See also: 

Complainant must be made aware that matter cannot be investigated unless at some point the journal Editor informs the corresponding (or complained-about) author (due process)

First stage must be a simple comparison of the relevant (two) texts ...

- a simple side-by-side comparison by the Editor for the simpler forms of plagiarism - a more thoughtful analysis by the Editor if paraphrasing or types of ‘self-plagiarism’ are alleged (see   Legal Guidelines for Editors Concerning Ethics Issues)

If the Editor reasonably determines that there is significant overlap of text...

... then the Editor should correspond with the corresponding (or complained-about) author—in some cases for confidentiality purposes it may be best for the Editor to summarize (rather than simply forward) the complaint

It may be useful for the Editor to involve other peer reviewers, editorial board members, or experts in the relevant field (anonymously as to the complainant and if possible as to the identity of the complained-about authors), using standard peer review procedures, to review the texts (especially if the allegation is a more complex form of plagiarism)

Legal review may be appropriate if the complainant or their publisher is alleging copyright infringement (Elsevier legal will provide this)

In the communication to the corresponding/complained-about author (see   Form Letter A), the Editor should indicate that it is possible that the matter may be referred to institution or company where research took place or any other relevant institution or agency (for example a funding agency) unless the author provides a reasonable explanation (accepted as reasonable by Editor)

Note that the Editor may believe that referral to the institution or agency is not necessary (unlike with respect to claims about authorship or fraud, where the institution has responsibility for the conduct at their institution and an obligation to investigate, plagiarism may simply be a mistake - perhaps a type of unattributed copying - or may be considered to be the personal responsibility of the author rather than an institutional responsibility)