Key takeaways
- Turnitin matches text strings against billions of sources—it does not judge plagiarism directly.
- The similarity percentage is the proportion of matched text, not a guilt score.
- Understanding the process helps you fix problems before submission.
Turnitin is the most widely used text-matching system in higher education. Understanding how it works removes much of the anxiety around similarity reports and helps you respond to flagged content intelligently.
The submission process
When you upload a document, Turnitin converts it to plain text and breaks it into searchable segments. It then compares those segments against its index of web pages, published articles, books, and previously submitted student papers.
What the similarity score means
The overall percentage represents how much of your document matched something in the database. A 30% score means roughly 30% of your text found a match—not that 30% is plagiarised. Properly cited quotations still count.
How matches are displayed
Each match is colour-coded and linked to a source. You can click through to see the original and compare wording side by side. Filters let you exclude quotes, bibliographies, and small matches.
What Turnitin does not do
- It does not determine intent or whether plagiarism occurred.
- It does not replace human review by supervisors or examiners.
- It cannot detect ideas—only overlapping text.
- Paraphrasing that changes every word may not match even if the idea came from a source.
AI writing detection
Recent Turnitin versions include AI-writing indicators alongside similarity. These are probabilistic estimates, not definitive verdicts, and institutions vary in how they use them.