Do I need FAQ schema to show up in AI Overviews

No. You do not need FAQ schema to show up in Google AI Overviews. Google’s documented floor is ordinary Search eligibility: the URL is indexed and allowed to show a snippet. There is no Overview-specific FAQPage type, and stuffing JSON-LD into the head is not a citation ticket. Write the questions on the page in HTML. That is the extractable object. The schema is optional labeling, not a pass.

This spoke sits under the Answer Engine Optimization playbook. The implementation layer for honest JSON-LD lives in schema markup for answer engines. If you already rank and still never appear as a supporting link, diagnose that with ranked but missing AI Overviews. Lane work is on /visibility. I have been SEO certified since 2021. That credential does not make FAQ JSON-LD an Overview ranking factor.

The short answer

• FAQ content (visible Q&A a human can read) still helps a model lift a bounded answer. FAQ schema (FAQPage JSON-LD) does not buy inclusion.
• Google deprecated FAQ rich results for most sites in August 2023, then stopped showing the feature in Search as of May 7, 2026. That is a SERP widget death. It is not a new Overview rule.
• Google’s generative-AI optimization guide is explicit: structured data is not required for generative AI search, and there is no special schema.org markup you need to add.
• Keep FAQPage only where the questions render. Delete sitewide injections. Unused markup is not a ranking penalty by Google’s 2023 wording — it is also not a ranking gift.
• Score the work on Overview citations and brand facts that stop being invented, not on a FAQ enhancement report that no longer exists.

Object	Still a job	Not a job
Visible FAQ HTML	Extractable Q&A on a money URL	Padding to chase a retired widget
FAQPage JSON-LD	Label the same questions the reader sees	Ticket into AI Overviews
FAQ rich result	Gone from Search as of May 7, 2026	Your KPI
Indexed + snippet-eligible URL	Google’s documented Overview floor	Optional if you “have schema”

What did Google actually deprecate in 2023 versus May 2026?

Two different events get mashed into one LinkedIn slide. Separate them or you will “fix” a widget that most sites already lost.

On August 8, 2023, John Mueller announced that FAQ rich results from FAQPage structured data would only be shown for well-known, authoritative government and health websites. For everyone else, the expandable SERP dropdown would no longer appear regularly. Google also said you did not need to rip the markup out: structured data that is not being used does not cause problems for Search, and also has no visible effects in Google Search. That is the “most sites” deprecation people still remember.

On May 8, 2026, Google’s Search documentation changelog recorded a deprecation notice on the FAQ rich-result docs. The reason line is the one to quote: the feature would no longer appear in Google Search starting May 7, 2026. On June 15, 2026 Google removed that documentation because the feature was no longer shown.

Search Engine Journal quoted Google’s deprecation banner as saying Search Console’s FAQ search appearance, rich-result report, and Rich Results Test support would drop in June 2026, with Search Console API support removed in August 2026. I am citing SEJ for that banner text because Google later deleted the FAQ docs page. Treat the June and August tooling dates as recap of Google’s own banner, not as a second primary source.

Date	What Google changed	Who still had a FAQ widget	What it did not change
Aug 8, 2023	FAQ rich results limited to well-known gov and health sites	Most commercial sites lost the dropdown	FAQPage as a schema.org type stayed valid
Aug 2023 (same post)	Unused FAQ markup does not need emergency deletion	Anyone who kept JSON-LD	It did not become an Overview ranking factor
May 7, 2026	FAQ rich results stop appearing in Search	Nobody should plan on the widget	Visible FAQ copy on your page
May 8, 2026	Changelog documents the deprecation	Same	AI Overviews eligibility rules
Jun 15, 2026	FAQ rich-result documentation removed	Same	schema.org FAQPage
Jun–Aug 2026 (banner via SEJ)	Reporting, Rich Results Test, then API support for the FAQ appearance	Engineering calendars	Citation in Overviews

Hedge: Google has not published a study that says “we turned FAQ rich results off, therefore Overviews now require FAQ schema” or the inverse. The two systems are not documented as a swap.

Operator rule: if your 2026 ticket still says “restore FAQ rich results,” close it. The feature is gone. Open a ticket about extractable answers instead.

