FAQ Schema Detection

What this parameter measures

Presence is the easy part; this parameter measures the quality of your FAQPage JSON-LD. friendly4AI parses the FAQPage block, counts the Question entries in mainEntity, measures the average length of questions and answers, and compares each question against your body content to detect duplication. It is a composite signal: presence is necessary but not sufficient, because a thin or copy-pasted FAQ helps AI systems far less than a well-formed one.

Why it matters for AI-readiness

FAQPage schema maps your answers directly onto the questions users ask, which is exactly the shape AI assistants need to quote you. A focused set of well-sized question-answer pairs gives ChatGPT, Gemini, and Perplexity clean, citable units to lift into their responses. Conversely, one-question stubs, bloated answers, or questions that just restate your body text give the model nothing distinct to extract, so the schema earns little credit.

How we score it

Under the v4.4 methodology, this Discovery parameter is scored by a five-part composite — the published Binary rubric is superseded by the v4.3 composite the processor actually runs. The composite weights presence at 50%, question count at 15%, question length at 10%, answer length at 10%, and low body-duplication at 15%. A valid FAQ with at least one question always reaches presence credit, so a presence-only FAQ scores 50 and lands in partial; a rich FAQ in the 4-8 question sweet spot with 8-20 word questions, 40-80 word answers, and little duplication clears 80 and passes. Anything that fails to parse, has no Question in mainEntity, or scores below 30 (one-question stubs, oversized answers, near-total duplication) fails at composite 0 or low. Counts at or above 16 questions are treated as spam and contribute nothing to the count sub-score.

How to fix common issues

• Add FAQPage JSON-LD with real Question/acceptedAnswer pairs in mainEntity; presence alone moves you off a hard fail.
• Aim for 4 to 8 questions — fewer underscores the count band, 16 or more is penalised as spam.
• Keep questions concise (roughly 8-20 words) and answers substantive but tight (roughly 40-80 words).
• Write answers that add information rather than echoing your page body; high overlap with body text triggers the duplication penalty.
• Validate the block at search.google.com/test/rich-results so a malformed FAQ does not score zero.
• Re-scan after publishing to confirm the composite has moved into the pass band.