Most FAQ sections read like they were written by a committee — vague questions, hedged answers, and half the queries your visitors are actually typing into Google left uncovered. Writing a genuinely useful FAQ section takes research most site owners skip: figuring out exactly what people ask before they buy, sign up, or contact support. AI tools shortcut that research and draft the answers, cutting a task that used to take an afternoon down to about twenty minutes.
In most sites I build, the FAQ section is one of the last things added and one of the first things visitors read. What’s changed recently is why it’s worth doing well — the SEO payoff most people assume comes with a good FAQ section quietly disappeared, and treating it as the main reason to write one will lead you to the wrong priorities.
The FAQ Schema Reality Check
If you’ve read older guides on this topic, you’ve probably seen FAQ sections framed as a way to win extra space in Google’s results with a rich snippet — the expandable question list that used to show up directly under some search listings. That feature is dead. Google restricted FAQ rich results in August 2023 to a narrow list of well-known government and health websites, after years of sites gaming it with artificial questions that had nothing to do with genuine user intent. Then, as of May 2026, Google removed FAQ rich results entirely — even for the government and health sites that had kept the narrow exception. Google quietly added a deprecation note to its FAQ structured data documentation rather than making a formal announcement, and the reporting tools and Rich Results Test support for it are being phased out through mid-2026.
None of that means FAQPage schema will hurt your site if it’s already there, and it doesn’t mean skip writing FAQ content. It means the actual reason to build a good FAQ section is the one that was always true underneath the SEO pitch: real visitors have real questions, and answering them well reduces support tickets and keeps people from bouncing off a page in confusion. If a plugin or an old guide is telling you to add FAQPage JSON-LD purely to “win a rich result,” that specific payoff no longer exists for ordinary sites.
Where the Real Questions Actually Come From
Before opening an AI tool, pull the actual questions people ask. Check support emails, comment sections, and any keyword research you’ve already done. If you sell a product, look at reviews — the confusion buyers mention publicly is exactly what belongs in your FAQ. I’ve seen product pages cut support tickets noticeably just by adding five well-chosen questions pulled from actual customer emails rather than guessed from scratch.
If you’re starting with nothing, ask the AI tool itself: “List the 20 most common questions someone would have before buying [your product/service].” Treat the output as a starting list to trim and verify against your actual policies, not a finished set — AI has no visibility into your specific refund window, shipping carrier, or cancellation process unless you tell it.
Prompting for Usable Drafts, Not Finished Copy
Give the AI tool specific facts — your actual pricing, policies, and timelines — rather than asking it to invent answers. A prompt like “Write a short, direct answer to ‘What’s your refund policy?’ using these terms: 30 days, unused items only, buyer pays return shipping” produces something usable. A vague prompt produces a generic answer that sounds fine and says nothing.
Ask for answers in the two-to-four sentence range. FAQ answers that run long defeat the purpose — visitors scanning for a quick answer skip past a paragraph.
Fact-Check Every Answer Before It Goes Live
This is the step that gets skipped and shouldn’t be. AI tools will confidently state a shipping window, a compatibility detail, or a policy term that isn’t accurate for your business. Read every answer against your actual current policies before publishing anything. An FAQ section with one wrong answer about refunds or pricing causes more support tickets than having no FAQ at all — it converts a visitor’s uncertainty into a false certainty that turns into a dispute later.
Structure Still Matters, Even Without the Rich Snippet
Losing the FAQ rich snippet doesn’t mean heading structure stopped mattering. Format each question as an H3 heading with the answer directly below it in a paragraph block — not bolded text inside one long paragraph. This is what lets Easy Table of Contents build a working outline automatically, and it’s still the structure most likely to get pulled into a featured snippet or an AI Overview answer, both of which are separate mechanisms from the old FAQPage rich result and are still very much active. Clean question-and-answer structure also just makes the page easier for a human to scan, which is the part that was never dependent on a Google feature in the first place.
If you’re weighing whether to bother with FAQPage schema markup at all now, the honest answer is: it’s low-risk to leave it in if you already have it, since it doesn’t get flagged as spam and could theoretically still matter for some future Google feature, but it’s not worth new engineering effort. Spend that effort on the actual content and heading structure instead.
A Worked Example: Trimming Twenty Questions to Eight
A small SaaS tool I reviewed recently asked its AI tool for the standard “20 most common questions before buying” prompt and got a competent but generic list — things like “Is there a free trial?” and “What payment methods do you accept?” Cross-referencing that list against six months of actual support tickets cut it down to eight questions, and three of the eight weren’t on the AI’s list at all: one about whether the tool worked with a specific accounting platform, one about data export on cancellation, and one about whether a solo freelancer needed the team-tier plan. Those three came directly from repeated support emails and would never have come from asking an AI to guess — they were specific to that product’s actual confusion points, not generic SaaS questions.
Practical Tips
- Keep the FAQ section to 6–10 questions on most pages — a wall of 30 questions buries the ones visitors actually need.
- Order questions by how often they’re asked, not alphabetically or by internal logic. The most common question goes first.
- Revisit the FAQ section every few months and add questions from new support tickets — it should grow with real usage, not stay static after launch.
- Avoid duplicating exact wording from your main page copy elsewhere on the page — search engines can treat heavily repeated blocks as thin content.
Common Mistakes
- Publishing AI-drafted answers without checking them against actual current policies or pricing.
- Adding FAQPage schema purely chasing a rich snippet that no longer exists for ordinary sites as of 2026.
- Writing questions the business wants to promote rather than questions customers actually ask.
- Letting every answer keep the same generic AI phrasing across the whole page, which makes the whole section read as impersonal.
When AI-Assisted Drafting Makes Sense
AI-assisted drafting works well for products and services with a stable, well-defined set of policies — the facts don’t change often, so the fact-checking step stays manageable. For fast-changing information — live pricing tiers, frequently updated shipping rules, or anything legally sensitive — write those answers yourself and use AI only for phrasing suggestions, since the risk of a stale or invented detail is higher. If your site is still finding its audience, start by building a dedicated FAQ page before adding shorter FAQ sections to individual product or service pages.
Use AI to speed up research and first drafts, treat every answer as unverified until checked against your actual policies, and stop building the section around a rich-snippet payoff that no longer applies. Real questions, fact-checked answers, and clean heading structure are what make an FAQ section useful — that was true before the schema changes and it’s still true now.
Etienne Basson works with website systems, SEO-driven site architecture, and technical implementation. He writes practical guides on building, structuring, and optimizing websites for long-term growth.