Checklist for the client or exec who still has a FAQ appearance screenshot from 2022:

• Date the screenshot. If it is pre-August 2023, it is a different product era.
• If it is between August 2023 and May 2026 and you are not a well-known gov or health publisher, treat the screenshot as a holdback, a misread, or a different search appearance — not as a promise the widget will return.
• If Search Console still lists a FAQ enhancement in an old export, do not rebuild the content calendar around it. Google’s May 8, 2026 changelog said the Search appearance was ending.
• Separate “we lost FAQ pixels” from “we are missing Overviews.” Those are two scoreboards.
• Put the 2023 Mueller post and the 2026 changelog URLs on the same slide as the screenshot.

Slide line that causes damage	Replace with
“Google removed FAQ schema”	“Google removed the FAQ rich result. FAQPage is still a schema.org type.”
“We need FAQ schema back to enter AI Overviews”	“Overviews require indexed, snippet-eligible pages. FAQ JSON-LD is not on that list.”
“Gov and health sites still get FAQ rich results”	True for a window after August 2023. Not a plan after May 7, 2026.
“Unused FAQ markup is a ranking risk”	Google’s 2023 post said unused structured data does not cause Search problems. Dirty, mismatched markup is a different issue.

Does Google require FAQ schema for AI Overviews?

No. The official pages do not list FAQPage as an Overview requirement.

Google’s AI features and your website page says a supporting link in AI Overviews or AI Mode must be indexed and eligible to show in Google Search with a snippet. It then says there are no additional technical requirements. The same page lists structured data that matches visible text as ordinary SEO hygiene, next to crawl access and internal links — not as a special Overview type.

The generative-AI optimization guide (added to Search Central on May 15, 2026 per the same changelog) has a mythbusting section. One line to keep on the slide: structured data is not required for generative AI search, and there is no special schema.org markup you need to add. Google still says it is a good idea to use structured data as part of overall SEO, because it helps with rich-result eligibility on Search. FAQ is no longer that rich result.

Claim you will hear	Google’s documented position	What you do
“Add FAQ schema or you cannot enter Overviews”	No extra technical requirements beyond indexed + snippet-eligible	Do not buy a schema plugin to “unlock AIO”
“FAQPage is the AI Overviews markup”	No special schema.org type for generative AI search	Write the answer in HTML
“Google killed FAQ schema”	Google killed the rich result. FAQPage remains a schema.org type	Stop treating a missing enhancement as an outage
“Keep schema for rich results, therefore keep FAQ schema for Overviews”	Remaining rich-result types still exist. FAQ is not one of them	Spend the schema budget on Organization, Article, Product, LocalBusiness where they still map to live features
“If we have FAQ JSON-LD we are snippet-eligible”	Snippet eligibility is robots / preview controls, not FAQPage	Check nosnippet and max-snippet:0 first

Checklist before anyone opens a “schema for AIO” sprint:

• Read the AI-features technical-requirements paragraph out loud in the standup
• Confirm the money URL is indexed (URL Inspection, not a vendor crawl of staging)
• Confirm the answer block is not wrapped in nosnippet or data-nosnippet
• Confirm the answer exists in fetched HTML, not only in a client-rendered drawer
• Only then decide whether this URL even has a visible FAQ worth labeling

If those boxes fail, FAQ schema is rearranging deck chairs. Eligibility is a floor. Schema is not the floor.

Why do teams still treat FAQPage as an Overview ticket?

Because the old play worked on a different scoreboard. From roughly 2019 through mid-2023, FAQ rich results were extra SERP pixels. Extra pixels meant extra CTR on some queries. Agencies sold “FAQ schema packages.” Plugins shipped sitewide FAQPage by default. When Overviews showed up, the same package got relabeled “AEO” without a new mechanism.

The confusion is understandable. Both objects involve questions. Only one of them was a Google Search appearance. Overviews are a generative unit that cites supporting links. They are not an accordion under your blue link.

