What a Proofing Score Actually Measures (and Why "Looks Fine" Isn't Enough)

A 200-page EOC can look completely fine on a visual skim. The formatting is consistent, the sections are in order, nothing jumps out as wrong. That is exactly the problem: several categories of error are specifically ones a skim will not catch, no matter how careful the reviewer is.

What “looks fine” doesn't catch

• Cross-reference and table of contents errors, which read correctly in isolation and only reveal themselves if someone checks the actual destination
• Broken or dead URLs, which look identical to working ones on the page
• Glyph and font defects, shape-level rendering errors that print production can introduce and the eye can miss
• Inconsistent terminology for the same concept used in different places in the same document
• Stray foreign-language text left behind by source production tools
• Numbering and heading sequences that break somewhere in a long document

What a proofing score is

A proofing score is one weighted number per document that reflects findings across all of these categories at once, cross-references, URLs, spelling and grammar, glyph defects, term consistency, non-English contamination, and numbering consistency, with full drill-down into every individual finding behind the score. It is not a substitute for reading the document. It is a way to know, before final submission, whether the document has the kinds of errors a visual skim structurally cannot surface.

A related but different question

A proofing score measures document quality and internal consistency. It is a different question from PDF/UA-1 accessibility conformance, which measures whether assistive technology can correctly read and navigate the document. A document can proof cleanly and still fail accessibility conformance, or vice versa. Both checks matter, and they are not substitutes for each other.

This is the specific gap ProofTag is built to close. It is currently in development.