Google Parses Raw Code Over Rendered Display. JSON-LD Schema Bloat Triggers Over-Optimization Penalties.

The algorithm reads raw text instead of the rendered layout, making hidden code elements a primary trigger for keyword over-optimization penalties, warns Charles Floate.

A URL might display a safe visible frequency of 40 to 55 keyword variations in the browser.

However, when practitioners inject the target keyword into every available JSON-LD field, the underlying document density can spike to 354 variations.

Because the crawler parses the complete raw source code rather than the visual viewport, a page with this schema bloat can fail algorithmic quality filters.

To audit this discrepancy, load the raw source code and run a direct search for the exact-match keyword to compare the true code-level density against the rendered count.