What it actually does
It matches text, and only text. The service compares a manuscript against large bodies of content and returns every passage that also appears somewhere else, with the source and the extent of each match. It makes no judgement about whether a match is a problem.
The content it compares against is the reason journals use this tool rather than a free web checker. Alongside indexed web pages it covers published scholarly literature deposited by publishers — in the Crossref Similarity Check arrangement, full text contributed by participating members — which is precisely the material a free tool cannot see.
It does not detect plagiarism, and the distinction is not pedantic. Plagiarism is a judgement about intent and attribution that a human makes. iThenticate finds overlapping strings. A correctly quoted, correctly cited passage matches exactly as strongly as a copied one.
iThenticate reports text overlap. Whether overlap is plagiarism is a judgement a person makes afterwards.
iThenticate is not the same as Turnitin
Both are Turnitin products, and they are frequently confused. Turnitin as most people meet it is the student-assignment product used inside universities, matching against a large repository of previously submitted student work.
iThenticate is the research and publishing product. Its comparison set is weighted toward published scholarly literature, and its workflow is built for manuscripts, grant applications and theses rather than coursework.
The practical consequence: a Turnitin report from your university is not the report a journal will see, and the two can return noticeably different figures on the same document. Neither is wrong — they are searching different collections.
Who can run a check, and when
Journals and publishers access it as part of editorial screening. Where a journal is a Crossref member using Similarity Check, the report is generated as part of the submission workflow, and authors generally never see it unless something in it prompts a query.
Many universities hold an institutional licence covering theses, grant applications and staff manuscripts — usually administered through the library or research office. This is the cheapest route by a distance, and a large number of researchers who pay for a check already had access to one.
Individual subscriptions are sold directly to researchers who have no institutional access, and a number of editing and consultancy providers — ourselves included — run a check as part of a pre-submission service. Fees and plans change, so check the current terms at the source rather than relying on a figure quoted elsewhere.
What to do with the report before you submit
Run the check early enough to act on it. A similarity report is useful in the week before submission and useless after acceptance, when a query from an editor has already become a problem to explain rather than a paragraph to rewrite.
Read the matches rather than the number. Open each substantial match, see what it is, and decide: correctly quoted and cited, standard methods wording, your own earlier paper, or genuinely a passage that needs rewriting and attributing.
Fix what needs fixing and leave what does not. Rewriting a correctly cited quotation to reduce a percentage makes the paper worse, not safer — and the number was never the thing being assessed.