Old FAQ-schema play	What it optimized	What Overviews actually select	Why the play fails now
Ten generic Qs in JSON-LD	Pixel height on the SERP	A passage that answers a fan-out sub-question	The pixels are gone
Same FAQ blob on every URL	Implementation convenience	The URL whose copy matches that sub-question	Sitewide mush looks like spam to a reader and a rater
Keyword-stuffed question strings	Query coverage theater	Query fan-out across real pages	Google’s guide warns against doorway variants for fan-out
Hidden answers in JSON-LD only	“Machines will find it”	Visible text, per Google’s own structured-data matching rule	Structured-data policies still forbid marking up what the user cannot see
Rich-result impression report	Proof the package “worked”	Citation in the Overview, or not	The report is being retired with the feature

John Mueller’s May 2025 Search Central post on succeeding in AI search still lists structured data that matches what people see. Matching visible text is the bar. Inventing a second copy in JSON-LD is the opposite of that bar.

I will not invent a percentage of Overview citations that “came from FAQ schema.” Google has not published that split. If a vendor shows you one, ask for the method, the date, and whether they confused FAQ content with FAQ markup.

Vendor artifact	What to ask	Acceptable answer	Walk-away answer
“FAQ schema package for AEO”	Which Google URL lists FAQPage as Overview markup?	They cannot find one; they reframe to extractable HTML	“Trust us, models prefer JSON-LD”
Yoast / Rank Math “FAQ block = AI Overviews”	Does the block render HTML the user can read?	Yes, and JSON-LD is generated from that HTML	JSON-LD only, or sitewide identical FAQs
Case study with a CTR spike in 2022	Was that FAQ rich result CTR?	They date it and admit the widget is gone	They relabel 2022 CTR as “AIO citations”
“Schema completeness score”	Completeness of which types, against which Google appearance?	Organization / Product / Article against live docs	A single score that treats FAQPage like Product

What is the difference between FAQ content and FAQ schema?

Two artifacts. One page. Mixing them is how a six-week schema project ships and nobody can quote your price.

FAQ content is HTML a person can read: an H2 or H3 that is a question, a short answer underneath, ideally 40–80 words that stand alone if a model lifts them. That is the same extractability job as the lead paragraph. It happens to be shaped like Q&A.

FAQ schema is a machine graph, usually JSON-LD, that says this page is an FAQPage and lists Question / acceptedAnswer pairs. schema.org/FAQPage is the type. It is valid whether or not Google shows a rich result. Validity is not eligibility for Overviews.

Job	FAQ content (HTML)	FAQ schema (JSON-LD)
Human reads it	Yes	No, unless they view source
Model can lift it without executing JSON-LD	Yes, if it is in the fetched HTML	Only if the crawler chooses to use the graph
Google rich result (2026)	Irrelevant	Feature gone as of May 7, 2026
Overview supporting link	Possible if the URL is eligible and the passage fits	Not documented as a requirement or a bonus
Wrong when mismatched	Stale answers confuse buyers	Mismatch with HTML is a policy problem
Worth a week	On the five money URLs	Only after HTML is true

Procedure to keep the two from drifting:

1. Put the questions in the component that renders the page. One array. One source of truth.
2. Generate JSON-LD from that same array, or do not generate JSON-LD at all.
3. If an editor can change the HTML without changing the graph, you do not have a source of truth. You have a future incident.
4. If the FAQ UI is removed, the graph must disappear in the same deploy.
5. Never paste answers into acceptedAnswer that are longer, cheaper, or more specific than the visible paragraph.

Hedge: some retrieval systems may use labeled Q&A in the graph when they already fetched the URL. Google has not said Overviews do that with FAQPage. Do not brief leadership as if they did.

A page with six tight, true answers and zero FAQ JSON-LD is doing the Overview job. A page with forty FAQPage nodes and a vague lead is doing the 2022 CTR job on a feature that no longer exists.

Worked match versus mismatch — same URL, two graphs:

Field	Honest pair (keep)	Mismatch (delete or rebuild)
Visible H3	“Do you work with Michigan manufacturers?”	“Do you work with Michigan manufacturers?”
Visible answer	“Yes. The studio is based in Northern Michigan and takes manufacturer retainers remotely. On-site kickoffs are scheduled.”	Same HTML
JSON-LD name	Exact H3 string	“best AI SEO agency near me”
JSON-LD acceptedAnswer	Same meaning, same bounds	“We serve clients worldwide with packages starting at $99” — a price the page never shows
Question count	6 in HTML, 6 in JSON-LD	6 in HTML, 18 in JSON-LD from a leftover plugin
Who can see it	Anyone who opens the FAQ	Only a crawler that reads the graph

If the right-hand column is your production HTML, you do not have an Overview strategy. You have a policy problem and a future hallucination with your brand on it. Rebuild from the left-hand column or ship HTML only.

