What a Turnitin Similarity Report Actually Shows (and Why You Cannot Buy One)

A Turnitin Similarity Report lists the passages in your document that match text Turnitin already holds, and names a source for each one. The percentage at the top is the share of your words that matched. It is not a verdict on plagiarism. Correctly quoted and correctly cited material counts as matched text too.

Ask ten students what that report is and nine will tell you it is the thing that says whether you plagiarised.

Turnitin says otherwise. Its own guide for students opens with a sentence that has no hedging in it at all.

"Turnitin does not check for plagiarism"

That is a quotation, not a criticism. The student guide to the similarity score begins:

Turnitin does not check for plagiarism. What we actually do is compare your submission against our database. Our database includes current and archived content from the internet, thousands of periodicals, journals, and publications, as well as a repository of works other students have submitted to Turnitin. We then highlight wherever your writing is similar to text from any of our sources.

Every misunderstanding about the report comes out of that paragraph. The software finds text that resembles other text. Whether a resemblance is a citation, a coincidence, or theft is a judgement it never claims to make. The report is evidence assembled for a person who will make that judgement, and it is built to be read rather than scored.

What Turnitin compares your document against

Three kinds of content, according to the same guide: current and archived internet pages, periodicals and journals and publications, and a repository of work other students have submitted. Turnitin publishes counts for each of them. Those counts are not repeated here, because the figures on Turnitin's own pages disagree with one another and some of them are several years old. Treat any specific number you meet as an order of magnitude.

The third kind is the one that matters. Any serious checker crawls the web, and several license journal content. Nobody else holds the student archive, because it is gathered from the institutions that run Turnitin, and Turnitin describes its index of student papers as the largest there is. That archive is why a match can point at an unpublished thesis submitted in another country six years ago. It is also a large part of why no outside tool can reproduce a Turnitin score, and why handing your draft to a stranger's account is a worse idea than it looks.

What a Turnitin similarity score actually measures

The similarity score is the share of your words that matched something in the database. That is all it is. Turnitin is explicit that correctly quoted and correctly referenced text still counts as matched text, so a well-cited literature review will always carry a score.

Two edge cases are worth knowing, because they look like faults and are not:

• A 0% score can still come with matches in the report. On a long enough document, a small quantity of matched text rounds down to zero.
• A 100% score is not proof of anything on its own. Turnitin's own list of possibilities includes someone else having submitted your essay before you did, and your own earlier drafts sitting in the repository.

The colour of the score is a display band and nothing more: blue or green at the low end, then yellow, orange and red as the percentage climbs. It marks how much text matched. It does not grade you.

On the question everyone actually asks, Turnitin's answer is that there is no fixed number to aim for, and that a zero may itself be a problem, because it can mean you cited nothing. Your institution sets the figure that matters. What similarity percentage is safe works through one national policy that sets two limits rather than one, and how each of them is applied.

How to read a Turnitin Similarity Report

The percentage is the least informative part of the report. These are the parts that repay reading.

Match Overview

A list of every match found, ordered from the largest to the smallest. Each entry carries a colour and a number, and the same colour and number appear on the matching text in your document, so you can move between the list and the page.

All Sources

Every source matched to your submission, sorted by percentage. Beside it sits a match breakdown: open a source from the overview and you are shown the other sources carrying that same passage. A sentence can appear in twenty places on the web, and this is where you find out that the source credited at the top is not the one you actually borrowed from.

Source details

Selecting a source tells you how many times it matched and walks you through each instance. For a web page or a journal article you can open a full source text view and read the original. For a match against another student's paper you cannot: that view is unavailable. You are told the match exists and you do not get to see it, which is the correct way to handle someone else's coursework and also the most frustrating result a student can get.

Match Groups

This one appears only in the newer version of the report, and it does the most work of anything on the screen. It splits the similarity into four categories based on how the matched text is punctuated and cited:

• Not Cited or Quoted
• Missing Quotations, meaning cited, but so exact that quotation marks are probably needed
• Missing Citation, meaning quoted, but with no source given
• Cited and Quoted

That is the breakdown a student wants, because it separates the matches that are working as intended from the ones that need an afternoon of work. If your report has this view, start there and leave the front page alone.

