Answer Engine Optimization for Solo Healthcare Practices

A web design client told me recently that they’d stopped worrying about SEO. It isn’t relevant anymore, they said. What they needed was to get optimized for AI instead.

They hadn’t invented that. Somebody told them, and whoever told them almost certainly had a retainer ready. It’s a good pitch, too. Search is dead, AI replaced it, and the person you’re already paying doesn’t do the new thing.

The reason it’s hard to argue with is that some of it is true. The measured changes are genuinely large, and I’ll show you them, because you should see the numbers before anyone asks you for money.

Here’s what the pitch leaves out. In 2026 Google published a page listing the popular answer engine optimization tactics you can safely ignore. Not a competitor’s blog post. Google’s own documentation. And the Forrester analyst who covers this called AEO “significantly, but not fundamentally, different from SEO.” Fundamentally is doing a lot of work in that sentence, and for once it’s doing it in your favour.

That’s three acronyms for one job now, and I’d bet somebody is workshopping a fourth.

Answer engine optimization is real. It just isn’t a new discipline you have to buy. Here’s what a solo practice actually has to change, what it can leave alone, and how to tell which vendor claim you’re being sold.

01What Is an Answer Engine, and What Is Answer Engine Optimization?

An answer engine answers the question instead of handing you a list of pages to go and read. Google’s AI Overviews, Google’s AI Mode, ChatGPT, Perplexity and Copilot are all answer engines. Answer engine optimization is the work of becoming a source those systems quote, rather than a link they skip.

That’s the honest definition, and it’s about as far as the agreement goes. There isn’t a settled line between AEO and its sibling acronym GEO, generative engine optimization. Practitioners, vendors and publications use the two interchangeably and define them differently from each other, which is a decent sign that the distinction is being sold rather than observed. Google refuses to treat either as a separate thing. Its guidance says that “from Google Search’s perspective, optimizing for generative AI search is optimizing for the search experience, and thus still SEO.”

So when a vendor tells you AEO is a new discipline, the correct response isn’t yes or no. It’s which engine, and what specifically.

Here’s the part that makes the terminology worth arguing about anyway. Nikhil Lai, the Forrester principal analyst who covers this, put it plainly in November 2025. “AEO is significantly, but not fundamentally, different from SEO.” He also named why the acronyms multiply. Point solutions popularising them, he wrote, “tend to exaggerate SEO and AEO’s differences.” Somebody sells the gap.

02AEO vs SEO vs GEO, in One Table

Three terms, and only one clear difference between them. If you only keep one thing from this article, keep the table. It’s the fastest way to check whether an answer engine optimization pitch is describing real work or renamed work.

Scroll the table sideways to compare all three.

The row that matters is the third one. Two of the three cells say some version of “keep doing the fundamentals.” That is not me being lazy. That is what the primary sources say, and section 04 is Google saying it in its own words.

03What Actually Changed for Healthcare, Measured

More than the sceptics say and less than the vendors say. Four findings, each dated, because the most recent measurement available isn’t always from this year.

Patients really are using AI to find providers. In its 2026 Patient Choice Report, rater8 surveyed 992 US adults in April 2026 and found 47% had used an AI tool like ChatGPT, Claude, Perplexity or Google’s AI Overviews to research a healthcare provider. Nine months earlier that figure was 31%. Reliance on AI to actually choose a provider more than doubled year on year, from 17% to 36%.

Clicks off Google are down, and not by a little. SparkToro’s analysis of Similarweb clickstream data found 68.01% of US Google searches ended without a click in the first four months of 2026, against 60.45% in 2024. SparkToro flags that its dated figures come from three different clickstream panels across the years it charts, Jumpshot then Datos then Similarweb, so read the two-year jump as a direction rather than a precise delta.

Seer Interactive tracked 53 brands, 5.47 million queries and 2.43 billion organic impressions through 2025 and into early 2026. On informational queries, a million impressions with no AI Overview returned about 33,500 clicks. With an AI Overview that cited the brand, about 20,743. With one that didn’t, about 9,445.

Read those last two numbers again, because they’re the whole argument for doing any of this. Seer puts it at 120% more organic clicks per impression when you’re cited than when you’re not. They’re also careful about what the data can’t show, and so am I: “We cannot claim causation. Higher-authority brands are also more likely to be cited.” Treat it as a gap worth closing, not a lever that pays out on contact.

Healthcare coverage is near saturation, except where it matters most to you. BrightEdge’s healthcare tracking found 89% of healthcare keywords returned an AI Overview as of December 2025, up from 59% in 2023. Treatment and procedure queries hit 100%. Pain queries 98%. Symptoms and conditions 93%.

