Patients Are Googling Far Less And Asking AI A Lot More.

Answer engine optimization (AEO) is the science of making your organization the source that AI platforms cite when it answers a potential patient’s question. It matters now because a third of Americans already use AI platforms for health questions, and roughly 90% of what those assistants cite is content the brand does not own or control. As a result, we are seeing substantial loss of web traffic to practice-level healthcare websites.

Key Facts About AI Prompts

1. 34% of Americans now use AI platforms for health and medical questions, including 25% who use them to help interpret symptoms and 15% who use them to decide whether to see a doctor at all.
2. Brand-owned domains account for only about 10% of AI citations, and just 2.2% on unbranded discovery queries. The rest comes from third parties.
3. The three tactics most commonly sold as “AEO” right now – schema markup, llms.txt files, and listings consistency – have weak or no evidence behind them.
4. AI is taking the research layer of search, not the transactional layer. Someone searching “orthopedic surgeon in Tampa” still gets a local pack. Someone asking “should I see a surgeon for a torn meniscus” gets an answer.
5. Citation share is the new ranking. Most healthcare organizations are not measuring it at all, which means they cannot tell whether they are gaining or losing.

What is answer engine optimization, and how is it different from SEO?

SEO earns you a position on a page of links. AEO earns you a mention inside an answer. The distinction sounds academic until you look at what happens after the answer appears: in the first four months of 2026, 68% of US Google searches ended without a click, up from 60% in 2024, according to SparkToro’s analysis of Similarweb panel data. The link you worked to earn is increasingly a link nobody presses.

That is the mechanical difference. The strategic difference is harder and more interesting. In classic SEO, you competed for a slot on a page. In AEO, you compete to be the source a model trusts enough to name, and the model is assembling its answer from many places at once, most of which are not your website.

Patients are already asking AI. This is no longer speculative.

The strongest data here is not from a vendor with a tool to sell. In August 2026, the Pew Research Center surveyed 3,488 US adults and found that 34% use AI chatbots for at least one health or medical reason. Twenty-eight percent use them for quick health information, 25% to help diagnose symptoms, 22% to understand a diagnosis a physician already gave them, and 15% to decide whether to see a doctor at all. Among adults aged 18 to 29, adoption reaches 44%.

Two months earlier, Pew found that 49% of US adults use AI chatbots generally, up from 33% in 2024, and that 60% have read an AI-generated summary at the top of search results. On the local side, BrightLocal’s March 2026 consumer survey put the share of consumers using AI for local business recommendations at 45%, up from 6% a year earlier.

Healthcare feels this sooner than most categories because search engines surface AI answers there more aggressively. BrightEdge’s tracking put AI Overview coverage on healthcare queries at 88% to 89% by December 2025, up from 59% two years earlier. A separate WebFX study of 130,070 US healthcare queries put it lower, at 51% as of July 2025. Both are vendor studies, and they don’t reconcile, because each tracks a different keyword set. The honest read is a range, and either end of that range is roughly double the cross-industry average.

Why most healthcare brands are invisible in AI answers.

Here is the finding that reframes the whole scenario. Foundation Marketing and AirOps analyzed 57.2 million citations across 5.1 million AI responses between December 2025 and February 2026. Brand-owned domains accounted for 10.15% of citations. On unbranded discovery queries, the kind a prospective patient actually asks, brand-owned citations fell to 2.2%.

Roughly nine out of ten citations point somewhere the brand does not control. In healthcare, that means PubMed, Mayo Clinic, health systems with strong editorial libraries, review platforms, Reddit threads, and news coverage. For local and location-level queries, the concentration is even sharper: Foundation and AirOps found Yelp took 72.5% of local citations inside Google AI Mode and 62.1% on Perplexity.

The visibility gap this creates for multi-location organizations is measurable. SOCi’s 2026 Local Visibility Index, built on roughly 350,000 locations across 2,751 multi-location brands, found AI assistants recommended between 1% and 11% of business locations, against 35.9% that appeared in Google’s local three-pack. ChatGPT recommended 1.2% of locations. Perplexity, 7.4%. Gemini, 11%.

If you run 60 clinics, the number of them an AI assistant will name today is closer to one than to twenty.

The Citation Surface: the four layers AI actually reads

We use a simple model with clients to make this concrete. Your citation surface is everywhere an AI engine can learn about you, and it has four layers. Most organizations invest heavily in the first and ignore the other three.

1. Your owned content. Service line pages, location pages, clinician bios, and editorial. This is the 10% layer. It still matters, because it is the only layer you write yourself, and because it is what gets pulled once someone asks about you by name.
2. Your profiles. Google Business Profile, Yelp, Healthgrades, Vitals, insurer directories, hospital affiliation pages. In local AI answers, this layer often outranks your own website as a source.
3. Third-party description. Reviews and their content, physician directories, local news, association listings, community forums. This is where most of the citation weight sits, and it is the layer almost nobody manages deliberately.
4. Entity coherence. Whether all of the above describes one consistent organization with consistent service lines, locations, and clinicians. Models are resolving an entity before they cite it. Contradictory information does not just fail to help; it actively suppresses you.

For service organizations absorbing acquisitions, layer four is usually the broken one. Every practice you acquire arrives with its own legacy profiles, its own review history, its own directory entries under a previous name. Two years of add-ons produce an entity an AI engine cannot resolve into a single organization, and the answer goes to a competitor with a cleaner record.

The transactional search is still yours. The research search is not.

One nuance keeps this from becoming a panic. AI is not consuming all of search evenly. Whitespark analyzed 540 local queries across three metros and six industries, and the split by intent is stark. Pure local intent, such as “dermatologists in Phoenix,” returned a local pack 93% of the time and an AI Overview only 15% of the time. Informational queries, such as “how much do dental implants cost,” returned an AI Overview 92% of the time. Hybrid queries, cost or condition plus a city, hit 97%.