Flags for review

On any document of at least 500 characters, spaces not counted, Turnitin also looks for signs that the text has been manipulated to defeat matching. Two kinds are named in its guidance. Replaced characters are look-alike letters borrowed from other alphabets, which Turnitin swaps back before scanning. Hidden text covers tricks such as white characters inserted between words so that a sentence reads as one long token. Turnitin notes that a flag is not necessarily a problem, but it is the first place a suspicious marker will look.

Why the AI writing score is not in the Similarity Report

The AI writing indicator is a separate number. Turnitin's own wording is that the percentage its AI model produces is different from and independent of the similarity score, and that AI writing highlights are not visible in the Similarity Report at all. Only instructors and administrators can see the indicator, and Turnitin states flatly that the indicator and its report are not visible to students. An instructor can choose to download the AI report and share it with you. Nothing you open yourself will show it.

This matters when somebody offers to sell you "your Turnitin AI score". Even a genuine student view of a genuine Turnitin report does not display that figure. If your worry is being flagged as AI, the mechanics are a separate subject, covered in how to check if an assignment is AI-generated.

Why the same document gets two different similarity scores

Here is the part that explains everything else. The similarity score is not a property of your document. It is a property of your document and a settings panel you have never seen.

Your instructor chooses which repositories the submission is compared against: the global student paper repository, your institution's own repository, current and archived internet, publications, or some subset of those. Turnitin states the consequence plainly. Deselect a repository and the score may fall. Your instructor also chooses the exclusions.

The bottom two rows exist only in the newer version of the report. On the classic version the instructor gets a blunter instrument instead, one that drops a whole source when its total contribution falls below a size they set.

Then there is storage. The instructor decides whether the paper goes into the global repository, into the institution's own repository, or into neither. They also decide whether you are allowed to see the report at all.

The same 8,000 words submitted to two classes in one department can therefore come back at 9% and at 22%, with nothing dishonest happening anywhere. One assignment excluded the bibliography and the quotes. The other did not.

There is a further trap for anyone reading their own report. If you apply the bibliography or quote filters yourself, Turnitin is explicit that your changes are temporary and do not alter the official score your instructor sees. It is a good way to understand your own document. It is not a way to fix your number.

Can you buy a Turnitin report as a student?

Turnitin's help centre answers this directly, on a page titled Can I purchase an individual Turnitin license?:

Turnitin does not sell individual licenses or single-use subscriptions. Our Similarity detection and grading services are sold exclusively at the institutional level to schools, universities, and organizations.

The same page names the one thing an individual can buy: iThenticate, where independent users purchase credits to check their own documents. That is a real Turnitin product and it is genuinely purchasable. It is also built for researchers and publishers rather than for coursework, and it is still not the report your department runs on your submission under your department's settings.

So every site selling "a Turnitin report" for an assignment is running your document through an account that belongs to someone else. Three things follow, and only the first one usually gets mentioned.

Your paper may end up in the repository, and you cannot get it out. If that account stores submissions, your draft joins the student archive. When you submit the real thing weeks later it can match itself, and Turnitin lists that exact shape in its own guidance: send multiple drafts to a repository and the final one comes back at 100%. Removal is where this turns bad. Turnitin will not accept a deletion request from a student, for contractual and data privacy reasons, and routes every request through your institution. You hand your instructor the submission ID, the class name and the assignment name, and an administrator acts on it. For a paper submitted by a website through somebody else's account, none of those three things is yours to give.

The score came out of settings you cannot inspect. You have no way to know which repositories were selected or which exclusions were switched on. A number produced under unknown settings cannot predict a number produced under your department's settings. It is a figure with no denominator.

Your unsubmitted work went to a stranger. One of the scenarios Turnitin lists for a 100% score is somebody acquiring a copy of your document and submitting it before you do. That is the risk you take on when you email a thesis to an account you cannot identify, and the consequence lands on you rather than on them.

None of that argues against checking your work before you hand it in. It argues against buying a fake of one specific report. An independent plagiarism check makes no claim to be Turnitin, never puts your draft into a repository Turnitin will read, and publishes its rates up front. How it works is the whole process in four steps. The rest of this article is what your own institution gives you first, and then what an outside check can honestly add.

