AI Visibility · 19 July 2026

By Eddie Eastwood, Founder

Writing Dental FAQs That ChatGPT and Perplexity Actually Quote (and the 2026 Rule That Rewrote the Advice)

Picture the moment a nervous patient finally decides to deal with the tooth that has been bothering them for months. They do not open ten browser tabs and compare practices. They ask an assistant a plain question, out loud or typed: "is a root canal painful, and who near me does them well?" In a couple of sentences the assistant tells them what to expect and names a practice.

If those sentences happen to be lifted from your website, you have just been recommended for free, at the exact second the patient was ready to act. If they are lifted from a competitor, you never find out the patient existed. There is no missed click in your analytics, no bounce to investigate. The enquiry simply went somewhere else.

That is the commercial weight sitting behind a question dental owners now ask us all the time: how do we write FAQ content that ChatGPT and Perplexity will actually quote? It reads like a small content question. It is really about where your most valuable enquiries are decided.

A single implant in the UK typically sits somewhere in the region of £2,000 to £2,500 per tooth (a figure that varies case by case and is given here only as a realistic illustration), a course of clear aligners commonly runs to several thousand pounds, and a patient who joins your books for routine care is worth years of check-ups, hygiene appointments and the occasional larger treatment.

These are precisely the decisions people now talk through with an assistant before they ever ring a practice. Getting your answer content into those conversations, or being shut out of them, is a real line on your future revenue.

There is also a specific change this year that flips the advice most existing guides still give, and because the topic is literally about FAQ pages, it matters here more than anywhere. The short version: the old SEO reason for building FAQ content has gone, and the AI reason is now stronger than it has ever been.

This guide explains why the FAQ format is the most quotable thing a practice can publish for AI answers, exactly how to write an entry that gets lifted, and how to stay correct on the 2026 reality that almost every rival page gets wrong. It is written for UK practices, it stays inside General Dental Council rules throughout, and every section answers a real question an owner has put to us, so you can read the one that fits your situation and act on it straight away.

Do FAQ pages actually get a dental practice cited by ChatGPT and Perplexity?

Yes, and the reason is structural rather than lucky. The FAQ format (a specific patient question paired with a short, self-contained answer) is the single easiest piece of content for an assistant to lift word for word and drop into its reply. It has already done the work the assistant would otherwise have to do itself.

Think about what these tools are built to produce. When someone asks an assistant a question, it returns a brief, direct response, often two or three sentences, sometimes read aloud. A well-written FAQ entry already is that response. The question matches what the patient typed, and the answer stands on its own without needing the paragraph above it for context. There is nothing for the assistant to summarise, trim or reassemble.

Compare that with the same fact buried halfway down a long treatment page, where the model has to locate it, separate it from its surroundings and compress it before it can use it. The FAQ has removed all of that friction in advance, which is why it gets quoted so readily.

The stakes for being the quoted source keep climbing. A Pew Research Center study of real browsing behaviour found that 58% of US users ran at least one Google search in March 2025 that returned an AI-generated summary, and that when a summary appeared, people clicked through to a traditional result in only 8% of visits, against 15% when no summary was shown ^[1].

More and more, the patient reads the answer and goes no further. When the click disappears, being named and quoted inside the answer is the visibility. And the assistants are not a fringe channel: BrightLocal's Local Consumer Review Survey 2026 reports AI tools like ChatGPT "surging into third place for local business recommendations" ^[2]. For a local service like dentistry, that is your prospective patients, asking exactly the questions a good FAQ answers.

The 2026 change every FAQ guide misses: Google removed the FAQ rich result

Here is the part that most dental marketing advice has not caught up with, and the single thing this guide most wants you to get right. For years the standard pitch was: add FAQ structured data so Google shows that expandable accordion of questions and answers under your listing, the "FAQ rich result". That visible badge was the headline reason marketers told dentists to build FAQ schema. As of 7 May 2026, it no longer exists.

Google's own changelog records it without ambiguity. The May 2026 entry is titled "Deprecating the FAQ rich result feature", noting that "this feature will no longer appear in Google Search starting May 7, 2026", and a follow-up entry in June 2026 confirms Google "removed documentation for the FAQ rich result feature" because it "is no longer shown in Google Search results" ^[3]. So if you are reading older guidance that promises a search-results accordion in return for FAQ markup, that guidance is describing something that has been switched off for everyone.

