How do you rewrite an FAQ block to get cited in AI Overviews?
Work question by question, applying the six attributes the study tested, starting with the two that carried the biggest lift. Each edit is small; together they compound, since questions carrying five or six attributes were cited at 3.4x the rate of questions carrying one or none.
- Pull the page's real queries. Use Google Search Console to list the searches the page already appears for.
- Rewrite each question to match a query word for word. "How much does IVF cost in India", not "Understanding IVF pricing" (+38%).
- Rewrite each answer's first sentence. Answer directly in 30 words or fewer, with no preamble (+31%).
- Add a specific fact. A number, a named rule or a proper noun the answer can be cited for (+22%).
- Trim or extend to 40-90 words. The band most often lifted into AI answers (+19%).
- Read it aloud. Natural spoken phrasing matches assistant and voice queries (+14%).
- Add FAQPage schema. Mark up the visible questions and answers exactly (+11%).
- Set the count. Six to eight questions on a commercial page, 8-12 on an informational one.
Google's own guidance on AI features and your website says there is no special markup requirement for AI Overviews, which fits the study's finding that the visible FAQ carries most of the lift. ICG applies this checklist in its AEO work.
What is the difference between an FAQ section and FAQ schema?
An FAQ section is the visible block of questions and answers that readers see on the page. FAQ schema (FAQPage structured data) is code in the page that labels those questions and answers for machines. The study measured both, and the visible section did more of the work.
| Visible FAQ section (HTML) | FAQPage schema | |
|---|---|---|
| Who sees it | Readers and crawlers | Crawlers only |
| Measured lift on AIO citation | About 6-8 percentage points | About 3-5 points on top |
| Effort | Writing and editing per page | One-time template change |
| Works alone? | Yes | Should only mark up content that is visible on the page |
Schema should always match the visible text; marking up questions that readers cannot see risks being treated as misleading markup. For healthcare pages, answers must also stay accurate and free of promotional claims, since AI answers quote them out of context. See ICG's guide on getting cited in AI Overviews.
TL;DR — FAQ, schema and AIO citation
- Pages with no FAQ cite at 8.4 percent in the sample. Pages with 5-8 FAQs cite at 17.6 percent. Pages with 9+ FAQs cite at 21.3 percent.
- The FAQ pattern carries the lift; the schema markup adds to it. HTML FAQ without schema lifts by ~6-8 pp. Schema on top adds another ~3-5 pp.
- Six question attributes account for the highest-cited questions. Query-verbatim phrasing (+38 percent lift), direct-answer opening (+31 percent), proper-noun-or-number content (+22 percent), 40-90 word answer length (+19 percent), spoken phrasing (+14 percent), FAQPage schema (+11 percent).
- Direct-answer opening is the highest lift-per-minute edit. Rewrite the first sentence of every answer to answer the question in 30 words or fewer.
- Efficient operating point: 6-8 questions per commercial page, 8-12 per informational page. Above that, marginal returns diminish and readability suffers.
- The schema is not dead. Even without rich-result rendering, FAQPage markup functions as a discovery signal for AIO, PAA panels and third-party assistants.
Cite this report
Inline (HTML):
Ichelon Consulting Group (2026). FAQ Schema & AIO Citation Correlation Study · 2026. https://ichelonconsulting.com/reports/faq-schema-aio-citation-correlation-study-2026
APA 7:
Sihag, H. & Ichelon Consulting Group. (2026). FAQ Schema and AIO Citation Correlation Study · 2026. Ichelon Consulting Group. https://ichelonconsulting.com/reports/faq-schema-aio-citation-correlation-study-2026
Licence: CC BY 4.0 — free to reuse with attribution.
Four numbers to anchor the study
Methodology
1,842 URLs from the ICG SIE dataset — 155 healthcare and healthcare-adjacent web properties actively managed on the SIE platform — were classified into four cohorts by FAQ depth on the page: no FAQ, 3-4 FAQs, 5-8 FAQs, 9 or more FAQs. Cohorts were balanced on baseline organic authority (Domain Rating band) and on category concentration before analysis, so the observed citation-rate differences reflect FAQ depth rather than authority differences.