Which schema types still earn their keep after the FAQ widget died?

Schema budget is not “add FAQPage because AEO.” Schema budget is “describe the page you actually shipped, for features that still exist, and for entity clarity.” The companion spoke is the full type list. Here is the Overview-myth cut.

Google’s remaining rich-result types still have documentation. Organization, Article / BlogPosting, Product / Merchant listings, LocalBusiness, and similar types map to live Search appearances or Merchant Center / Business Profile surfaces. FAQPage no longer maps to a Google Search appearance. That does not make Organization optional. It makes FAQ-as-ticket optional.

Type	Keep when	Drop or skip when	Overview relationship
Organization / LocalBusiness	You have a legal name, NAP, sameAs you can defend	The node disagrees with About	Entity facts Overviews invent if you leave them blank
Article / BlogPosting	The URL is a post	You slap Article on a pricing page	Dates and authorship; not an Overview key
Product / Offer	A public product with a price you will honor	Fake reviews, hidden prices	Shopping-ish answers; still not FAQ
FAQPage	Visible Q&A on this URL, generated from the same source	Sitewide GTM, leftover after the FAQ UI died	Labeling only. Not a ticket
QAPage	A thread where users submit answers, matching Google’s QAPage docs	You marked a publisher FAQ as QAPage to “get a rich result”	Different type. Still not Overview markup
HowTo	Almost never for Google Search appearance	You kept it hoping for a How-To carousel	HowTo rich results were already gutted in 2023; do not confuse with FAQ

QAPage is the usual mix-up. Google still publishes that documentation for pages where users submit answers — forums, product Q&A — not for a brand writing its own FAQ. Marking your pricing FAQ as QAPage to chase a different widget is the same category error as treating FAQPage as Overview markup.

Checklist for a schema hour that is not a myth hour:

• Organization @id is stable and matches the About page
• sameAs URLs return 200 and are actually you
• Article dates match the visible byline, including dateModified after a real refresh
• Product price in JSON-LD matches the cart
• FAQPage question count equals visible FAQ count, or FAQPage is absent
• No second FAQ graph from a plugin plus your theme

If you only have an hour, fix Organization and the money-page offer sentence. Do not “complete” FAQ schema on 200 URLs.

What fails first when a team “adds FAQ schema for AEO”?

The failure is almost never “JSON-LD failed to validate.” The failure is a mismatch, a sitewide blob, or a week spent on markup while the answer still is not on the page.

Concrete pattern I have seen on production sites (hundreds shipped; I will not invent an Overview recovery rate for this one): a marketing site injects a global FAQ graph from a tag manager. The visible FAQ lives on pricing. Blog posts, the homepage, and a JS-rendered service template all emit the same ten questions. A retrieval pass on a blog URL can see Q&A the reader never saw. Google’s structured-data quality rules call that out. A structured-data manual action, when it happens, removes rich-result eligibility. It does not “rank-ban” the host. It also does not make an Overview trust you.

Failure	What you observe	What it costs	What you do instead
Sitewide GTM FAQPage	Every URL’s view-source has the same mainEntity	Conflicting answers vs the actual page; wasted engineering	Emit FAQPage only from the template that renders the FAQ
JSON-LD only, HTML empty	Rich Results Test used to flicker; now it ignores FAQ	You thought machines had the answer. Users did not	Render the questions. Then maybe label them
Stale answers in the graph	Editor updated HTML; schema still says the old price	Overview or chatbot quotes the old price with your brand attached	One array, or delete the graph
Forty questions of fluff	“What is SEO?” on an HVAC pricing page	Dilutes the three answers a buyer actually asks	Six hard questions. Kill the glossary
Measuring FAQ impressions	Search Console FAQ appearance goes to zero in 2026	Panic that “AEO died”	Log Overview citations. That was never this report
Schema sprint, no extractability	Valid FAQPage, 400-word wind-up, no table	You still lose to a uglier page with a procedure	Fix the lead. Then talk markup

Procedure when this ticket lands:

1. View source on the URL that is supposed to win the Overview, and on a URL that should not have a FAQ.
2. Count visible questions. Count Question nodes. If they differ, stop the AEO narrative.
3. Search the codebase and the tag manager for FAQPage. Kill global injection first.
4. Quote-test the lead and the first FAQ answer: can a stranger use those 60 words without the rest of the page?
5. Request indexing only after fetched HTML contains the answers.
6. Do not file a Sev-1 because the Rich Results Test no longer celebrates FAQPage. Google told you that would happen.

