Direct answer
A plagiarism scan is an automated comparison of your document against a large corpus of sources — web pages, academic publications, and previously submitted student papers — that returns a similarity report listing every matched passage and its source [1]. The scan itself does not decide whether you plagiarised; it produces a percentage plus a source-by-source match list that a human then interprets. What matters most is knowing which database was searched and reading the matches rather than the headline number [1].
What Database Does a Plagiarism Scan Check Against?
A plagiarism scan does not search "the internet" in the way a browser does. It compares your text against a structured corpus, and in Turnitin's case that corpus is built from three distinct pools: a continuously crawled index of web content, a library of subscription publications and journals, and a student paper repository made up of work previously submitted through institutional accounts [2]. Each pool behaves differently, which is why two scans of the same paragraph can return different results depending on which one you used.
The student paper repository is the pool students ask about most, and it is also the most misunderstood. A paper enters that database only when the submitting institution's account is configured to retain it; a check run without repository storage does not add your file to the pool that future students are matched against [2]. That distinction matters for your own draft: a non-repository check gives you the same match detection without turning your essay into a permanent source for someone else's report.
The corpus is also moving. Web crawling and new institutional submissions mean the database is updated constantly, so a document that scored 12% in one term can score 19% in the next without a single word changing [2]. Treat any similarity figure as a snapshot tied to a date and a specific database configuration, not as a fixed property of your writing.
Finally, no scan covers everything. Paywalled databases, unpublished lecture notes, and work submitted to institutions outside the network are outside the corpus, so a clean report is evidence of low overlap with the searched sources — not a guarantee of universal originality [2]. Reading the report as "what was found in this database" is the accurate interpretation.
How Do You Read the Similarity Score in a Plagiarism Scan Report?
The similarity index is the percentage of your document's text that matched something in the database, and it is a measure of overlap, not of misconduct [3]. A 30% score can be entirely legitimate if your paper quotes three long passages with full attribution, because correctly cited quotations still count toward the percentage. Conversely, a 4% score can hide a single copied paragraph if the rest of the document is original, which is why instructors read the match list rather than the number [3].
Open the report and work through the sources individually. Each matched source is listed with its own contribution, and clicking a match shows the exact highlighted strings, so you can see whether the overlap is a quotation, a common phrase, a reference entry, or an unattributed block of prose [3]. The reference list and bibliography are frequent and harmless contributors; many institutions exclude them with a filter, and you should check whether the report you are reading has that filter applied before you panic about a double-digit figure.
Pay attention to the pattern of matches, not just the total. A cluster of small matches spread across generic phrasing suggests the detector is catching stock academic language, while one large contiguous highlight from a single source is the pattern that needs rewriting [3]. That single-source block is also the one most likely to draw an academic integrity conversation, regardless of what the overall percentage says.
Can You Run a Plagiarism Scan Before You Submit Your Paper?
Yes, but access depends entirely on how your institution has configured its account. In many university setups, draft checking is deliberately restricted and students see a similarity report only after the graded submission has been processed, which leaves no opportunity to revise [4]. Where an instructor has enabled a draft folder, you can submit once and view the report before the final deadline — but that pre-submission copy may itself be stored, depending on the repository setting.
The practical value of scanning early is that it converts an anxious guess into an editable list. A pre-submission scan shows you which sentences are matching, so you can re-paraphrase, add missing attribution, or move a quotation into proper block format while the work is still yours to change [4]. Students who scan only after the deadline are reading a diagnostic they can no longer act on.
If your institution does not offer a draft check, an independent non-repository scan reproduces the same comparison against the same content pools without adding your file to the student paper database [4]. That keeps the diagnostic benefit — real matches, real source list, real percentage — while leaving your draft out of the pool that later students are matched against. Whichever route you take, scan the version you actually intend to submit, including the reference list, because a scan of an earlier draft tells you nothing about the file your instructor will receive.
If you want that same source-by-source match list on your own draft — before the deadline, not after — turnitin0 runs the check and returns the actual Turnitin similarity and AI reports, so you can see exactly which passages are flagged and fix them while it still counts.
※ Turnitin0.com - Actual Turnitin AI Report Cover, Score, Flag And Similarity Summary
Get [Real Turnitin](https://www.turnitin0.com/) AI & Similarity Report
FAQ
Does a plagiarism scan check against published journal articles?
Yes. Subscription publications and academic journals are one of the three main content pools a Turnitin scan searches, alongside the web index and the student paper repository [2]. Journal coverage depends on the publisher agreements in the subscription, so it is broad but not exhaustive.
Is a 20% similarity score bad?
Not inherently. Twenty percent means a fifth of your text matched something in the database, and that text may be correctly cited quotations or reference entries [3]. Look at the individual matches before drawing any conclusion.
Will my draft be added to the student paper database if I scan it?
Only if the account running the check is configured to store submissions. A non-repository check compares your document against the database without adding it to the pool [2], so your draft does not become a future match source for other students.
Why did my score change when I scanned the same paper again?
The database is updated continuously as new web pages are crawled and new student papers are submitted, so the comparison set is different each time [2]. Filters and settings applied to the report also change the displayed percentage.
Can I see the exact sentences that matched?
Yes. The report highlights each matched string and lists the source it came from, which is the part instructors actually read [3]. That match list is what tells you whether to paraphrase, add a citation, or leave the passage alone.