Almost every UK student has asked it, usually at 1 a.m. before a deadline: “My Turnitin score is 24% — am I in trouble?” The honest answer is: the number alone cannot tell you. The similarity index is one of the most misunderstood figures in university life, surrounded by myths about “safe” percentages that no university actually uses. This guide explains what the score really measures, why there is no magic safe number, what genuinely matters in the report behind the number, and how to lower similarity the legitimate way — by writing better, not by trying to fool a machine.
What the similarity index actually measures
Turnitin compares the text of your submission against an enormous database: published academic work, books and journals, websites, and a vast archive of previously submitted student papers. The similarity index is simply the percentage of your text that matches text in that database, in matching strings of words.
That is all it is: a text-matching measure. Turnitin does not detect plagiarism, does not judge intent, and does not know whether a match is properly quoted and referenced (legitimate) or copied and passed off as your own (plagiarism). A human — your marker, and in serious cases an academic integrity panel — makes that judgement by reading the report. Understanding this one point dissolves most Turnitin anxiety: the score is a screening tool that tells a human where to look, not a verdict.
Why there is no magic “safe” number
You will see websites and forum posts claiming that “under 15% is safe” or “anything below 20% is fine”. These numbers are folklore. In reality:
- Universities set their own expectations, and they vary by institution, department and assignment type. Many publish no fixed threshold at all, precisely because the number is meaningless without context.
- Assignment type changes everything. A reference list alone can contribute several percentage points of matching text. A law essay full of case names and statutory phrases, a lab report with standard method descriptions, or a dissertation with a long bibliography will naturally score higher than a reflective piece — with every match entirely legitimate.
- The report’s settings matter. Whether the bibliography and quoted material are excluded from the score changes the figure significantly; the same essay can show very different percentages under different settings. Comparing your number with a friend’s on another module is comparing different instruments.
So the question “what is a good Turnitin score?” has no single answer — and anyone who sells you one is guessing. The better question is: what does my report actually show?
What matters more than the number: the pattern of matches
Markers look at the report, not the headline figure. What reassures them, and what alarms them, is the shape of the matching:
Usually fine:
- Lots of small matches (1–2% each) spread across many sources — the natural footprint of scholarship: phrases, references, common terminology.
- Matches concentrated in the reference list, quotations, and standard technical or legal phrasing.
- Matches to sources you have cited, used as evidence inside your own argument.
Genuine red flags:
- One source accounting for a large share of the match — whole paragraphs tracking a single article, website or another student’s paper.
- Long unbroken blocks of matched text, even at a modest overall percentage. A 12% score made of three copied pages is far worse than a 30% score of scattered phrases.
- Matches to essay mills, sample-essay sites or uncited sources.
- Matching text with the citations stripped out — your words follow a source’s structure sentence by sentence, with only synonyms changed. This is mosaic plagiarism, and markers are trained to spot it.
Two essays can share the same percentage and be at opposite ends of integrity. Pattern, not percentage, is what you should check.
What markers actually do with your report
It helps to see your submission the way your marker does. They open the report, glance at the percentage, and then — in any case that matters — scroll the match list: the biggest sources, the longest blocks, the places where matched text sits without a citation nearby. Routine, well-referenced work is waved through in seconds, whatever the headline number. Where a report triggers a conversation, it is almost never because a percentage crossed a line; it is because a marker found a passage that raised a question the number alone could not answer. University academic integrity processes are built the same way: the software flags, humans investigate, and the student’s own account of how the work was produced — drafts, notes, reading — becomes part of the evidence. This is one more reason genuine process matters: a student who really researched and drafted their work can show it.
How to lower similarity legitimately
All of the following reduce your score for the right reason — because the work becomes more genuinely yours:
- Paraphrase properly. Read the source, close it, and write the idea from your understanding, in your own sentence structure, then cite it. Changing a few words in the original sentence is not paraphrasing; restructuring the thought is — and it both lowers matches and is what markers actually want. If you cannot explain the idea without the source open in front of you, you do not understand it yet, and that is the real problem to fix.
- Quote sparingly and correctly. Direct quotes should be the exception — a definition, a contested claim, a line you are about to analyse — in quotation marks, cited, with a page number. An essay padded with quotes is both weaker and higher-matching.
- Reference everything you use, accurately. Proper referencing does not remove a match from the raw report, but it converts it from “suspicious” to “scholarship” in the reader’s eye — which is where the judgement is made. Our Harvard referencing guide covers the system most UK students use.
- Write from multiple sources. Work built on one or two texts inevitably shadows them. Reading more widely spreads your matches into the harmless 1–2% pattern and improves the work itself.
- Leave time to draft twice. Most heavy similarity is a time problem disguised as a writing problem: the night-before essay leans on its sources because there was no time to digest them. A rough first draft written from notes, then revised, produces natural, original prose almost automatically.
What not to do
Be wary of “solutions” that promise to beat the detector: synonym-spinning tools, paraphrasing software, swapping characters, translating text back and forth. Three honest warnings. First, they produce garbled, unnatural prose that markers flag on reading, even before any software runs. Second, universities increasingly treat deliberate concealment as an aggravating factor — an attempt to deceive, on top of any original offence. Third, the work itself gets worse: spun text cannot hold an argument, and you learn nothing from it. The only durable way to a clean report is the unglamorous one: understand the material, write it yourself, cite it properly.
If your score comes back higher than you expected
Do not panic, and do not resubmit in a scramble. Open the report and look at the biggest matches one by one. Are they your reference list and quotes? Then you are fine. Is one source dominating? Rewrite those sections from your notes, from understanding, and diversify the sources behind them. Is the match a piece of standard phrasing your field cannot avoid? Leave it — markers in your field see it in every legitimate paper. If your university lets you view the report before final submission, use that window calmly and for exactly this review; that is what it is for.
The bottom line
There is no good Turnitin score, in the way there is no good number of words in a sentence — only good writing, which the report then reflects. Aim for that: your own argument, built from sources you understood, paraphrased in your own words, quoted rarely, and referenced precisely. Students who work that way rarely need to ask what their percentage is.
That is also the principle behind everything we publish and produce: model work written from scratch, designed to be studied and learned from rather than submitted — you can read how we work on our academic integrity page, and see the standard of original writing we mean on our samples page.