The cost is not a Google “schema penalty” in most cases. The cost is a quarter of engineering on a retired widget while competitors ship answer-first HTML.

How do you write FAQ copy an Overview can lift?

Write it like the rest of an extractable page. The H3 is the query. The first sentence is the answer. The next two sentences are the constraint, the date, or the next step. Stop.

Google’s AI-features doc says Overviews and AI Mode may use query fan-out — related searches across subtopics. Your FAQ is a cheap way to cover the objections that fan-out already shows in the unit: price bounds, who it is for, what you do not do, how long it takes. That is content architecture. It is not a JSON-LD trick.

Weak FAQ answer	Extractable FAQ answer
“It depends, reach out for a custom quote.”	“Visibility audits start as a scoped engagement. We do not publish a fake starter price in schema. Book a visibility audit and we will say the number on the call.”
“We offer a wide range of premium digital solutions.”	Vague brand line plus no number. Rewrite or delete.
400-word essay inside acceptedAnswer	40–80 words that match the HTML paragraph
Question stuffed with every synonym	The question a buyer would actually type
FAQ that repeats the H1	A sub-question the Overview already lists and your lead ignored

Numbered pass on one money URL:

1. Open the Overview (or ten frozen prompts) for the query you care about. List the bullets and the follow-up questions.
2. For each bullet you should own, write one H3 that ends in ? and one answer that can be lifted.
3. Put facts that must not be invented — price bounds, cities, certifications, hours — in a table above or beside the FAQ, then point the FAQ at the table. I will cite receipts I can defend: SEO certified since 2021, 600+ automations built and 500+ still live, 20,000+ hours on agentic systems, 35,000+ hours saved for clients, hundreds of production sites. I will not invent your close rate.
4. Keep the visible FAQ to the questions this URL should answer. Move cluster questions to their own URLs instead of stuffing one graph.
5. If you emit JSON-LD, copy the question string and the answer meaning exactly.
6. Re-fetch with URL Inspection. If the FAQ is behind a tab, confirm Googlebot still gets the text. Expandable sections are allowed when the user can open them; they are not a license to hide the only true price.

Checklist for each FAQ item:

• H3 ends with a real question a buyer types
• First sentence answers it without a wind-up
• Answer is 2–4 sentences, not a blog post
• No claim you cannot source
• Same words in HTML and JSON-LD, or no JSON-LD
• Next step is a protectable URL, not “read more”

Do not write FAQ items for queries this URL cannot win. That is how you recreate doorway pages inside a <dl>.

When should you keep FAQPage JSON-LD, and when should you delete it?

Keep it when it is honest, generated from the visible FAQ, and cheap to maintain. Delete it when it was installed to chase a widget, when it disagrees with HTML, or when nobody owns the sync.

Google’s 2023 line still matters for the keep case: unused structured data does not by itself cause Search problems. That is permission to leave a clean graph in place. It is not a reason to keep a dirty graph. The 2026 deprecation removes the last Google-Search reason to add FAQPage you do not already have.

Situation	Keep FAQPage	Delete FAQPage	Do this week
Pricing page with six real Qs, one template array	Yes	No	Quote-test the answers
Blog post, no FAQ UI, plugin emits FAQPage	No	Yes	Kill the plugin default
FAQ UI removed in the last redesign	No	Yes	Confirm the graph died
Two FAQ graphs with different answers	No	Delete the one that does not render	Then keep or drop the survivor
Gov/health site that still hoped for the 2023 exception	The widget is gone for them too as of May 2026	Optional either way	Stop planning on the dropdown
You have never shipped FAQ HTML	Do not add schema first	N/A	Write the HTML

Decision list:

1. Does this URL render questions a user can open without View Source? If no, no FAQPage.
2. Can you generate the graph from the same source as the HTML? If no, no FAQPage until you can.
3. Are you adding it because a 2024 playbook said “required for AEO”? If yes, do not add it.
4. Are you deleting it because the Rich Results Test stopped showing a FAQ enhancement? If that is the only reason and the graph is honest, you can leave it. You can also delete it. Google has not documented a ranking difference. Pick the option your engineers can maintain.

