Direct answer
A plagiarism checker compares your document against three large bodies of text — a repository of previously submitted student papers, a collection of academic journals and publications, and archived and live web pages — then returns a similarity report that lists every source it matched and highlights the overlapping text. It does not decide whether you plagiarised. It produces evidence for a human to read: a similarity index, a numbered source list, and colour-coded matches you can inspect line by line [1]. Understanding that distinction is the whole game, because the number people panic about is a measurement of overlap, not a verdict on your integrity.
What does a plagiarism checker actually compare your paper against?
The comparison set is the part most students never think about, and it is the part that determines every result you will ever see. Turnitin's core comparison database is built from three distinct collections: the student paper repository, which holds papers previously submitted to institutional accounts; a subscription collection of academic journals, periodicals and publisher content; and an internet archive that crawls and stores billions of web pages, including pages that have since been taken down [2]. Your draft is fingerprinted and matched against all three at once, which is why a report can surface a 2014 essay from a university on another continent alongside a journal article and a Wikipedia revision.
The student paper repository is the source that surprises people most. When a paper is submitted to an institutional Turnitin account, it is added to that institution's repository by default, so a later submission can match it even if the original was never published anywhere [2]. This is not a punishment mechanism — it is how a checker detects recycling between students and reuse of essays bought or shared from previous cohorts. It also explains why a genuinely original piece of writing can still show matches: the phrasing you arrived at independently may already exist in a paper someone else submitted.
What the checker does not compare against is just as important. It does not read your intentions, it does not know whether you cited a source correctly, and it does not distinguish between a quotation you marked properly and a passage you lifted without attribution — both appear as matched text until a human looks at the context [2]. That is why the report is structured as a set of clickable sources rather than a single score. The tool's job is to find overlap and show you where it is; the interpretation of that overlap is deliberately left to the reader.
What does the similarity percentage on a plagiarism report actually mean?
The similarity index is simply the proportion of your document's text that matched something in the database. It is not a plagiarism score, and Turnitin states plainly that a high similarity index is not evidence of misconduct on its own [3]. A paper with a 34% index might be entirely properly cited, with the matches concentrated in a block-quoted methodology section and a long reference list; a paper with 8% might contain one unattributed paragraph copied from a source the checker only partially indexed. The percentage tells you how much text overlapped, not whether that overlap was legitimate.
Two failure modes follow from this, and both are common. The first is treating a high number as a confession: students rewrite perfectly good sentences to push a score down, damaging their own argument while the underlying citation problem stays untouched. The second is treating a low number as a clean bill of health — a 0% result does not prove originality, because a source may simply sit outside the indexed collections, and paraphrasing that changes vocabulary without changing structure can slip under the matching threshold [3]. Neither direction of misreading helps you.
The practical skill is reading the source list rather than the headline figure. Open each match, look at what percentage of the document it accounts for, and ask a single question: did I intend this overlap? Bibliography entries, direct quotations with quotation marks, common technical phrases and standard headings routinely inflate the index and are normally excluded or discounted by the person assessing the report [3]. What matters is the unattributed block of prose sitting in the middle of your analysis. That is the thing worth fixing, and it is visible only in the highlighted matches, not in the number.
Can you see your own Turnitin similarity and AI report before you submit?
Usually not through your institution. Students cannot normally self-submit to a live Turnitin assignment; whether you can preview a report depends entirely on your instructor enabling a draft or self-check assignment in the course [4]. Many courses never turn that option on, which means the first time a student sees a similarity report is often after the grade has already been recorded. By then the report is diagnostic rather than corrective — useful for learning, useless for the deadline you just missed.
Where self-check is unavailable, students use a pre-submission checking service to generate the same two documents on their own draft: a similarity report and an AI writing report, structured the way an instructor would see them [4]. The value is timing. Seeing the source list a day before the deadline turns an abstract worry into a specific, fixable list of paragraphs — you can paraphrase deliberately, add the missing citation, or cut a passage that was doing less work than you thought. Previewing also removes the guesswork from the AI side of the report, which is a separate signal from similarity and is not visible in a similarity percentage at all.
The honest limit of any preview is that it is a snapshot of the database at that moment. Repositories grow, web archives update, and a paper submitted after your check can create a match that did not exist when you ran it [4]. What a preview reliably gives you is a clear picture of your own citation hygiene and phrasing habits — the parts of the outcome you actually control.
If you would rather see the source list and the score before your instructor does, turnitin0 generates the real Turnitin AI detection report and similarity report on your own draft in most cases within 5–15 minutes — the same two PDFs your professor sees in the LMS, checked without adding your file to the student paper database, so previewing your work never creates the match you were trying to avoid.
※ Turnitin0.com - Actual Turnitin AI Report Cover, Score, Flag And Similarity Summary
FAQ
Does a plagiarism checker prove I cheated?
No. It reports text overlap and lists the sources involved; the judgement about intent and citation practice is made by a person reading the matches in context [1]. A high index is a prompt to look closer, not a finding.
Why does my original essay show a similarity match?
Because the database contains millions of student papers, journal articles and archived web pages, and common phrasing, standard terminology and reference lists frequently coincide with existing text [2]. Open the match and check whether the overlap is a citation, a quotation, or your own prose.
Is a 0% similarity score good?
It means nothing matched in the indexed collections — not that the work is original [3]. A source outside the database, or heavy paraphrasing that preserves structure while swapping words, can both produce a low score.
Does the checker also detect AI-written text?
Similarity and AI detection are separate outputs from separate models. Turnitin's AI writing report estimates how much text appears machine-generated and is displayed independently of the similarity index, which is why a draft can be 5% similar and still carry an AI flag [3].
Will checking my paper before I submit flag me later?
Not with a non-repository check. A pre-submission service that does not add your file to the student paper database leaves no stored copy for a future submission to match [4].