← Back to Blog

How to Reduce Similarity Score: 5 Proven Steps to Get Below 15%

Target: Below 15% total similarity for most Q1 journals, below 10% for high-impact journals. These 5 steps address the sections responsible for 90% of similarity issues — in order of impact.

If your iThenticate report shows a similarity score above 15%, the path to fixing it is systematic — not random. Different sections of your manuscript contribute differently to similarity, and different techniques work better for different source types. This guide gives you the highest-impact fixes in the right order.

Step 1 — Fix self-plagiarism from prior papers Highest impact

Self-plagiarism from your own prior publications is the highest-impact source of elevated similarity. The most vulnerable sections are methodology (reusing validated instruments) and literature review (reusing background from prior papers).

How to fix: Identify which prior paper is being matched. Rewrite the methodology section from a different angle — instead of describing what you did, explain why you chose each method. For the literature review, reorganize the structure around your new research question rather than copying your prior framing.

❌ Self-plagiarism (copied from prior paper)

Data were collected using a structured questionnaire distributed to employees at three manufacturing firms. The questionnaire consisted of 45 items measured on a 5-point Likert scale.

✅ Rewritten

The measurement instrument comprised 45 items across six constructs, each rated on a five-point scale anchored by "strongly disagree" and "strongly agree." Items were adapted from validated scales (see Table 1 for sources). Three manufacturing firms in the automotive sector provided access to their employee populations for data collection.

Step 2 — Rewrite your literature review framing Moderate impact

Literature reviews have high similarity because many papers describe the same foundational studies using similar language. The fix is not to avoid citing these papers — it's to describe them from your analytical perspective rather than summarizing their content.

How to fix: Instead of summarizing what each paper found, organize your literature review around themes or debates. Write "Studies examining X consistently find Y (Smith, 2020; Jones, 2021; Lee, 2022)" rather than three separate paragraphs summarizing each paper.

Step 3 — Eliminate boilerplate academic phrases Lower impact

Certain phrases appear in thousands of academic papers and contribute to similarity scores. Replace these with more specific, original phrasing.

  • "This study aims to investigate" → "This study examines [specific relationship] in [specific context]"
  • "The findings of this study contribute to" → "This study extends [theory name] by showing [specific contribution]"
  • "Future research should examine" → Remove entirely from body text; keep only in limitations section
  • "It is worth noting that" → Delete and make the point directly
  • "As shown in the literature" → Cite the specific literature instead
Step 4 — Handle your preprint match Variable impact

If you posted a preprint on SSRN, ResearchGate, or arXiv before journal submission, iThenticate will match your manuscript against your own preprint — potentially adding 10–30% to your similarity score. This is technically self-plagiarism.

Options: (1) Substantially revise the manuscript before journal submission — 40%+ changes reduce the match significantly. (2) Remove the preprint before submission if the platform allows it. (3) Disclose the preprint in your cover letter — many journals accept this with proper disclosure.

Step 5 — Configure iThenticate exclusions correctly Quick win

iThenticate has configurable exclusions that most authors don't know about. Ensure these are enabled before your final check:

  • Exclude bibliography/references
  • Exclude quoted material
  • Exclude small matches (under 8 words)

If your journal includes references in the similarity check, a paper with 70 citations can show 5–8% similarity purely from reference formatting — before any content is matched. Knowing whether your journal excludes references changes your target threshold.

Sections by Similarity Contribution

Across typical manuscripts, similarity tends to concentrate in a few sections rather than spreading evenly. The rough distribution below reflects common patterns, not a controlled measurement — your own report may differ:

Focus your rewriting effort on the literature review and methodology first — these two sections alone account for 60–80% of total similarity in most manuscripts.

Identify Your Highest-Similarity Passages

ScholarAI highlights the specific passages driving your similarity score and suggests targeted rewrites — so you fix what matters, not everything.

Analyze My Similarity Risk Free →