The trap is drawing the wrong conclusion from this. What Google retired is the SERP feature, the decorative panel in the search listing. It did not retire the FAQ content format, and it has no bearing at all on how ChatGPT, Perplexity or an AI summary treat your page. So this change does not weaken the case for writing answer-first question-and-answer content.

It sharpens it, because the entire value of an FAQ now sits where it always mattered most: being the clean, liftable answer an assistant extracts and a voice device reads aloud. The badge is gone. The reason that actually carries the decision is more important than ever. Our companion piece on dental blog posts versus FAQ pages works through how the two formats divide the labour once you stop chasing that retired feature.

How do ChatGPT and Perplexity decide which FAQ answer to quote?

The plain answer is that they do not all decide the same way, and the differences are worth knowing because they change where your content needs to be strong. What follows is observed behaviour rather than anything the engines publish as a rulebook, so treat it as a working model rather than gospel.

ChatGPT, when it searches the live web, leans heavily on Bing's index and on the established directories and sources Bing already rates. That means an FAQ answer of yours is far likelier to surface through ChatGPT if your practice is consistently and clearly represented across the web, not just on your own site. Perplexity behaves differently: it favours fresh, current web content, shows its working with numbered citations, and will happily pull from recent pages and community discussion. Google's AI summaries weight content that is clearly structured and that answers the question directly and unambiguously on the page.

Pull those three together and the common thread is obvious. Every one of them rewards the same thing: a clean, self-contained answer to the literal question the patient asked, sitting on a website and an entity the model already trusts. The FAQ format nails the first half of that by design. The trust half depends on the wider picture of how your practice appears across the web, which is the subject of a later section. Get both and you are the answer the assistant reaches for; get only one and you are passed over for a practice that has both.

What makes a dental FAQ answer get lifted? The formula competitors leave out

This is where most pages on this topic stop short. They tell you the FAQ format is extractable and leave it there. The useful part is the method, so here it is as a checklist you can hold every answer against.

• Use the patient's exact words in the question. People do not ask about "endodontic discomfort management". They ask "does a root canal hurt?" Write the question the way a patient would actually say it, because that is the phrasing the assistant is matching against. The closer your wording is to the spoken query, the likelier the match.
• Put the direct answer in the very first sentence. Do not warm up. The opening line should resolve the question on its own, so an assistant can take that one sentence and have something complete. Everything after it is supporting detail.
• Make every answer self-contained. No "as mentioned above", no "as we covered earlier", nothing that depends on a neighbouring paragraph. The assistant lifts the block in isolation, so it has to make sense in isolation.
• Keep it to two to four sentences, specific not vague. Long enough to be genuinely useful, short enough to be liftable whole. A one-line dodge gives the model nothing worth quoting; a six-paragraph essay is no longer an FAQ answer. Specifics (a realistic price range, a timescale, a clear yes or no) are what make it worth extracting.
• Make sure the answer is genuinely on the page. This sounds obvious, but it is the rule that catches people out, and Google states it plainly (covered in the next section). You cannot conjure a quotable answer with markup alone; the words have to really be there for a visitor to read.

A quick test gathers all of this into one move: read the answer aloud. If it sounds like a complete, natural reply to the question when spoken, a voice assistant can use it. If it trails off, hedges without landing, or only makes sense after the sentence before it, it will not extract cleanly. That read-aloud check is the fastest quality gate you have, and it costs nothing.

Weak FAQ versus citable FAQ: worked dental examples

Abstract advice is easy to nod along to and hard to apply, so here are real dental questions shown twice, once in the buried or evasive form that gets ignored, and once rewritten to be lifted. The underlying facts are the same; the shape is what changes.

"Does a root canal hurt?" Weak (buried, hedged): "Many patients have concerns about discomfort, and it is natural to feel anxious. Our caring team will always do everything possible to make your visit as comfortable as it can be, and we are happy to discuss your individual needs." That answers nothing. There is no fact to extract. Citable: "A root canal is carried out under local anaesthetic, so most patients feel little or nothing during the treatment itself. Mild tenderness for a day or two afterwards is normal and usually settles with ordinary pain relief." Direct, specific, complete in itself.

"How much does a dental implant cost in the UK?" Weak: "Implant pricing depends on a number of factors, so please contact us for a personalised quote." True, but useless to a patient comparing options, and nothing an assistant can quote. Citable: "A single dental implant in the UK typically costs somewhere in the region of £2,000 to £2,500, depending on your individual case and whether any additional treatment is needed first. A full assessment gives you an exact figure before anything goes ahead." It gives a realistic, clearly illustrative range and stays candid about variation.

