Our Blogs

Turnitin Similarity Checker: How It Works and What Your Score Means

Turnitin is a text matching tool, not a plagiarism judge. It compares your writing against student papers, journals and web pages, then shows a similarity score as a percentage. The score flags overlap so a person can review it. A high number is not proof of cheating, and a low number is not a free pass. If your university uses Turnitin, your work passes through it the moment you submit, and for a lot of students that is a nervous moment. The score feels like a verdict, and a number you did not expect can ruin your week. The good news is that Turnitin is far less mysterious than it looks. Once you understand what it actually measures, the fear fades and the report becomes a tool you can use rather than a result you dread. This guide walks through what Turnitin is, how it works step by step, what the similarity score really means, and how to read your report with a clear head. What is Turnitin? Turnitin is the similarity checking service that most universities and colleges use to review submitted work. When people say their essay was checked for plagiarism, Turnitin is usually the tool behind that statement. It has been around for more than two decades, and over that time it has built one of the largest databases of academic writing in the world, made up of student submissions, published articles and pages from across the open web. The single most important thing to understand is that Turnitin does not decide whether you cheated. It is not an examiner and it has no opinion. All it does is find passages in your writing that match text it has seen before, then present those matches to your tutor. The judgment about whether a match is acceptable, accidental or a problem belongs to a human being who reads the report and knows the context of your assignment. Turnitin simply points at the overlaps and says, look here. This distinction matters because it changes how you should react to a score. A high number does not mean your tutor has caught you doing something wrong. It means there is more matched text to look at, and the reasons for that could be completely innocent, such as a long quotation or a detailed reference list. Keeping this in mind takes the panic out of the process. How does Turnitin work, step by step? The process is more mechanical than most people imagine. Here is what happens from the moment you hand in your work to the moment your tutor sees a report. Notice that the only automated part is the matching. The interpretation is human. That is why two students can have the same score and very different outcomes, because what counts is not the number but the nature of the matches behind it. For a closer look at how to make sense of the colours and source list, see our guide on Turnitin report colours explained. What does the similarity score mean? The similarity score is the percentage of your writing that matches other sources in the database. If your score is 18%, that means roughly eighteen out of every hundred words sit inside a passage that matched something else. It does not mean eighteen per cent of your work is plagiarised. It means eighteen per cent overlapped with text Turnitin recognised, and some of that overlap is almost always legitimate. There is no universal pass mark, because every institution sets its own expectations, and many do not publish a hard figure at all. As a rough guide, a score under 15% is usually fine, a score between 15% and 25% is worth a closer look, and a score above 25% needs careful review before you submit. Even these bands are only a starting point. What matters far more than the headline number is the spread of the matches. Score What it usually signals Under 15% Normally fine, especially when matches are small and scattered 15% to 25% Review the largest single matches before submitting Above 25% Needs careful checking, though it may still be explainable To see why the spread matters, picture two essays that both score 20%. In the first, the matches are spread across thirty different sources, none more than two per cent each, which is the natural result of citing widely and quoting a little. In the second, a single source accounts for eighteen of the twenty per cent, which suggests one passage was leaned on far too heavily. The first is healthy. The second needs work. The number is identical, but the story is completely different. We unpack this fully in what counts as a good Turnitin score. What does Turnitin compare your work against? Turnitin checks your text against three broad pools of content. The first is its archive of previously submitted student papers, which is enormous and grows every time a student anywhere submits work with archiving switched on. The second is a large collection of published material, including journal articles, books and periodicals that Turnitin licenses. The third is the open web, the pages a search engine could reach. What it does not do is check everything. It cannot see inside paywalled content it has not licensed, it cannot read text trapped in images, and it does not understand meaning, so it will not catch an idea you borrowed and reworded heavily unless the wording itself matches. This is the reason a low score is never a guarantee of originality. You can score zero and still have an uncredited idea sitting in your argument. For the detail on the databases and the gaps, read what Turnitin actually checks against. Does Turnitin detect AI writing? Yes, but through a separate feature. Alongside the similarity report, Turnitin offers an AI writing indicator that estimates how much of your text reads as if it were generated by a tool like ChatGPT. This is a different measurement from the similarity score and it answers a different question.