Every URL was probed against a matched keyword set — 4 to 14 keywords per URL, selected as the top-intent queries the page targets — on a weekly AIO probe crawl. Citation rate is defined as the share of AIO-carrying result pages, for the matched keyword set, in which the URL appears as a listed citation. 21,400 citation checks were executed across the four-month window. Cohort citation rates are the sample mean across all URLs in the cohort.
Attribute analysis. Every FAQ question inside every cohort was scored on six attributes: query-verbatim question phrasing, direct-answer opening sentence, proper-noun-or-number content, answer length in the 40-90 word band, spoken phrasing, and FAQPage schema markup presence. Attribute lift is the ratio of citation rate for questions carrying the attribute versus questions in the same cohort without it. Attribute effects are additive within observed limits — questions carrying five or six attributes cite at 3.4x the rate of questions carrying one or zero.
Anonymisation. No client names, page titles or URL strings appear anywhere in this report. Cohorts are reported at aggregate level.
Cohort results · FAQ depth versus citation rate
| Cohort | n URLs | FAQ presence | AIO citation rate | Lift vs baseline |
|---|---|---|---|---|
| No FAQ schema | 612 | 0% | 8.4% | — |
| FAQ present · 3-4 questions | 384 | 100% | 11.9% | +3.5 pp |
| FAQ present · 5-8 questions | 511 | 100% | 17.6% | +9.2 pp |
| FAQ present · 9+ questions | 335 | 100% | 21.3% | +12.9 pp |
Source: ICG SIE dataset (155 properties, 1,842 URLs) · AIO probe crawl, April-August 2026 · citation rate is share of AIO-carrying result pages where the URL is a listed citation.
FAQ depth versus AIO citation rate
Sample citation rate by FAQ-depth cohort
Six question attributes that predict citation
| Attribute | Citation lift | Why it matters |
|---|---|---|
| Question restates the query verbatim | +38% | Direct match between H3/question text and the search query |
| Answer opens with a direct answer sentence | +31% | First sentence answers the question in <=30 words, without preamble |
| Answer contains a proper noun or number | +22% | Attribution-friendly content Google can cite as a source-of-fact |
| Answer between 40 and 90 words | +19% | Sweet spot for AIO citation snippet length |
| Question uses natural spoken phrasing | +14% | Matches voice and AI-assistant query patterns |
| FAQPage Schema.org markup present | +11% | Structured signal on top of the HTML pattern |
Attribute lift = citation rate for questions carrying the attribute / citation rate for questions in the same cohort without it. Effects are additive within observed limits.
What to change on your existing FAQ blocks this week
1 · Rewrite the first sentence of every answer
Direct answer, 30 words or fewer, no preamble, no "in this article we will explore". This single edit accounts for +31 percent lift and takes about ten minutes per page.
2 · Restate the query verbatim in every question
Not paraphrase. Verbatim. If the target query is "how much does IVF cost in India", the question is exactly that, not "understanding IVF pricing in India". +38 percent lift.
3 · Push depth to 6-8 questions per commercial page
Below five, the lift is modest. At six to eight, citation rate roughly doubles vs the no-FAQ baseline. Above eight, marginal returns diminish.
4 · Add FAQPage schema even where no rich result renders
The schema still functions as a discovery signal for AIO and third-party assistants. Adding it is a one-time template change, worth ~11 percent lift on top of the HTML pattern.
Frequently asked — FAQ schema and AIO citation, 2026
Does FAQPage schema actually help with AI Overview citations?
How many FAQ questions should a page carry?
What makes a question likely to be cited in an AI Overview?
Is Google now ignoring FAQPage schema in search results?
What is the biggest single change to make on an existing FAQ block?
How does ICG operate FAQ writing against these findings?
Related ICG research reports
Want your FAQs audited against the six-attribute checklist?
Share your URLs. You will get a per-page FAQ scorecard, the six-attribute audit, and a citation-lift projection. Retainers custom-scoped per engagement, from Rs 20,000/month equivalent.