"Can I get an emergency dental appointment the same day?" Weak: "We understand emergencies happen and aim to help wherever we can." Vague reassurance, no information. Citable: "Yes, we keep appointments aside each day for dental emergencies and aim to see patients in pain the same day where possible. Phone the practice as early as you can, as same-day slots fill quickly." Concrete, actionable, liftable.

"Is Invisalign better than fixed braces for an adult?" Weak: "Both have their advantages and the right choice depends on your circumstances." Technically correct, completely unquotable. Citable: "For adults, clear aligners like Invisalign are removable and far less visible, which suits work and social life, and they typically treat mild to moderate alignment issues. Fixed braces can be the better option for more complex cases. A consultation will confirm which is right for your teeth." It actually helps the patient decide and gives the assistant something to say.

Notice the pattern in every pair. The weak version protects the practice from committing to anything; the citable version answers the question a real person asked. The citable ones are also, not coincidentally, the ones a patient finds genuinely helpful, which is the whole point. You are not writing for the machine at the expense of the person. The same qualities that make an answer extractable make it useful.

Where should FAQ content live: a dedicated page or sections on each service page?

The instinct is to build one big FAQ page and pile every question onto it. Usually that is the weaker choice. FAQ content tends to perform better as focused sections sitting on the relevant treatment or service page, because the answer is then surrounded by the very topic the assistant is already reading about.

Here is why context matters. When an assistant is weighing up your implants page, a cluster of implant questions answered right there (cost, pain, healing time, lifespan) reinforces that this page is a thorough, trustworthy source on implants specifically. The same questions stranded on a generic catch-all FAQ page, three clicks from anything about implants, carry far less of that signal. The answer is the same; its setting is what changes how confidently the model attributes it to your expertise on that treatment.

There is one rule to hold to whichever way you go: give each question and answer a single canonical home. Repeating an identical Q&A across a service page, a dedicated FAQ page and a blog post, with no clear sense of which is the authoritative version, muddies the picture and wastes the effort.

Decide where each answer lives, put it there properly, and link to it from the other places rather than duplicating it wholesale. For how this fits the broader question of which content format to build first, our piece on blog posts versus FAQ pages lays out a publishing order that stops you doing the work twice.

Do I still need FAQ schema if there is no rich result?

Yes, but for straightforward reasons now, not the old one, and it pays to be clear about exactly what the markup does and does not buy you in 2026.

FAQPage is still a valid type in the structured-data vocabulary; it did not vanish when the rich result was switched off ^[4]. Adding it continues to do a few genuinely useful things. It labels your question-and-answer content clearly for any system that reads markup, it lets you validate that your FAQ blocks are well-formed, and engines beyond Google may still parse it. Pairing it with the Dentist type (a formal Schema.org type, a subtype of LocalBusiness and MedicalBusiness, that carries your address, phone, opening hours and the rest) ties those answers to a clearly defined practice ^[5].

What FAQPage markup no longer does, as of 7 May 2026, is produce that expandable panel in Google's search results, because the feature is gone for everyone ^[3].

So the sensible position is to add the markup for correctness, not for badges. Mark up FAQ content that genuinely appears on the page, keep it accurate and matched to the visible words, and then put the bulk of your effort into the thing that actually wins the citation: the clarity and usefulness of the answers themselves.

Do not let anyone sell you FAQ schema on the promise of a rich result, because there is no longer one to deliver. If you want the implementation detail, the copy-paste JSON-LD and the review-markup rule most guides handle badly, our dental schema markup guide is the companion to this article and covers the how-to in full.

Why your FAQ will not get cited if your practice looks fuzzy to AI

A perfectly written FAQ answer can still go unquoted, and when that happens the problem is usually not the answer. It is the foundation underneath it. An assistant does not just need a good answer; it needs to be confident about whose answer it is. If your practice looks ambiguous or inconsistent across the web, the model hedges by quoting a directory or a competitor it trusts more, even if your answer was better.

That confidence is built from the things that establish you as a clearly defined entity: a consistent name, address and phone number everywhere your practice appears, a complete and accurate Google Business Profile, the Dentist schema that spells out who you are, and a steady flow of recent reviews. BrightLocal's 2026 survey makes the link explicit, describing reviews as evidence that a business is "worthy of prominent visibility and citation within traditional Google search and LLMs like ChatGPT, and AI search", and noting that 97% of consumers still lean on reviews to guide their decisions ^[2].