Read that carefully, because it tells you where to spend. AI is taking the research layer, where patients form their shortlist and their expectations. Your local pack presence, your reviews, and your booking flow still own the transactional moment. The risk is not that AI steals the appointment. It is that by the time the patient reaches the transactional search, a competitor has already been named as the expert.

How do you measure AEO?

Clicks were never a perfect metric, but they were at least countable. Citation share is harder and more important. Most healthcare marketing teams we talk to have no baseline at all, which means they cannot tell whether the last two quarters helped or hurt.

What you measured before	What to measure now	Why it matters
Keyword rankings	Citation frequency by engine, by service line, by market	91% of AI citations appear on only one engine, per Kevin Indig’s H1 2026 analysis. Winning on Perplexity tells you nothing about ChatGPT.
Organic sessions	Share of voice against named competitors in AI answers	Absolute citation counts move with query volume. Relative share tells you whether you are gaining.
Branded vs. non-branded traffic	Mentions without citations	Semrush and Kevin Indig found 61.7% of appearances were “ghost citations,” where a source is cited but the brand is never named. You can be the source and get no credit.
Referral traffic	AI referral traffic, tracked separately	It is small today. Chartbeat data reported in March 2026 put ChatGPT referrals at under 1% of total referrals even after growing more than 200%. Track it because the trend line matters, not the current number.

MDG has developed a methodology in-house rather than just trusting a reseller dashboard, and this includes building our own connections into Search Console and GA4, so the citation data sits next to the acquisition data instead of in a separate tool nobody opens. If you want the mechanics of the underlying acquisition math, our breakdown of calculating true patient acquisition cost is the companion piece to this one. Our approach is all informed by more than 25 years of doing multi-location, high-growth marketing for PE-backed companies.

Where to start in the next 90 days

Four moves, in order, and none of them require a new platform subscription.

1. Baseline your citation share. Pick your top revenue-driving queries per service line. Run them across ChatGPT, Perplexity, Gemini, and Google AI Mode. Record who gets named. This takes a week, and it is the only way to know where you actually stand.
2. Fix entity coherence first. Reconcile the organization name, service lines, locations, and clinician records across your site, your profiles, and every directory carrying a legacy practice name. This is the highest-yield work for any group that has acquired in the last three years.
3. Rebuild your highest-intent content as answers. Cost, condition, procedure, and “should I” questions, answered directly in the first 40 words of a section, with the question phrased the way patients actually ask it. Our guide to entity-based SEO covers the structural side in more depth.
4. Manage the third-party layer deliberately. Review velocity and review content on the platforms AI engines actually pull from, physician directory accuracy, and earned coverage. This is the 90% layer, and it is where the durable advantage sits.

Healthcare is a YMYL category, which means search engines and AI models apply a higher bar for authority and trust here than in almost any other vertical. That cuts both ways. It is harder to earn citation share in healthcare, and much harder for a competitor to take it from you once you have it. Right now the field is wide open, because most healthcare organizations are still optimizing for a results page their patients are increasingly skipping.

Frequently asked questions

What is answer engine optimization?

Answer engine optimization is the practice of structuring content, profiles, and third-party presence so that AI assistants such as ChatGPT, Perplexity, Gemini, and Google AI Mode cite your organization when answering a relevant question. Unlike SEO, which competes for a position in a list of links, AEO competes to be named inside the answer itself.

Is AEO different from GEO?

The terms overlap and are often used interchangeably. In practice, generative engine optimization (GEO) usually refers to being referenced by generative platforms broadly, while answer engine optimization (AEO) refers to being the extracted answer in AI Overviews, AI Mode, and voice results. Both sit alongside SEO rather than replacing it, and the underlying work is largely shared.

Does schema markup help my practice get cited by AI?

The best available evidence says the effect is minimal, but we expect it to become more significant with adoption. Ahrefs tracked 1,885 pages that added schema between August 2025 and March 2026 and found AI citations barely moved, and Google’s documentation states structured data is not required for generative AI search, but it is still being documented as a best practice. Schema still earns rich results in classic search, so it remains worth implementing. It is not an AI visibility strategy on its own.

How many of my locations show up in AI search?

Probably very few. SOCi’s 2026 Local Visibility Index found AI assistants recommended only 1% to 11% of business locations across 2,751 multi-location brands, compared with 35.9% appearing in Google’s local three-pack. The only way to know your own number is to test your locations directly across each engine.

How long does it take to see results from AEO?

Entity and profile corrections can change AI answers within weeks because they update sources the models re-crawl frequently. Content and authority work runs on a longer cycle, generally one to two quarters before citation share moves measurably. Both require a documented baseline first, or you will not be able to prove either one worked.

Should healthcare organizations worry about accuracy in AI answers?

Yes, and it is an underrated operational risk. Research reported in mid-2026 found roughly two-thirds of patients using AI to research providers encountered incorrect provider information. Wrong locations, outdated affiliations, and retired service lines circulate in AI answers, and correcting them means correcting the underlying sources rather than the AI itself.

Request an AEO Visibility Audit

MDG delivers AEO alongside SEO, GEO, and local search as one in-house discipline: entity and schema engineering, structured content, listings accuracy across 60 or more directories, and AI visibility auditing and tracking. We work with multi-location healthcare organizations and their sponsors, currently serving more than 2,700 client locations across 48 states, with an average client relationship of eight years.

An AEO Visibility Audit shows you how often AI engines cite your brand versus your competitors, by service line and by market, with a prioritized fix list. Start the conversation here, or read more about how we support multi-location healthcare organizations and private equity sponsors and their portfolio companies.