The Complete AEO Playbook for Physical Therapy

Answer Engine Optimization is the practice of structuring your online presence so that AI platforms — ChatGPT, Perplexity, Google Gemini, and others — recommend your practice when someone asks for a physical therapist. It is fundamentally different from traditional SEO. With SEO, you compete for a spot on a list of ten blue links. With AEO, you compete for one of three to five direct recommendations that AI gives as its answer. There is no page two. There is no scrolling past ads. AI picks a handful of names and presents them as the best options. If your practice is not in that handful, the patient went somewhere else before they ever knew you existed.

This shift is already well underway. Millions of people now ask AI for local provider recommendations instead of typing queries into Google. The behavior is especially common in healthcare, where people want a direct answer — not a list of websites to evaluate. When someone asks an AI assistant for the best physical therapist for knee pain in their city, the AI does not show them a search results page. It gives them names, explains why it chose those practices, and often provides contact information. That is a fundamentally different kind of competition than ranking on Google, and it requires a fundamentally different approach.

For physical therapy specifically, this matters because most PT practices rely heavily on physician referrals and word-of-mouth. Those channels still work, but they are no longer the only way patients find providers. Patients with direct access — which is now available in all 50 states for evaluation and in most states for treatment — are increasingly searching on their own. When those patients ask AI, your practice needs to be positioned to show up. The practices that figure this out first will capture patient volume that competitors do not even realize they are losing.

We studied exactly this question across 21 U.S. cities and three major AI platforms. The findings were clear and, for many practice owners, surprising. The single most important factor in whether AI recommends a physical therapy practice is how the practice's website is structured. Not how many backlinks it has. Not how old the domain is. Not whether the practice is affiliated with a hospital system. Website structure — how information is organized, labeled, and presented — is the strongest predictor of AI citations by a wide margin.

The data is striking. Domain authority, the metric that has driven SEO strategy for over a decade, showed a correlation of r = 0.002 with AI citations. That is essentially zero. A brand-new practice with a well-structured website can outperform an established clinic with thousands of backlinks. Meanwhile, 75.7% of all AI citations in our physical therapy study went to independent practices rather than chains or hospital-affiliated clinics. This is not because AI has a preference for independent practices — it is because independent practices are more likely to have websites with detailed, specific content about the conditions they treat and the approaches they use.

Perhaps the most important finding for practice owners to understand is the low cross-model agreement. ChatGPT, Perplexity, and Gemini each pull from different data sources, weight different signals, and arrive at different recommendations. A practice that gets recommended by ChatGPT may be completely absent from Perplexity's suggestions, and vice versa. This means that optimizing for one AI platform captures only a fraction of the opportunity. A complete AEO strategy accounts for all three major platforms and the different signals each one prioritizes.

When an AI model evaluates your website, it is not reading it the way a human does. It is parsing structure. It looks at your heading hierarchy to understand what your practice does and how your services are organized. It examines your navigation to determine the scope of your offerings. It reads your page titles and meta descriptions to classify your content. If your website has a single page called "Services" with a bulleted list of everything you do, AI has very little structured information to work with. It cannot confidently recommend you for knee pain rehabilitation if that phrase only appears once in a bullet point buried on a general services page.

The website structure that performs best in our research follows a clear pattern. The homepage clearly identifies the practice, its location, and its primary focus areas. The main navigation includes direct links to individual condition or service pages — not a dropdown that leads to a single services overview. Each condition page has its own URL, its own title tag, and its own content. The heading hierarchy on each page is clean: one H1 that names the condition or service, H2s that break down key aspects like symptoms, treatment approaches, and what to expect. This structure gives AI exactly what it needs to match your practice to a patient's question.

Think of your website as a filing cabinet that AI needs to navigate. If everything is stuffed into one drawer with a label that says "Physical Therapy," the AI cannot find what it needs. If every drawer is clearly labeled — knee pain, shoulder rehabilitation, post-surgical recovery, sports injuries, balance and fall prevention — the AI knows exactly where to look and can confidently recommend you for the specific query a patient asked. The practices that appear in AI recommendations consistently have this kind of clear organizational structure.

One critical detail many practices miss: your website's internal linking matters. When your knee pain page links to your post-surgical rehab page with descriptive anchor text, you are helping AI understand the relationships between your services. When your homepage links directly to each condition page, you are telling AI that these are important, distinct offerings. A well-linked website is easier for AI to crawl, understand, and ultimately cite.

Dedicated condition pages are not optional for AI visibility. Our research consistently shows that practices with individual pages for specific conditions receive significantly more AI citations than practices that lump everything onto a single services page. The reason is straightforward: when someone asks AI about physical therapy for a torn ACL, the AI looks for pages specifically about ACL rehabilitation. A dedicated page titled "ACL Reconstruction Rehabilitation" with detailed content about the recovery process, treatment timeline, and expected outcomes gives AI exactly the kind of focused, authoritative information it needs to make a recommendation.

