How RDFa works
RDFa stands for Resource Description Framework in Attributes, a W3C standard for embedding machine-readable meaning in ordinary markup. Search engines support the simplified subset, RDFa Lite, which comes down to a handful of attributes: vocab declares which vocabulary you are borrowing words from, typeof says what kind of thing this element describes, and property labels each detail within it.
The declaration step is what separates RDFa from its neighbours. Microdata names its type with a full web address every time; RDFa names the vocabulary once and then uses short terms from it. That design comes from the linked-data world, where a document might legitimately borrow terms from several vocabularies at once and needs to say which word came from where.
You have almost certainly seen it without noticing. The Open Graph tags that control how a page looks when it is shared on social platforms use the RDFa property attribute in the page head, which is why they read property rather than name like other meta tags.
Why RDFa matters
For most site owners it matters as recognition, not as a choice. Some content management systems and older enterprise templates shipped RDFa in their themes, so it turns up in audits of sites that nobody has rebuilt in years. If you cannot recognise it, you will either miss working markup that is already doing its job, or you will add a second description of the same page on top of it.
Search engines read it. Valid RDFa that describes the page honestly is treated the same as valid markup in any other format, so there is no penalty for having inherited it and no ranking bonus for switching away.
Where RDFa goes wrong
The vocabulary declaration is the usual trap. Terms are resolved against whatever vocab is in scope, so an element that sits outside the declaration, or inside a nested block that redeclares a different vocabulary, silently produces properties that mean nothing. The page still renders perfectly; only a validator reveals the gap.
Beyond that, the failures are the ones every inline format shares. Attributes attached to visible elements break when a designer restructures the template, hidden content marked up as though it were visible breaks the guidelines, and layering a fresh JSON-LD block over surviving RDFa leaves two descriptions of one page that can disagree with each other.
What to do about it
Run the page through the Schema Markup Validator before you touch anything, so you know what the existing markup actually says rather than what the theme documentation claims. Check the rendered page rather than the source file, because a template that assembles content from partials may not produce what you expect.
If it validates and matches the page, leave it. If you are rebuilding the template, or the markup is broken beyond a quick fix, write the replacement in JSON-LD — it is the format search engines recommend, it lives in one block, and a developer can change it without going near the design. Either way, treat structured data as part of the technical SEO layer of the site and keep one description per entity, in one format, on one page.