Now the two numbers from that same study that the headline hides. Mental health and behavioural content sits at 63%, well below the healthcare average and barely moved from 58% in 2023. And local or provider-intent queries sit at 0%, down from 100% in 2023 and 14% in 2024. Google appears to have pulled AI Overviews off “therapist near me” style searches entirely.

One caveat I’d rather state than bury. BrightEdge published a second report in January 2026 whose own table puts “near me” healthcare queries at 11% for December 2025, not 0%. Same publisher, same query class, incompatible endpoints. Take the direction as solid and the exact figure as contested: AI Overviews are rare on provider-finding searches and near-universal on clinical ones.

But being cited no longer follows from ranking. In March 2026 Ahrefs looked at 863,000 keyword SERPs and 4 million AI Overview URLs and found only 37.9% of cited URLs also appeared within the first ten result blocks, which counts SERP features and not just the organic ten. In July 2025 that figure was around 76%. The overlap roughly halved in eight months.

That last one is the finding worth sitting with, and it cuts both ways. A practice sitting at position 15 for a clinical question is no longer locked out of the answer. A practice sitting at position 3 is no longer safe in it.

04What You Can Safely Ignore

Google published a section of its documentation called “Mythbusting generative AI search: what you don’t need to do.” It names AEO and GEO directly, says that “many suggested ‘hacks’ aren’t effective or supported by how Google Search actually works,” and then lists them. If a vendor is selling you one of these five for Google visibility, you now have the citation to send back.

llms.txt files and other “special” markup. Google: “You don’t need to create new machine readable files, AI text files, markup, or Markdown to appear in Google Search.” It goes further and says maintaining one “will neither harm nor help your site’s visibility or rankings in Google Search, as Google Search ignores them.” Ahrefs looked at 137,210 domains and found that of the roughly 38,000 publishing a valid file, 97% saw no requests for it at all during May 2026. Their own summary of what that means for you: “If you publish an llms.txt file today, the most likely outcome by far is that nothing ever fetches it.”

“Chunking” your content. Google: “There’s no requirement to break your content into tiny pieces for AI to better understand it.” And on the word count arms race, “There’s no ideal page length.”

Rewriting your pages for machines. Google: “You don’t need to write in a specific way just for generative AI search.” Its systems handle synonyms, so you don’t need to have captured every phrasing a person might use.

Chasing mentions. Buying or manufacturing mentions across the web “isn’t as helpful as it might seem,” because the spam systems and the ranking systems both feed the AI features.

Schema stuffing. Google: “Structured data isn’t required for generative AI search, and there’s no special schema.org markup you need to add.” Then somebody measured it. Ahrefs tracked 1,885 pages that added JSON-LD between August 2025 and March 2026 against 4,000 control pages. AI Overview citations moved by −4.6%, AI Mode by +2.4%, ChatGPT by +2.2%. Their conclusion: “Adding schema produced no major uplift in citations on any platform.” Two limits worth knowing before you throw your schema out: the study pooled schema types rather than testing them separately, and it looked at pages already picking up heavy citation volume. It tells you schema is not the thing that turns it around. It does not tell you a page nobody cites will stay uncited with it.

One correction to what this post used to say, and to a lot of therapy-website advice still circulating. FAQ schema is not the citation lever people think, and its decline started three years ago. In August 2023 Google said FAQ rich results “will only be shown for well-known, authoritative government and health websites” and that for all other sites “this rich result will no longer be shown regularly.” A solo practice is a health website but not a well-known authoritative one, so in practice that was the end of it. Google finished the job on 7 May 2026, when the rich result stopped appearing for anyone, and pulled the documentation on 15 June. The markup is still valid Schema.org, and Google’s own position on leaving it in place is that “structured data that’s not being used does not cause problems for Search, but also has no visible effects in Google Search.” So leave it where it is. Whether other retrieval systems read it is a separate question nobody has measured well. But if somebody is charging you for FAQ schema as an AEO service, that’s the conversation to have.

Keep your schema. Keep your FAQ page, because clients read it. Just stop expecting either to be the thing that gets you quoted.

05The Four Things Worth Changing

Here’s the finding everyone quotes at you, and the part of it nobody quotes. When Ahrefs ran correlations across 75,000 brands in December 2025, the strongest predictors of AI visibility weren’t on the website at all. YouTube mentions correlated at roughly 0.737. Branded web mentions came in between 0.66 and 0.71. Domain Rating managed 0.266 to 0.326, and backlinks were weaker still.

