Answer engines
Writing an FAQ an answer engine will actually quote
An answer is quotable when it survives being removed from the page. That means it names its own subject, answers in the first sentence, and contains no pronoun or phrase that depends on what came before it.
The test: does it survive removal
Take any answer on your site, paste it into an empty document, and read it as a stranger would. If you cannot tell what it is about, no model will quote it — it has no way to reattach the context you left on the page.
Answers fail this test in predictable ways: they open with “we”, “it” or “this”, they refer to “the above”, or they answer a question they never restate.
Restate the subject, then answer
“Yes, we do” needs the page to mean anything. “Yes, we install aluminium apertures in domestic properties across Malta and Gozo” does not. The second is longer and slightly repetitive to a human reader scrolling the page. It is also the one that can be lifted, and the small redundancy is the price of that.
Answer in the first sentence
Put the answer first and the qualification second. A model extracting a claim reads the opening sentence as the claim; if your first sentence is context and your third is the answer, the extraction gets the context.
Only publish answers you would stand behind verbatim
An FAQ answer is the sentence most likely to appear somewhere you do not control, stripped of every hedge around it. Write each one as though it will be read alone and attributed to you, because that is precisely the intended outcome.
Article FAQs
Still have questions?The short answers.
How long should an FAQ answer be?
Long enough to be complete and short enough to be lifted whole — in practice two to four sentences. Answers past roughly a hundred words tend to get truncated mid-thought, and a truncated answer is worse than a brief one because it misrepresents you.
Does FAQ structured data guarantee we get quoted?
No. Structured data makes an answer easier to find and parse; it does not make it worth quoting. Marking up a vague answer just helps a model discover that the answer is vague.