Hedge: I cannot quote a controlled test that says keeping honest FAQPage after May 2026 moved Overview citations. I also cannot quote one that says deleting honest FAQPage moved them. SearchPilot and similar vendors have published tests on removing FAQ schema in the 2023–2024 window for classic organic traffic; those tests are not Overview citation studies, and I will not recycle their numbers here as if they were. Your frozen query log is the experiment.

How do you measure this without a FAQ rich-result report?

You do not replace a dead enhancement report with a vibe. You replace it with the Overview scoreboard you should have been using anyway.

Google counts AI Overviews and AI Mode inside Search Console’s Web performance data. In June 2026 Google launched a Search generative AI performance report. The help-center writeup described a rollout to a subset of properties, impression-led, with Search Labs excluded. If your property does not have it yet, do not invent a workaround metric. Use the Web report plus a manual citation log.

Metric	Source	Tells you	Does not tell you
Overview present on the query	Manual SERP log, same device class	Whether the unit fired	Whether you were cited
Your URL cited / mentioned / absent	Same log, screenshot	The actual KPI for this myth	A schema “score”
Generative AI impressions	Search Console, if rolled out	Reach inside AI features	That FAQPage caused them
Clicks from Overview SERPs	GSC click counting rules	Residual traffic	2022 informational CTR restored
FAQ rich-result impressions	Legacy appearance report	That a retired widget is gone	Anything about Overviews
Rank in the top 10	Classic GSC	Candidate status	Citation — see the missing-Overviews diagnostic

Procedure for a 30-query panel:

1. Freeze 20–40 queries that already trigger Overviews in your category. Write them down. Do not swap them weekly.
2. Once a week, log: Overview yes/no, cited domains, your URL cited or not, your classic position.
3. Annotate any FAQ HTML or schema deploy on the candidate URLs. Date it.
4. After 30 days, look for citation flips on the URLs you actually edited. If citations moved on URLs you only “schema’d,” check whether the HTML answers also changed. They usually did.
5. If nothing flipped, you did not fail “FAQ schema.” You failed extractability, eligibility, or fan-out coverage.

Checklist so the log is honest:

• Same queries for 90 days
• Device class noted (desktop vs mobile SERPs differ)
• Schema-only deploys tagged separately from content deploys
• No celebrating a FAQ enhancement disappearing as an Overview loss
• Brand queries as a control

If leadership wants a percentage, give them your citation rate on the frozen set, dated. Do not give them a vendor “FAQ schema lift for AIO.”

What should you skip if you only have a week?

Skip the sitewide FAQPage rollout. Skip the plugin that “adds 12 schema types.” Skip waiting for a FAQ impression report to come back. Skip rewriting every blog post’s JSON-LD.

Do the extractability work on the URLs that can actually be cited.

Day	Do	Done looks like	Skip
1	Inventory GTM, theme, and plugin FAQPage emitters	List of templates that emit the type, with question counts	A new schema vendor
2	Kill global injection that does not match HTML	Blog and homepage no longer claim the pricing FAQ	“We’ll clean it in Q4”
3–4	Write or rewrite six FAQ items on one money URL	Lead + FAQ pass a 60-word quote test	Forty questions of fluff
5	Align Organization and offer facts with About	Price, city, legal name match	New FAQPage on 50 URLs
6	URL Inspection on the money URL	Fetched HTML contains the answers	Index request on stale HTML
7	Start the 20-query citation log	Spreadsheet with today’s SERP screenshots	Declaring victory because JSON-LD validates

Decision list if the week is even shorter:

1. If nosnippet or noindex is on the money URL, fix that before any FAQ work.
2. If the money URL has no answer in the first 80 words, write that before schema.
3. If FAQPage is only on URLs with no FAQ UI, delete it. That is a half-day. It is not an Overview strategy, but it stops a lie.
4. If honest FAQ HTML already exists and the graph matches, leave the graph. Spend the remaining days on fan-out coverage, not on a second graph.

