Somewhere there is a forum thread with your book in it, posted by an account that bought exactly one copy. That is the shape of ebook piracy at publisher scale: not a heist, a purchase, followed by a strip and an upload, after which the file reproduces for free forever.
Authors argue about DRM as if it were a lock quality question. It is not. Every mainstream ebook DRM has public removal tools, so the lock delays a determined pirate by minutes. The real question is what the leaked file can tell you, and what the buying experience costs the people who paid.
A visible and hidden watermark pair does two jobs. The visible mark, buyer name or order reference on the title page or in the footer, raises the social cost of sharing: posting the file means posting your name with it. The hidden marks, small variations and identifiers embedded through the file, survive casual retyping and format conversion well enough to identify the source copy when a leak surfaces.
Deterrence by attribution works because the first leaker is usually not a professional. They are a reader in a community, with a reputation and an account at your store, and the knowledge that a shared copy names them stops most of the casual pipeline before it starts. Publishers using social DRM consistently treat the absence of leaks from watermarked stores, next to leaks from unwatermarked ones, as the practical evidence.
Platform DRM, the kind tied to a retailer ecosystem, buys friction inside that ecosystem and nothing outside it. It inconveniences legitimate readers who change devices, read across platforms, or lose account access, and those readers are disproportionately your most committed buyers, the ones with libraries.
There are exceptions. Textbooks with institutional licensing, high value professional references, and titles under contractual DRM requirements from a distributor may not have the choice. The argument here is for the default: when the decision is yours, spend the effort on watermarking, store level enforcement, and fast takedown, rather than on a lock whose removal instructions rank on the first page of search results for your own title.
Confirm the source before accusing anyone. Extract what the watermark tells you, verify the corresponding order, and consider innocent routes first: a shared household account, a stolen copy, a review copy that travelled. A false public accusation against a reader is a reputation event you cannot retract.
Then work the distribution points, not the file. Takedown notices to the forum, the file host, and search engines, citing the specific URLs, shrink a leak's audience to people who already know where to look. New leaks of the same title get faster with practice and templates. It is maintenance, not victory, and publishers who budget it as maintenance keep their sanity and their backlist revenue.
Not in practice for a determined leaker: major DRM schemes have known removal tools. DRM can slow casual copying inside one retail ecosystem, but it also restricts legitimate readers, which is why many publishers choose watermarking instead.
Watermarking each sold copy with the buyer's identity, visibly and through embedded identifiers. The book remains fully readable and portable, and a leaked copy can be traced to the purchase it came from, which deters sharing at the source.
Visible marks can be cropped or retyped away with effort, which is why hidden identifiers distributed through the file matter. The goal is not an unremovable mark; it is raising the effort and preserving attribution for copies that circulate casually.
Identify the source copy from its watermark, verify the order behind it, and file takedowns against the hosting and linking URLs. Avoid public accusations until the source is confirmed, because copies travel through hands other than the buyer's.