Direct answer
A plagiarism checker is a text-matching tool: it compares your document against a database of web pages, academic publications and previously submitted student papers, then returns a similarity report listing every passage that overlaps with a source [1]. The percentage it produces is an investigative signal, not a verdict — matches must be read in context, because correctly quoted and cited material also appears in the report [1]. For a university draft, the useful question is not "which tool has the biggest database" but "does this tool return the same kind of similarity report my institution will run, and can I see it before I submit?"
How Does a Plagiarism Checker Actually Detect Copied Text in a Document?
Detection is mechanical text comparison, not comprehension. A checker segments your document into sentences and phrases, indexes them, and looks for identical or near-identical strings in its corpus of web content, journal and book databases, and prior student submissions [2]. When a string match clears the tool's threshold, the passage is highlighted and the source is attached to it. Nothing in that process knows whether you meant to copy — which is exactly why the report is evidence to be interpreted rather than a score to be obeyed [2].
That limitation shapes what the tool can and cannot catch. Direct copy-paste is easy to flag; paraphrased and mosaic plagiarism — where sentence structure and vocabulary are swapped but the argument is lifted — is far harder, because the words no longer match the source string [2]. A low similarity score therefore does not prove originality, and a high one does not prove misconduct.
The database is the other half of the equation. A checker can only match what it has indexed, so coverage differs sharply between tools: open web crawls, subscription journal indexes, and institution-specific student paper repositories produce very different results on the same draft [2]. This is why two checkers can return 8% and 31% for one essay without either being "wrong" — they searched different corpora.
Finally, legitimate overlap is normal and expected. Quotations, terminology, standard method descriptions and reference lists all match sources by design, and a well-built report separates those out so you can see what remains [2]. Reading the matched text, not the headline number, is the actual skill.
What Should a Similarity Report Look Like, and How Do I Read the Percentage?
A standard similarity report has three layers: an overall similarity index, a match overview listing sources in descending order of matched volume, and in-text highlights colour-coded to each source [3]. The index is the share of your submission that matched something in the database — and institutions commonly configure it to exclude the bibliography and quoted material, so the same document can show different numbers under different settings [3]. Always check which exclusions are switched on before you compare two reports.
Context decides whether a number is a problem. A 30% score driven by a long block-quoted literature review and a standard ethics statement is a very different thing from a 30% score driven by three unmarked paragraphs from one journal article [3]. The match overview tells you which of those you are looking at, and the highlighted text tells you whether the borrowing was cited.
Two practical habits make the report usable. First, work source by source rather than chasing the total: open each match, decide whether it is quoted, cited, common knowledge or genuinely unattributed, and fix only the last category [3]. Second, treat the report as a drafting tool — the value is in the revision it triggers, not in the score itself, and a report reviewed two days before a deadline is worth far more than one reviewed two hours before it.
Can Students Check Their Own Draft Before Submitting Without It Being Stored?
Yes, in principle — but the answer depends entirely on which system you use and how it is configured. In institutional Turnitin setups, whether a student paper is retained in the standard repository is an administrative setting chosen by the institution, not something the student controls [4]. Draft checking and formal course submission are separate events, and running a check on a draft does not itself count as handing the assignment in [4].
That distinction matters for two reasons. Repository retention affects future students: a paper stored in the database becomes a match source for later submissions, including your own resubmissions, which can inflate your similarity score the second time you check the same file [4]. And institutional policy varies — some departments explicitly permit draft checks through the official route, others restrict them, so confirming the local rule before you upload anything is basic due diligence [4].
If you use an independent checking service instead, ask one direct question before paying: is the file added to a student paper database? A non-repository check returns the same style of similarity and AI reports without turning your draft into a future match source, and you should also be able to delete the file from your account afterwards. Combined with a turnaround measured in minutes rather than days, that is what makes a pre-submission check genuinely useful rather than a gamble on your own record.
Reading the report is the easy part; getting a real one on your own draft before the deadline is where students get stuck. That is the gap turnitin0 was built for: upload a .docx, .pdf or .txt file of 300–30,000 words, and within roughly 5–15 minutes you receive the same two PDFs your professor sees — a similarity report with highlighted matches and an AI writing report — generated without adding your file to Turnitin's student paper database, with no subscription and files you can delete from your account.
※ Turnitin0.com - Actual Turnitin AI Report Cover, Score, Flag And Similarity Summary
FAQ
Is a high similarity percentage automatically plagiarism?
No. The index counts matched text, and quoted, cited, or standard-phrase matches are counted until your institution's exclusions remove them [3]. Open the match overview and judge each source on whether the borrowing was attributed.
Which database does a plagiarism checker use?
It varies by tool and is the main reason scores differ between checkers. Turnitin-style systems compare against web content, publications and previously submitted student papers, and institutional settings determine whether student work is added to that repository [1][4].
Can I check my own draft without it being submitted to my course?
Yes — a draft check and a course submission are separate actions [4]. Confirm your department's policy on draft checks, and if you use an outside service, verify that it does not add your file to a student paper database.
Why did my similarity score go up when I re-checked the same essay?
Because a stored copy of your earlier upload can become a match source for the later check [4]. Re-checking a revised file through a non-repository service avoids that self-match effect.
Does a plagiarism report also tell me if my writing looks AI-generated?
Not usually — similarity and AI writing detection are separate analyses, and a draft can score low on similarity while still being flagged for AI use. Pre-submission checks that return both reports together give you the fuller picture before you submit.