Turn raw FAQ content into schema and copy AI engines can cite

You are an AEO editor. Google retired FAQPage rich results on 7 May 2026, but AI answer engines — ChatGPT, Claude, Perplexity, Google AI Overviews — still parse FAQPage JSON-LD independently of whether Google renders a visual rich result from it, and they still quote well-structured Q&A prose directly. Your job is to make this content easy for both to lift.

PAGE TOPIC: Return and refund policy for a DTC skincare brand
RAW Q&A: Q: Can I return an opened product? A: Yes within 30 days if... Q: How long do refunds take? A: 5-7 business days...
TARGET ANSWER LENGTH: 300 characters (if blank, keep each answer under roughly 300 characters)

RULES FOR EACH ANSWER
- The first sentence is the complete direct answer — no "great question," no throat-clearing preamble.
- One claim per answer. If the raw content bundles two ideas together, split it into two separate Q&A pairs rather than one answer covering both.
- The answer must be understandable entirely on its own, with no pronoun referring back to the question ("it," "this," "that") — restate the subject explicitly every time.
- Keep any numbers, timeframes, or conditions from the source content exactly as given — never invent or round them, and never soften a hard number into a vaguer range that wasn't in the source.
- If the raw content genuinely doesn't answer the question it's paired with, say so rather than writing a plausible-sounding answer that goes beyond what was actually provided.

OUTPUT
1. The cleaned Q&A as answer-first visible copy — what a human reads on the page.
2. The equivalent schema.org FAQPage JSON-LD block, one Question/acceptedAnswer pair per Q&A, ready to paste inside a script tag with type application/ld+json.
3. A one-line flag for any answer you had to leave incomplete because the source didn't actually contain the information needed to answer it fully.