Turnitin Similarity Filters and Exclusions: What Changes Your Score — and Who Can Do It

The similarity score is not a fixed number stamped onto your paper — it is a view of matches that can be refined. Turnitin's own guide describes filters and exclusions that exclude small matches, quoted text, bibliographies, and entire sources, and the score regenerates and recalculates after those exclusions are applied. One important boundary: these controls are on the instructor's side of the report. You do not have a button that removes matches from your own report — and Turnitin's guide says explicitly that exclusions "don't replace human judgment." Understanding the mechanics is not the same as being able to change your own number.

The report exists to help an instructor review matches — not to announce a verdict. That is why Turnitin built filters and exclusions into the report itself. From the official guide:

"Using Filters and Exclusions will help you refine a Similarity Report to only include matches that are irrelevant to you."

The same guide states the practical consequence: "Applying filters and exclusions may or may not change the overall similarity percentage depending on several factors." That means the number you see on submission day is one version of the report — an instructor can produce a more refined version, and the percentage can change.

The most common source of "why is a 5-word phrase counted?" is the small-match category. Turnitin's guide describes the exclusion directly:

"Excluding small matches allows you to exclude matching against sources with a small amount in common with the submission. For example, it's possible to exclude matches fewer than 20 words in total, or less than 10% in total."

The next sentence explains how the percentage is calculated:

"The percentage is based on the total word count of the student's submission. For example, if a student submitted a 5,000 word submission, and you have this set to 10%, it will exclude sources of less than a 500 word match."

So a five-word phrase sits at the bottom of the size scale. It can be a match — and it is precisely the kind of match that an exclusion threshold can remove, because it is below the cutoff. Whether the instructor applies such a threshold is their choice; the default report includes the match.

The second category is the one students ask about most: properly formatted quotes and reference lists showing up as matches. The guide says:

"Exclude quoted text, bibliographic content, and reference materials from within a Similarity Report to remove those matches from the report. This can possibly affect the score percentage, and help provide a more focused assessment."

In the Similarity product version, the same idea is stated with the reasoning:

"Exclusions hide specific types of text that shouldn't count toward similarity, such as properly formatted quotations, bibliographies, or in-text citations. Exclusions help make similarity scores fairer and more accurate by ignoring expected or properly referenced text."

This is why a reports page can look alarming ("my whole references page matched!") while the instructor's refined view is much smaller. The matches are real — the report simply can exclude expected text when reviewing.

Beyond content types, a whole source can be removed from the calculation. The guide describes it as:

"Excluding a source removes an entire source and all associated matches from the similarity calculation, preventing any of its text from contributing to the overall similarity percentage."

And a single match can be removed the same way:

"Excluding a match removes a specific text match from the similarity calculation, so it no longer contributes to the overall similarity percentage."

After any exclusion, the report is regenerated: "Select Apply Changes … to confirm and regenerate the report with your change now reflected in the assignment." The percentage you see after that point is the refined view — not the original submission view.

One detail from the guide connects directly to resubmission behavior:

"If a student resubmits an assignment, their original assignment will show up under 'excluded sources' in order to prevent matching against the original draft."

This is the same self-match protection described in Turnitin's matching help center: your later draft does not match your earlier one because the earlier one is treated as excluded. Instructors can restore excluded sources from the report's Excluded Sources list — including, if they choose, the self-match they normally would not want to see.

The same official guide states the boundary in one sentence:

"While exclusions refine the process, they don't replace human judgment; it's still important to review the remaining matches to determine whether they reflect proper attribution or potential plagiarism."

For you as a student, this is the whole point. The report is an investigative tool for your instructor, not an output you control. The takeaway in practice:

• You cannot exclude matches from your own report — the controls are on the instructor's side.
• A refined report (after exclusions) is not a "pass" — it is a clearer view of what the instructor wants to review.
• If a match bothers you — in the bibliography, in quotes, or as a tiny phrase — the right move is to ask your instructor about their review view, not to try to remove it yourself.

The similarity score you see is one version of the report. Turnitin's filters and exclusions can remove small matches (below a word-count or percentage threshold), quoted and bibliographic text, and entire sources — and the score recalculates after each change. Those controls belong to the instructor, and the guide's own words are the boundary: exclusions "don't replace human judgment." If you see matches in quotes, references, or short phrases, they are real matches that the report is designed to be able to exclude — you just are not the one holding the controls. Read the report, understand the mechanism, and ask the question that matters: how does your instructor review it?