If you finish the table above before Friday, do not “use the extra days” to spray FAQPage across the blog. Use them like this:

• Screenshot the Overview on five money queries and paste the bullets under the matching H3s you still owe
• Confirm Organization sameAs and the About page still agree after the FAQ rewrite
• Add one citeable table on the money URL (scope, pricing bounds, or turnaround) so the FAQ is not the only extractable object
• Request indexing on the URLs whose fetched HTML actually changed
• Write the exec sentence: “We did not install Overview markup. We wrote answers Google can quote.”

A week of honest Q&A on one service URL beats a quarter of FAQ schema theater. If you cannot name the money URL, you do not have a schema problem. You have a page-job problem.

When is FAQ schema work not worth doing yet?

Most of the time, “FAQ schema work” is not worth doing yet because the page still cannot be quoted. Markup cannot rescue a URL that is blocked, snippet-killed, or written as a brand myth.

Do the eligibility and extractability jobs first. Google’s AI-features page puts those in front of structured data. The ranked but missing diagnostic is the longer version of this table.

Blocker	Why FAQ schema waits	First fix
noindex, robots block, or not in the index	Overviews cannot cite what Search cannot show	Indexing, then come back
nosnippet or max-snippet:0 on the answer	Google says supporting links need a snippet	Remove the preview clamp on the passage you want quoted
Answer lives only in client-rendered JS Googlebot does not get	Fetched HTML is empty of the FAQ	SSR or prerender the questions
Lead is 200 words of origin story	Fan-out will pick a page that answers in sentence one	Rewrite the lead; then write FAQ items
Offer facts disagree across Home, About, and schema	Models invent a third version	Align entity facts
You sell a commodity explainer you refuse to date	Overviews prefer a specific, checkable passage	Add a dated table, then Q&A
The only “FAQ” is ten synonyms of the H1	You are stuffing, not covering sub-questions	Map the Overview bullets first

Checklist — if any box is unchecked, do not open a FAQPage ticket:

• Money URL returns 200 to Googlebot
• URL Inspection shows the answer text in crawled HTML
• Preview controls do not zero out the answer
• First 80 words answer the primary query
• Price, city, and legal name match About
• You can name the three sub-questions the Overview already shows

When it is worth a small schema pass: the HTML FAQ is live, the array is the source of truth, and a plugin is emitting a second, contradictory graph. That is hygiene. Hygiene is not a growth strategy. Do not brief it as “we shipped AEO.”

Skip FAQ schema work entirely when the site has no visible Q&A, when the CMS cannot keep HTML and JSON-LD in sync, or when the quarter’s actual problem is that you are not indexed. Adding FAQPage to an unindexed URL is a costume.

How does this connect to ranking but missing Overviews?

Ranking in the top 10 and missing the Overview is a different incident from “we forgot FAQ schema.” Classic rank says the URL is relevant enough for a blue link. The Overview still has to be allowed to quote you, able to lift a faithful passage, and willing to use that passage for a fan-out sub-question. FAQ JSON-LD does not finish any of those three.

If you rank and you are absent from the unit, run the diagnostic spoke, not a schema plugin. If you do not rank and you are absent, FAQ schema will not promote you. If there is no Overview on the query, you are looking at a feature-trigger decision, which Google says happens only when the unit is additive to classic Search.

What you see	Likely cause	FAQ schema’s role	Next move
Rank 1–10, Overview cites others	Passage selection or fan-out	Near zero	Quote-test the lead and add the missing sub-question as visible FAQ content
Rank 1–10, no Overview on the query	Feature may not have fired	None	Confirm the unit exists for a week; do not “schema” a missing unit
Rank 1–10, you are ineligible	nosnippet, noindex, unindexed	None until eligibility is clean	Fix robots and preview controls
Not in the top 10, not cited	Relevance or crawl	None	Classic SEO first
Cited after you added FAQ HTML (not just JSON-LD)	Extractability	Labeling optional	Keep the HTML; keep or drop matching JSON-LD
Cited after you only added JSON-LD, HTML unchanged	Coincidence, recrawl, or you are misreading the log	Do not claim causation	Check whether the Overview question even matches a Question node

Procedure when someone files “add FAQ schema, we rank but we are missing Overviews”:

1. Screenshot the SERP. Confirm an Overview is present.
2. Confirm your URL’s index and snippet state with URL Inspection.
3. Quote-test the current lead. If it cannot stand alone, rewrite it. Do not touch JSON-LD yet.
4. List the Overview’s bullets. Write visible FAQ items that answer the ones you should own.
5. Only if those items already existed in HTML and a graph would stay in sync, emit FAQPage.
6. Re-test the same query on day 0 / 7 / 14 / 30. Log citation separately from position.

