Every landing page and every blog post on this site carries a block of
FAQPage structured data. Five months ago that would have been an unremarkable
SEO checklist item. Today the visible payoff it was designed for no longer exists, so
it is fair to ask why it is still there.
Here is the honest version, including the part where our original reasoning was wrong.
What FAQPage used to do
FAQPage markup tells a search engine "this page contains these questions with these answers". For years Google rewarded it with a rich result: expandable question-and-answer dropdowns directly under your listing, which pushed competitors further down the page. That made it one of the most copied tactics in SEO, and predictably it got abused — pages stuffed with marketing questions nobody asked.
Google narrowed it twice:
- August 2023 — FAQ rich results limited to well-known, authoritative government and health websites. Everyone else stopped getting the dropdowns.
- May 7, 2026 — the feature stopped appearing in Google Search entirely, for every site. Google's FAQ structured data documentation now carries a notice saying exactly that.
So for a small site like ours, the rich result had already been unavailable for nearly three years before Google switched it off for everyone.
Why we added it anyway — and why that reason did not hold up
We added FAQPage to the homepage on May 17, 2026. The commit message gives the rationale in four words: "AI Overviews eligibility". The theory, widely repeated at the time, was that clean question-and-answer markup makes a page more likely to be quoted in AI-generated answers.
Google's own documentation for AI features says otherwise:
"There's also no special schema.org structured data that you need to add."
It states there are no additional requirements and no special optimisations for appearing in AI Overviews or AI Mode. So the specific reason we wrote down was, at best, unproven. We are not going to claim the markup earns AI citations, because nothing we can see says it does.
What we have measured: nothing. We have not run a before-and-after on FAQPage, and we could not isolate its effect if we tried — it shipped alongside new page copy, new landing pages and a dozen other changes. Any post claiming FAQPage moved its traffic by a precise percentage deserves the same scepticism.
The reason we are keeping it
The markup turned out to be useful for something other than search features: it forced the pages to actually answer questions.
Google's AI-features guidance has one structured data recommendation, and it is not about which types to use: make sure your structured data matches the visible text on the page. That rule is easy to state and surprisingly easy to break. We broke it ourselves.
In August we audited our blog and found that five of the eight posts carried four to six FAQ questions in their JSON-LD that appeared nowhere on the page — only search engines could see them. That is precisely the kind of markup Google spent three years phasing out. We fixed it by rendering every FAQ block as a visible accordion at the end of the post, using the same question and answer text as the structured data. The "Answers, briefly" section at the bottom of this page is one of them.
Once the markup has to match something a reader can see, it stops being a search trick and becomes an editing constraint:
- Every answer has to stand on its own. A FAQ answer cannot lean on the paragraph above it, so it gets written as one complete, self-contained statement. That is also the shape any answer engine — Google's or anyone else's — is best at quoting.
- Copy changes have to happen twice. When we removed a "no tracking" claim from our pages in August, because Google Analytics runs sitewide, the same sentence existed in the visible FAQ and in the JSON-LD. Both copies had to change together. Structured data that silently disagrees with the page is worse than none.
- It costs almost nothing to keep. FAQPage is still valid schema.org vocabulary. A few hundred bytes of JSON per page, no render cost, no maintenance beyond keeping it in sync with text we were going to write anyway.
An audit problem we did not expect
One practical lesson for anyone checking their own markup: many fetch tools strip
<script> tags before you see the page, and JSON-LD lives in a
<script type="application/ld+json"> tag. Our weekly automated SEO review
fetched pages that way and repeatedly reported that pages had no structured data at all,
when every one of them had three blocks.
We fixed it by parsing the JSON-LD on the server and reporting, per page, which schema types are present — so nobody has to trust a tool that cannot see the thing it is auditing. If your checker says your schema vanished, look at the raw HTML before you believe it.
What we would tell you to do
- Do not add FAQPage expecting a search feature. There is no rich result to win any more, and Google says AI features need no special markup.
- If you already have it and the questions are visible on the page, keep it. It is valid, cheap, and matches what Google asks of structured data.
- If your FAQ markup describes questions that are not on the page, fix that first — either show them or delete the markup. Hidden-only FAQ schema was the pattern Google retired the feature over.
- Measure the page, not the markup. Google reports traffic from AI Overviews and AI Mode inside the normal Search Console Performance report, under the Web search type. Watch whether the page is earning impressions; do not expect to attribute them to one JSON-LD block.
That is the whole decision. We kept FAQPage not because it does anything clever, but because writing answers that work as stand-alone answers is a good discipline, and the markup keeps us honest about it.
If you came here for the product rather than the SEO: ShareNotes is a free, no-account way to share text and code on a memorable link, with self-destructing notes and a REST API that needs no key.