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