]. Where no self-check exists, students use a service that returns an instructor-formatted report and revise the flagged passages first [4].
Does checking my draft add it to a student paper database?
That depends entirely on the service. A non-repository check does not add your file to Turnitin's student paper database and does not share reports with third-party databases — worth confirming before you upload, so your own draft cannot later match itself.