In other words, the review profile that reassures patients is the same signal that reassures the assistant it is safe to name you.

The practical takeaway is that FAQ work and entity work are not separate projects. A brilliant answer on a practice the model is unsure about is a wasted brilliant answer. Get the foundation solid and your liftable answers finally have a trusted source to be lifted from. We go deeper on this in our guide to how dental practices earn AI citations, which covers the trust side of the equation in detail.

How do I keep dental FAQ answers compliant with GDC rules?

A citable answer and a compliant answer are the same answer, which is a relief rather than a constraint. The General Dental Council expects claims to be accurate and not misleading, and the assistants reward exactly the same candour, so writing within the rules tends to improve your extraction rather than hold it back.

Keep a few things front of mind. Avoid superlatives you cannot stand behind, the "painless", "best", "cheapest" kind of language, because they are both a compliance risk and the sort of empty claim an assistant has no reason to quote. Do not promise clinical outcomes or guarantee results; describe what a treatment typically involves and what a patient can reasonably expect.

Where you give a price, present it as a realistic, clearly illustrative range and make plain that the exact figure follows an assessment, which is both candid and more useful than a flat refusal to indicate cost. None of this makes your answers weaker. An answer that is accurate, specific and free of overclaiming is precisely the kind a model is comfortable extracting and a regulator is comfortable seeing. Compliance and citability pull in the same direction here.

What are the mistakes that stop dental FAQs getting cited?

Most wasted FAQ effort comes down to a handful of avoidable errors. Run through these before you commit time to writing or rewriting.

• Burying the answer instead of leading with it. If the first sentence is preamble and the actual answer arrives in sentence four, the assistant has nothing clean to take. Lead with the answer, every time.
• Writing in jargon rather than patient language. A question phrased in clinical terms will not match how a patient actually asks, so the match never happens. Use the words people use.
• Thin, hedged answers with nothing to extract. A question followed by a vague non-answer is the commonest failure of all. If there is no fact, range or clear yes or no in there, there is nothing to lift.
• Building FAQs expecting the rich result. That feature has been gone since May 2026, so any effort aimed at the accordion badge is spent on nothing. Build for extraction and patient clarity instead.
• Duplicating the same Q&A everywhere with no canonical home. Repeating answers across pages with no authoritative version confuses which page should be quoted and dilutes them all.
• Marking up answers that are not on the page. Schema describing invisible content breaks Google's own rule and earns you nothing, as the next point makes clear.

That last one deserves spelling out, because it is the shortcut people most want to take. Google's structured-data guidance is explicit: "Don't create blank or empty pages just to hold structured data, and don't add structured data about information that is not visible to the user, even if the information is accurate" ^[6]. You cannot mark up an FAQ answer that a visitor cannot actually read on the page. The markup is only ever permitted to describe content that is genuinely there, which means format and schema can never substitute for the work of writing a real, useful answer.

How can I check whether AI already quotes my FAQ content?

You can run a useful self-audit today, free, in about a quarter of an hour, and it tells you plainly whether your answer content is pulling its weight.

Start by asking the assistants the questions your patients ask. Open ChatGPT, Perplexity and Google's AI Mode and put in the real queries: "does getting a dental implant hurt", "emergency dentist near me open today", "how much is Invisalign". Note whether your practice is named at all, and when your content does appear, notice whether the assistant is lifting one of your FAQ answers or paraphrasing something else. Run each prompt a few times, because the responses shift between runs, and it is the recurring pattern rather than any single answer that tells you where you stand.

Then read your own FAQ answers aloud, one by one. This is the quickest test of extractability there is. If an answer sounds like a complete, natural spoken reply, an assistant can use it. If it depends on the line before it, hedges without resolving, or fizzles out, it will not extract, and that is a fixable problem you have just found for free.

Where the assistants are quoting competitors and skipping you, the cause is almost always one of two things: your answers are not liftable, or your practice is not a clearly enough defined entity for the model to cite with confidence. This audit shows you which, and that points straight at what to fix first.

How QBiz gets your answers quoted, on your site and across the web

Most providers in this space sell you one slice of the job and leave the rest to chance. One firm writes a few FAQs; another tidies your markup; nobody owns the thing that actually matters, which is whether your content is being drawn into the answers patients read. QBiz is built around closing that whole gap, and the work runs on two connected fronts.

