A content warning tells you what is in a story before you read it: violence, grief, coercion, specific kinds of sex. In erotica it does a second job as well, because for many readers a warning is also a wish list, and the same tag that makes one person close a book makes another open it.
Who writes them
Usually the author, sometimes the platform, often nobody. Archives built around tagging put the burden on the writer and give them a vocabulary to do it with. Commercial apps tend to prefer a single age rating and a genre label, which tells you almost nothing about what a particular story contains.
That difference is worth knowing before you choose where to read. If you need to avoid a specific theme, a platform with real tags is a better tool than a polished app with none.
Why they disagree
There is no standard. One platform's warning list runs to dozens of entries; another has four categories. The same story can carry a detailed warning on one site and none on the store that sells it, because a store listing is marketing and a warning is not.
Using them well
Read the tags, not just the heading. A story tagged for a theme may use it as background or as the whole plot, and the tag will not say which. Where a platform lets you filter tags out, set that up once rather than checking every title. And treat the absence of a warning as no information: it usually means nobody wrote one, not that there is nothing to warn about.
Warnings and ratings are not the same
A store age rating describes an app, a content warning describes a story. An app rated 18+ can contain a story with nothing in it, and a story with a long warning list can sit inside an app rated for teenagers, because the two systems are filled in by different people for different reasons. If you are choosing what to read tonight, the warning is the useful one.
See also the glossary.