16 September 2026 · 8 min read
A plagiarism report is not a percentage. The percentage is a summary of a document that runs to several pages, and the pages are where the useful information is: which passages matched, what they matched against, how long each match is, and what the report was configured to ignore. Students who only ever see the number are reacting to the least informative part of it.
The layout differs between platforms, but every similarity report carries the same four things, and it is worth knowing what you are looking at before you panic about the headline figure.
Read the source breakdown first, not the number. A 24% made of forty sources at under 1% each is a normally referenced document. A 12% where one source contributes 11% is a single lifted passage, and it is the second one that gets referred.
Before changing a single word, classify. Most students start rewriting at the top of the report and waste their time on matches that were never a problem.
That fourth group is where the actual risk lives, and it is where an examiner looks. One uncited paragraph following a source sentence for sentence is a more serious finding than a high percentage spread across correct citations, whatever the two numbers look like side by side.
Rewriting means rebuilding the sentence from the idea. Read the passage, close the source, write what it said in your own words and your own structure, then cite it. Text produced that way is genuinely yours. Text produced by swapping synonyms with the source open is structurally identical to the original, still matches conceptually, and now carries a second problem: paraphrasing tools are generative, and their output has its own signature on the AI indicator.
Paraphrased material still needs a citation. Students often assume that once wording has been changed the source no longer needs crediting, which is precisely the misunderstanding that produces a low similarity score and an academic conduct query at the same time.
Text carried from your approved proposal into the final thesis will match the proposal if the proposal was submitted to the same repository. It feels like it should not count — it is your writing — and some departments treat it seriously anyway. The same applies to material reused between two assignments on different modules. Check the assembled document rather than finding out what your department thinks of self-matching at submission.
A similarity report finds text that matches its database. It cannot tell you whether an idea was taken without credit, whether a translated passage came from somewhere, or whether a citation supports the claim attached to it. Nor is it a judgement: it is evidence a human being interprets, and the human being in question is your supervisor. A low number is not a clearance and a high number is not a verdict.
This is exactly why we issue the report and the explanation rather than a figure. We do not promise a percentage, we do not sell a way around a detector, and on a well-referenced draft the honest answer is frequently that nothing needs doing at all.
Related:
Send the report you were given, with the document, and we will go through it passage by passage and tell you which matches actually need your attention.