Content SEO
Duplicate Content Checker
Paste two pieces of text — a draft vs. the published version, your article vs. a competitor's, or your page vs. a syndicated copy — and see exactly how much overlaps, sentence by sentence.
Matching is word-shingle based (contiguous word sequences), so it works for any Latin-script language. Languages written without spaces between words (e.g. Chinese, Thai, Khmer) get a coarser signal — treat the score as directional rather than exact for those.
What this tool checks
Search engines are wary of pages that substantially repeat content already indexed elsewhere — whether it is scraped, syndicated without attribution, or an old draft that never got rewritten. This tool measures real textual overlap between two documents: an overall similarity score, which individual sentences are word-for-word duplicates or close paraphrases, and the single longest verbatim phrase shared by both.
How the score is calculated
The engine uses w-shingling — the same family of technique search engines and plagiarism tools use for near-duplicate detection — rather than a naive character diff, so reordered paragraphs and minor edits are still caught.
Document similarity (Jaccard on shingles)
Both texts are broken into overlapping 6-word sequences ("shingles"). The Jaccard index — shared shingles divided by total unique shingles — gives an overall resemblance score that is robust to paragraph reordering.
Content overlap (containment)
Jaccard alone under-reports when one text is much shorter than the other (a short quoted excerpt inside a long article scores low on Jaccard even if 100% of the excerpt is copied). Containment measures what share of the shorter text's shingles appear in the longer one, in each direction, catching that case.
Sentence-level matching
Every sentence in Text A is compared against every sentence in Text B using the same shingle-overlap technique at a finer grain, then classified as an exact match, a near-duplicate (paraphrase-level overlap), or unique.
Longest verbatim run
Among matched sentence pairs, the tool finds the longest exact contiguous word-for-word sequence shared between the two texts — a single long verbatim run is a stronger duplicate-content signal than a moderate overall score.
Frequently asked questions
How much overlap counts as "duplicate content" to Google?
Google has never published an exact threshold, and near-duplicate detection is only one signal among many. As a practical guide: overlap in the 30-55% range is worth reviewing, 55% and above (especially with several exact-match sentences or one long verbatim run) is a real risk of the two pages competing with or cannibalizing each other, or of one being seen as unoriginal.
What should I do if two of my own pages score high?
Either differentiate the content meaningfully (different angle, more depth, different intent) and cross-link them, or pick the stronger page and 301-redirect / canonicalize the weaker one to it. Publishing both as near-identical pages competing for the same query rarely helps either one rank.
I am checking syndicated content — is duplication always bad?
No. Search engines handle legitimately syndicated content fine as long as the canonical page is clear — use a cross-domain rel="canonical" tag pointing at the original on the syndicated copy, or ensure the syndication partner does.
Does this tool send my text anywhere?
Both texts are sent once, over HTTPS, to our server to run the comparison (needed to keep the matching logic server-side and consistent), and are not stored beyond the request.