How to check your Turnitin score before you submit

There are three, and Turnitin names all of them. Which ones you have depends on your institution and on the assignment.

1. Draft Coach, if your institution has enabled it. This is the one most students have never heard of. It runs Similarity, citation and grammar checks on your draft inside Google Docs or Microsoft Word, before you submit anything. Ask your library or your instructor whether you have it.
2. An assignment that allows resubmission. Without Draft Coach, your only route is submitting a draft to an assignment that permits resubmissions, where each upload overwrites the last. Turnitin does throttle this. On classic assignments the first three attempts return a report immediately and the fourth onwards makes you wait 24 hours; on standard assignments you get three resubmissions in a 24 hour period and then wait for the next day. Three honest revisions are plenty. Twenty attempts at gaming the number are not available to you.
3. A practice assignment. If resubmission is switched off, your first attempt is final, and the only remaining option is to ask your instructor to set up a separate temporary assignment for checking. People do say yes to this. Very few students ask.

When you do get a report in front of you, read it in this order. Open Match Groups, or All Sources on the classic report, before you look at the percentage, because the front page tells you least. Apply the bibliography and quote filters to see what the body of your text matches on its own, remembering that this changes your view and not your score. Then read every match above a couple of percent and be able to say in one sentence why each one is there.

What an independent plagiarism check can and cannot do

If your university submits to Turnitin, that report is the one you are judged on, and only your institution can produce it. Nothing on this site changes that, and any tool claiming otherwise is claiming access it does not have.

An independent check is a different instrument with a narrower job. Ours cannot see the student paper repository, so a match against an unpublished thesis is invisible to it. It has no bibliography exclusion and no per-source totals, so any rule your department sets about a single source is theirs to measure rather than ours. The database is different, so the figure will not be the same as theirs.

What it does give you is an overall similarity score, your full document with the matched passages highlighted in it, and a numbered card for each match carrying the source link and a similarity figure for that passage. In plain terms: which paragraphs are matching, and where each one came from.

That is the part that carries across. A department's exclusion settings change the percentage. They do not change the fact that your third paragraph is four sentences out of a 2019 conference paper. Fix the paragraphs and the institutional score follows.

Our plagiarism check exists for that window, and the timing is most of its value. Run it while there is still a week to act on the answer, never the night before. Pay per document, read what comes back, then go and rewrite the paragraphs it points at. Each check is its own purchase, so budget for two if you intend to look again after revising.

Frequently asked questions

What is a good Turnitin similarity score?

Turnitin's own answer is that there is no fixed number to aim for, and that a zero may itself be a problem, because it can mean you cited nothing. The figure that matters is your institution's, and some policies set two: an overall limit, and a cap on any single source. What percentage is safe works through one of them.

Why did my similarity score change between two submissions?

Because the score is not a property of your document alone. Your instructor chooses which repositories the comparison runs against, and which exclusions are switched on for bibliographies, quotations, citations and small matches. Change the settings and the same words come back at a different percentage. Nothing about your writing has to change at all.

Can I buy a Turnitin report for my assignment?

No. Turnitin's help centre says it does not sell individual licences or single-use subscriptions. Any site offering one is running your document through somebody else's institutional account, which risks your draft entering the repository and produces a score from settings you cannot inspect.

Does the Turnitin report show my AI writing score?

No. The AI indicator is a separate number, it is visible to instructors and administrators rather than to students, and its highlights never appear in the Similarity Report. An instructor can download the AI report and share it with you. How to check if an assignment is AI-generated covers what that number measures.

Does a high similarity score mean I plagiarised?

Not on its own. Correctly quoted and correctly cited material counts as matched text, so a well-referenced literature review always carries a score. What matters is the shape of the score: how much came from any single source, and whether the matches are quoted, cited, both or neither. If your report has the Match Groups view, read that before the front page; on the classic report, All Sources is the nearest equivalent.

Can I check my document before I submit it?

Sometimes, through your own institution: Draft Coach if it is enabled, an assignment that allows resubmission, or a practice assignment your instructor sets up. Failing those, an independent plagiarism check gives you an earlier read on which paragraphs match, and our FAQ covers how a check is run. It will not reproduce your department's figure. Best plagiarism checkers compared sets out what the alternatives cost.