Failure mode: a sprint that “ships FAQ schema for AEO” on the ranking URL while the Overview is quoting a competitor’s table you never wrote. You will still be missing. The missing object was the table.

What does good look like 90 days later?

You will not get a FAQ dropdown back. You should be able to show that the myth is closed and that citations, if they moved, moved with extractable HTML.

90-day artifact	Pass	Fail
FAQPage inventory	Every emitter mapped; no sitewide GTM blob	“The plugin handles it”
HTML vs graph	Question counts match, or FAQPage is absent	Two answers for the same question
Money URL quote test	First 80 words plus one FAQ item lift cleanly	Origin story, then a JSON-LD dump
Frozen query log	12 weeks, Overview + citation marked	Ad-hoc screenshots after a schema deploy
Rich-result report	Ignored, or noted as retired	Used as the AEO KPI
Exec one-pager	Quotes Search Central: no special Overview markup	“We added FAQ schema so we will be in Overviews”

Weekly ops after the first month:

Cadence	Owner	Artifact
Weekly	SEO lead	10-query Overview citation log
After each template deploy	Engineering	HTML question count equals JSON-LD question count, or graph omitted
Monthly	Content	Two money-URL FAQ items rewritten from live Overviews, not from a keyword tool
Quarterly	Marketing lead	Schema types vs live Google appearances — FAQ is not on the appearance list

If citations did not move and you only changed JSON-LD, you ran the wrong experiment. Run the HTML experiment next. If citations moved after you wrote real answers, keep writing answers. Optionally keep the matching graph. Do not reverse the order next quarter.

I have been SEO certified since 2021. The certification is useful for reading Search Central without a vendor translation. It is not a reason to reinstall a 2022 FAQ widget as 2026 AEO.

FAQ

Do I need FAQ schema to show up in AI Overviews?

No. Google’s AI-features documentation requires an indexed, snippet-eligible page and lists no extra Overview markup. The generative-AI optimization guide says structured data is not required for generative AI search and that there is no special schema.org type to add. Visible FAQ copy can still help a model lift a bounded answer. FAQPage JSON-LD is not a ticket.

How do I measure whether FAQ content is helping AI Overviews?

Log a frozen query set weekly: Overview present or not, your URL cited or not, classic position. Annotate the date you shipped HTML answers versus the date you shipped JSON-LD. Use Search Console’s generative AI report if the property has it; otherwise the Web report plus screenshots. Do not use a FAQ rich-result impression report as the KPI. Google retired that appearance.

What usually fails first when teams try this?

Sitewide FAQPage from a tag manager or plugin, answers that exist only in JSON-LD, and a schema sprint while the lead still cannot be quoted. The graph disagrees with the HTML, or the HTML never had a real answer. Fix the visible Q&A and the eligibility floor before you emit another Question node.

How long does this take to show results?

Eligibility fixes can show up on the next recrawl, which Google says can take days to months depending on the URL. Extractable FAQ copy is a recrawl plus a selection problem, not a schema compile. Give a frozen panel 30 days before you declare a citation win or loss. Do not expect a FAQ rich result to return. That feature stopped appearing in Search as of May 7, 2026.

What should I skip if I only have a week?

Skip sitewide FAQPage, schema-pack plugins, and waiting for a retired enhancement report. Inventory emitters on day one, kill mismatched injection on day two, write six real FAQ items on one money URL, align Organization facts, inspect the fetched HTML, and start the citation log. Leave honest matching JSON-LD in place if it is already cheap to maintain.

When is this not worth doing yet?

When the money URL is not indexed, when preview controls strip the snippet, when the answer is missing from crawled HTML, or when Home / About / schema disagree on the offer. FAQ schema cannot buy an Overview citation in those states. Write the answer, fix eligibility, then decide whether a matching FAQPage graph is even worth generating.

CTA

FAQ schema is not your Overview ticket. Write the answers, keep the graph honest or omit it, and measure citations.

Lane: /visibility · Book a visibility audit.