I built this after reading a lot of threads from people fighting Word's mail merge: Word for the web doesn't have it at all, and desktop Word's "Finish & Merge" gives you one long document rather than a separate, sensibly named file per record. The usual workarounds are VBA macros, Graham Mayor's add-in (Windows desktop only), or Google Docs add-ons that need your data in Drive.

DocPerRow does it client-side: you write {Column name} placeholders in a .docx, drop in an xlsx/xls/ods/csv, pick a file-name pattern like {LastName} {InvoiceNo}, and get a ZIP with one .docx per row. It's a static site using docxtemplater, PizZip and SheetJS. Your files never leave the browser. The only outgoing requests are cookie-less GoatCounter analytics (page views and event names such as "run") and a same-origin counter that stores the event name and UI language; no file contents or names are sent.

docxtemplater already copes with Word splitting a placeholder like {Name} across several XML runs, which is what I feared most. The work went elsewhere:

- Excel date cells: the "locale default" date format is stored as m/d/yy, so dates have to be re-rendered in the user's locale or a Polish user gets 10/20/26.

- CSVs saved by Ukrainian/Polish Excel are often CP1251/CP1250, not UTF-8.

Free up to 20 documents per run, no sign-up. It outputs .docx only (no PDF yet). I'd like to hear what's missing for your use case, and whether the 20-document limit or the template syntax gets in the way.