On your own site, we make your answers genuinely liftable. That means writing FAQ content where each answer is self-contained, leads with the point, and uses the words patients actually say, placed in context on the service pages that carry your most valuable treatments rather than dumped on a catch-all page. We mark it up correctly, including FAQPage and Dentist schema for clarity and validation rather than for a rich result that no longer exists, so the engines read your real content without guessing and attribute it to your practice.

And we make sure the substance is there underneath the format, because, as Google's own rule insists, markup can only ever describe answers a visitor can really see.

The second front is the one on-site work alone cannot reach: getting your practice and your content out across the wider web so the assistants trust you enough to quote you. A clean, well-written site is necessary but not sufficient, because an assistant's confidence in naming you draws on seeing your details and your content referenced consistently across the directories, profiles and sources the models actually read.

So we also do the distribution work, getting your content published and circulated, and your practice details agreeing with each other everywhere they appear, which is what turns solid groundwork on your own pages into the off-site corroboration that earns a citation. The on-site answers and markup are the foundation; distributing your content and presence across the web is what converts that foundation into answers that name you rather than the practice down the road.

It begins with a QBiz AI Visibility audit. We ask the assistants the questions your prospective patients ask, show whether your practice is named at each stage, check whether your FAQ answers are genuinely being lifted, and confirm that your markup matches the visible page and your details agree across the web. You get back a prioritised, GDC-safe plan covering both the on-site answer work and the distribution, with the highest-value cases first. It is the no-cost first step, and it shows you exactly where the patients you should be winning are slipping away.

Frequently asked questions

Do FAQ pages help dentists get cited by ChatGPT and Perplexity?

Yes. The FAQ format (a specific patient question with a short, self-contained answer) is the most extractable content there is for an AI summary or a voice answer, because the assistant can lift the whole block cleanly without having to summarise a longer page. The same answer buried in a long article is harder to extract. To get cited, write each answer to stand alone in two to four sentences, lead with the point, and phrase the question the way a patient actually asks it.

Is FAQ schema still worth adding now that Google removed the FAQ rich result?

It is worth adding for correctness, not for a badge. FAQPage is still valid markup: it labels your question-and-answer content clearly, lets you check it is well-formed, and may be parsed by engines other than Google. What it no longer does, since 7 May 2026, is produce the expandable FAQ panel in Google's search results, because that feature was removed for everyone. Mark up the answers genuinely on your page, then spend most of your effort on the quality of those answers.

How long should a dental FAQ answer be for AI to cite it?

Two to four self-contained sentences is the sweet spot. The first sentence should answer the question directly, and the whole thing should make sense on its own without depending on the paragraph above it. A one-line dodge gives an assistant nothing to extract, and a long essay stops being a liftable answer. Read it aloud: if it sounds like a complete spoken reply, it is the right length.

Should I put FAQs on one page or on each service page?

Usually on each service page, as focused sections, rather than on a single catch-all FAQ page. An answer sitting in context with the treatment it is about reinforces that the page is a thorough source on that topic, which an assistant rewards. Give each question and answer one canonical home, and link to it from elsewhere rather than duplicating it across several pages.

Can I pay to get my FAQ content cited by ChatGPT?

No. There is no paid placement that buys you a citation in ChatGPT, Perplexity or an AI summary. Being quoted is earned through clear, useful, self-contained answers on a practice the model trusts, backed by consistent details and reviews across the web. What you can pay for is the work of getting your content and your practice into the state where the assistants choose to cite you, which is a different thing from buying the citation itself.

Where to start

If you take three things from all of this, take them in order. First, rewrite your most important treatment FAQs so each answer leads with the point, stands on its own in two to four sentences, and uses the words a patient would actually say, because that is what gets lifted into the answer a prospective patient reads.

Second, stop judging FAQ content by the rich result, which has been gone since May 2026, and judge it by whether it extracts cleanly and helps a real person, marking it up for correctness rather than for a badge. Third, make sure the practice underneath those answers is a clearly defined, well-reviewed entity across the web, because the best answer in the world goes unquoted if the model is not confident whose it is. Each step sits comfortably inside GDC rules, and each makes the next one count for more.

If you would rather see exactly where your practice drops out of the AI answer, whether the gap is FAQ answers too thin to lift or an entity too fuzzy to cite, that is the job of a QBiz AI Visibility audit. We ask the engines the questions new patients ask, show whether your practice is named at each stage, and return a prioritised, GDC-safe list of what to fix, on your own site and across the wider web. It is the no-cost first step before any spend.

Get your AI Visibility audit →