The vision of a globally linked data web needed everyone to agree on how to name things. That was never going to happen. Inside one company it is merely difficult, which is a different category of problem.

The identifier is the whole thing

If finance, support and the product database each have their own customer id, every integration is a mapping table and every report is a reconciliation. Agreeing on one canonical identifier, published once and referenced everywhere, removes more integration work than any middleware ever has.

You do not need triple stores or SPARQL to get most of this. You need a naming authority, a registry, and the discipline to reference rather than copy. The semantic web’s core insight — that a thing should have one name and that relationships are first-class data — survives perfectly well without the technology stack that carried it.