Content Similarity Checker

Compare the right text

Two pages can share a menu, a footer and a contact form without answering the same question. Start with the article or main content, then read the extracted text before interpreting the percentage. A clean comparison of the wrong section is still the wrong comparison.

For a published page, choose Live URL. For an unpublished draft, paste text or HTML. You can mix formats, such as a live article in Source A and its proposed revision in Source B. The checker reads server-returned HTML; it does not run JavaScript or see your logged-in browser session.

Auto selection uses the first main element, then the first article, then the body. You can choose one of those areas yourself. Navigation, headers, footers, sidebars and forms are excluded by default. That can remove useful text when a template uses those elements inside an article, so the Source text view makes the selection visible. Paste the intended copy when the page structure is ambiguous.

Read the match

The percentage measures aligned words in the two selected texts. The calculation is twice the number of aligned words divided by their combined word count. For example, 60 aligned words across two 100-word texts produces a 60% word match. It is a wording measure, not a judgement about quality or meaning.

Punctuation and spacing are ignored. Letter case is also ignored unless you switch that option off. Unicode normalization helps equivalent written forms compare consistently. Numbers remain part of the comparison. The report labels identical normalized word sequences separately; this does not mean their HTML, layout or punctuation is identical.

Word order matters. Moving a paragraph can reduce the alignment even when its wording survives elsewhere. A rewritten paragraph may communicate the same idea with few matching words. For that reason, use the percentage to open a review, not to close it.

Follow the passages

Matching passages show the wording and its word positions in A and B. Raise the minimum passage length to concentrate on longer blocks. This filters the evidence list without changing the underlying percentage. Search within the results when you need to locate a particular claim or repeated sentence.

The coverage figures answer a slightly different question: how much of each source was aligned? A short summary can be mostly contained in a long article while accounting for only a small part of that article. Reading both coverage figures avoids treating a short excerpt and a full page as interchangeable.

Differences lists unaligned and changed spans. Check names, dates, quantities and qualifications carefully. A small change such as adding a negative can reverse a claim while leaving most words untouched. A high wording match is no substitute for that editorial check.

Decide what to change

For two drafts, use the report to check whether a revision added useful information or mainly changed phrasing. For two service pages, look for a clear difference in audience, scope, examples and the decision each page helps someone make. Standard terms or an intentionally repeated quotation may need no change at all.

Google explains that some duplicate content is normal. This tool does not reproduce Google’s canonical selection or determine which URL is indexed. Do not delete a useful page merely because its wording matches another one. Review purpose, performance and technical signals together; use the Canonical Tag Checker when you need to inspect declared canonical information.

This is a two-source comparison, not a web-wide plagiarism search or semantic analysis service. It does not determine authorship, licensing or whether material has been copied without permission. Short sources, common phrases and shared templates can all produce strong matches.

Keep the evidence

Download matching passages as CSV for a review sheet, or save the JSON report for the extracted texts, settings, differences and response context. Reports include the comparison time. The page does not automatically save your drafts, and submitted content and results are not stored in WordPress.

Protected URL requests have size and time bounds. If a source is incomplete, blocked or too large, the checker asks you to choose a narrower area or paste the intended text. It does not label an unread page as unique. Avoid submitting confidential drafts or credentials. Shared abuse protection keeps a salted IP hash, tool identifier and usage date; hosting and verification-provider policies apply separately.

For a single-draft editorial review, use the SEO Content Optimizer. For overlapping query performance across several pages, use the Keyword Cannibalization Tool. Those are different questions from how closely two texts match.

Questions