A file goes off to be checked exactly as you sent it. Nothing is stripped out first, nothing is cleaned up, nothing is ignored for being obviously administrative. If your name is on page one, your name is part of the document being checked.
That single fact explains every item on this list.
Take the cover page out
A cover page is the worst offender, because it is made almost entirely of text that appears on thousands of other cover pages: the university name, the department, the module title, the words "submitted in partial fulfilment of the requirements for". All of it matches something. All of it counts towards your similarity score.
Worse, it does nothing for you. Nobody is checking your cover page for originality. It inflates the number and tells you nothing.
Start the file at the abstract, or at the first real paragraph of the work itself.
Take your name, ID and institution out
Your name, student number, supervisor's name, department, university — none of it belongs in a file you are checking.
Partly it is the same problem: institution names match other papers. Mostly it is a better reason. A check is a look at your writing; there is no reason to attach your identity to it. We delete the file and both reports 24 hours after your results are ready, and we never add anything to a plagiarism database — but the best way to keep a detail private is not to send it.
Empty the header and the footer
Running heads repeat on every page. Rahman, Consumer Behaviour, Module MKT-403 at the top of forty pages is that phrase counted forty times, and it drags the percentage up for no reason at all.
A plain page number is fine. Words are not.
Leave out the front matter
Declaration, acknowledgements, certificate of approval, letter of recommendation, dedication, table of contents. These are conventional pages made of conventional sentences, which is a polite way of saying they match everything.
Nobody assesses the originality of "I hereby declare that this thesis is my own work." Cut them and start at the abstract.
Leave the appendix out too
Questionnaires, transcripts, tables of raw figures, published instruments you reproduced — an appendix is often the largest match in a whole thesis, because much of it is quoted or standard by definition.
Check the body of the work. The appendix will still be there when you assemble the real submission.
Delete blank pages, especially at the start
Several empty paragraphs or a page break before the writing begins can leave the check with nothing to read at the start of the document, and occasionally with nothing to read at all. Take them out.
Watch the word count for the AI report
An AI-writing result only comes back for files between 300 and 30,000 words. Below that there is not enough text to judge; above it the file is too long for the model to process in one piece. Outside that range the similarity report still arrives, but the AI box will be empty.
If the work is longer than 30,000 words, check it in parts — by chapter is usually the natural split.
Make sure the text is really text
If your file is a scan, or a PDF made of page images, there is no text in it to read. The check comes back saying it could not be read, and you have a report about nothing.
A quick test: open the file and try to select a sentence with your mouse. If nothing highlights, it is a picture of writing rather than writing. Export it again from Word as DOCX or PDF, or run the scan through OCR first.
What happens if you forget
We read DOCX files before anything is sent and tell you what we found — a cover page, a name on page one, words in the header, a word count outside the range — while the slot is still yours. You can fix the file and try again, or send it as it is if you know better than we do. PDFs we cannot read inside, so with those the list above is on you.
None of this is about rules. It is about getting a report that answers the question you actually have — is there anything in my writing I need to fix? — rather than a number made mostly of your own cover page.
The checklist
Before you upload:
- Cover page removed
- Name, student ID, university and department gone
- Header and footer empty except page numbers
- Declaration, acknowledgements, contents and dedication cut
- Appendix left out
- No blank pages at the start
- Between 300 and 30,000 words, or split into parts
- Text you can select with the mouse
- File name short, ending in .docx or .pdf