Now the part that matters more to you than any of those numbers. To get into that sample, a domain needed a Domain Rating above 40 and its highest-volume keyword pulling at least 800 searches a month. Those thresholds skew hard towards established brands. Ahrefs doesn’t report how the sample breaks down by business type, so nobody can say how many practices would have qualified, and I’m not going to guess. What I will say is that this is a real finding about established brands and it is not evidence about what works for a practice with one clinician and a name nobody searches yet. Ahrefs adds its own caution too, that correlation isn’t causation and improving these metrics won’t automatically boost anything.

I’m keeping it in anyway, because it’s the best evidence available and because the direction is worth knowing. Just check whether your own site would have met the study’s entry requirements before you let anyone sell you a mention-building retainer on the strength of it. Here are four things worth doing, and only one of them is technical.

Answer the literal question, in the first sentence under the heading. Not because AI demands a 40 to 60 word block. Google’s current guide doesn’t ask for one, and it says there’s no ideal page length and no requirement to break content into tiny pieces. Because a page that answers “how much is a therapy session in Portland” in its opening line is easier to quote, easier to skim, and better writing. The tactic survives the mythbusting. The pseudo-technical justification for it does not.

Make sure the crawlers can actually reach you. This is the one genuinely technical item on the list, and it’s the one that gets skipped. A permissive robots.txt file doesn’t prove your host or CDN is letting the bots through. If a bot can’t fetch the page, nothing else in this article matters. how to check what ChatGPT can see

Audit the reviews you already have. Reviews are public off-site information about you, sitting on platforms that can surface in AI answers, though which platform feeds which engine varies by query and nobody has mapped it properly. For a clinician this is an ethics question before it’s a marketing one, the codes differ by licence, and the rule is to follow your own board rather than a marketing checklist. what your ethics code says about asking for reviews

Be consistent about who you are and where. Same practice name, same address, same phone number, same modalities, everywhere they appear. Assistants assemble a picture of your practice from fragments across the web. Contradictory fragments can contribute to a confident, wrong answer about you.

06The Thing the AEO Pitch Leaves Out

Go back to the rater8 report, because it asked a second question that almost nobody quotes. It asked the 714 respondents who had actually searched for a new doctor in the past year what influenced their choice, letting each pick up to three. Here’s the full ranking.

Friends and family, 43%. Insurance website, 43%. Review sites, 40%. AI tools, 36%. Google search results, 34%. Doctor referral, 32%.

AI is fourth. It beat Google search, which is the genuinely startling part and the number the vendors will quote at you. It also lost to a directory your insurer maintains and to a review site, and it’s roughly level with the friend who said “she was really good.”

Read the fine print on that third row before you act on it. rater8’s “review sites” were Healthgrades, Vitals and WebMD, which are physician directories. If you’re a therapist, those directories may not map onto where your clients actually look. My own read is that Psychology Today is the closer equivalent, though that’s my judgement rather than anything rater8 measured. The survey also asked about doctors, not therapists. So treat the ordering as a shape rather than a scoreboard for your own practice.

The shape is still worth having. This next part is my recommendation rather than something the survey establishes, since it measured reported influence and not what any of this costs or how fast it moves. For a solo practice with finite hours, I’d fix the insurance directory listing and the review footprint before I’d spend a penny on AEO. They’re two of the three highest-ranked influences in the survey, and they’re the two you can actually go and change. Friends and family ties for first and isn’t something you can fix on a Tuesday.

how to build a review footprint your board actually permits

07Where This Advice Runs Out

Three limits, stated plainly.

Google’s mythbusting is about Google. Every “you can ignore this” in section 04 is scoped to Google Search. Other engines behave differently. Semrush’s June 2026 index of 126 million US prompts found ChatGPT averages about 15 sources per answer while Gemini averages 3. Those are not the same system with a different logo, and advice tuned to one won’t transfer cleanly. how ChatGPT specifically retrieves and cites

The measurement is thin, and thinnest where you most want it. Search Console’s Generative AI performance report shows you Google’s AI surfaces. ChatGPT has no first-party citation report comparable to it. You can still see ChatGPT referral traffic in your analytics, and third-party tools sample citations, but sampling is not the same as a dashboard. If someone quotes you a “ChatGPT ranking,” ask what tool produced it and what that tool actually samples.

The correlation study is a correlation study. Section 05’s numbers describe brands that are already visible. They don’t prove that starting a YouTube channel causes citations, and Ahrefs says so themselves.

None of that makes the work pointless. It means answer engine optimization is a thing you do, not a thing you buy, and it makes the retainer worth reading carefully.

Frequently asked questions

Sources