For a physical therapy practice, the core condition pages should cover the problems you treat most frequently and the ones patients are most likely to search for. At minimum, most PT practices should have dedicated pages for: knee pain and knee injuries, shoulder pain and rotator cuff issues, back pain and spinal conditions, neck pain, post-surgical rehabilitation, sports injuries and sports medicine, balance and fall prevention, and any specialty areas like TMJ disorders, vestibular rehabilitation, pelvic floor therapy, or pediatric physical therapy. Each of these pages should be a substantive resource — not a paragraph with a stock photo, but a thorough explanation of the condition, how your practice approaches treatment, what patients can expect, and why your team is qualified to help.

The anatomy of an effective condition page follows a consistent pattern. Start with a clear H1 that names the condition. Follow with a brief overview that describes the condition and who it affects. Then cover your treatment approach — not in vague terms, but with enough specificity that a reader (or an AI) understands what you actually do. Include information about what a first visit looks like, how long treatment typically takes, and what outcomes patients can expect. If your therapists have relevant specializations or certifications, mention them on the page. If you use specific techniques or technologies, describe them. The goal is to create the most useful, complete page about that condition in your local market.

A common mistake is creating condition pages that are too thin — a few sentences and a call to action. These pages do not give AI enough information to work with. They also do not help patients, which means they do not help you. On the other hand, you do not need to write a medical textbook. Three to five well-written paragraphs that cover the condition, your approach, and what to expect will outperform both a thin page and a 5,000-word article that buries the useful information. Be specific, be clear, and write for the person who has this condition and wants to know if you can help them.

Having the right pages is necessary but not sufficient. The content on those pages needs to be structured in a way that AI can easily extract and reference. AI models work by finding passages that directly answer questions, so your content needs to contain clear, self-contained answer blocks. An answer block is a paragraph or short section that directly responds to a question someone might ask. For example, if someone asks "How long does physical therapy for a rotator cuff tear take?" your shoulder page should contain a section with a clear heading and a direct, specific answer — something like "Rotator cuff rehabilitation typically takes 12 to 16 weeks, beginning with gentle range-of-motion exercises and progressing to strengthening as healing allows." That kind of direct, factual statement is exactly what AI pulls into its responses.

Question-format headings are one of the most effective content structures for AI visibility. Instead of a heading that says "Treatment Duration," use "How long does physical therapy for shoulder pain take?" This mirrors the way people actually ask questions to AI, and it signals to the AI model that the content below is a direct answer to that specific query. Throughout your condition pages, use headings that match real patient questions: "What causes knee pain after running?" "Do I need a referral for physical therapy?" "What should I wear to my first PT appointment?" Each question-heading followed by a clear, concise answer creates a building block that AI can extract and cite.

Patient education content is another powerful tool. When you explain common conditions in accessible language, describe how specific treatments work, or provide guidance on what patients should know before starting therapy, you are creating content that AI can reference as authoritative. This is not about keyword stuffing or writing content purely for search engines. It is about being genuinely helpful in writing — the same way you are genuinely helpful in person. Describe your treatment approaches honestly. Explain what makes your practice different without marketing hyperbole. Provide the kind of clear, practical information that a patient would want to know. AI models are trained to identify and recommend this kind of substantive, useful content over generic marketing copy.

AI models do not rely solely on your website when making recommendations. They pull information from across the internet — Google Business Profile, healthcare directories, review platforms, social media, and other sources. When your practice name, address, and phone number are consistent across all of these sources, AI has higher confidence in its information about you and is more likely to recommend you. When there are discrepancies — a different phone number on Yelp than on your website, an old address on a directory listing, a practice name that varies between "Smith Physical Therapy" and "Smith PT & Wellness" — AI has less confidence and may skip you entirely in favor of a practice with cleaner data.

Your Google Business Profile is the single most important off-site asset for AI visibility. Make sure it is complete, accurate, and active. Your primary and secondary categories should be set correctly (Physical Therapy Clinic as primary, with relevant secondary categories like Sports Medicine Clinic or Rehabilitation Center). Your services should be listed individually, not as a generic description. Your hours, phone number, and address must match what is on your website exactly. Post updates regularly — Google Business Profile posts signal that the practice is active and engaged. Respond to reviews, because review content and responses are another data source AI models reference.

For physical therapy specifically, there are several directories that matter. Healthgrades, Vitals, WebMD, Zocdoc (if applicable), Psychology Today (for practices offering pelvic floor or pain psychology services), and your state PT association directory all feed into the information ecosystem AI draws from. Your APTA profile, if you are a member, should be complete. Insurance directory listings should be accurate. The goal is not to be on every directory — it is to ensure that every place your practice appears online has the same, correct information. Consistency across these sources builds the kind of data confidence that leads to AI citations.

Structured data markup — the code that explicitly tells search engines and AI what your content represents — is an important layer of your AEO strategy. For a physical therapy practice, the most relevant schema types are LocalBusiness (or the more specific PhysicalTherapy subtype), MedicalBusiness, and Service. These structured data types allow you to explicitly declare your practice name, location, hours, accepted insurance, areas of specialization, and the conditions you treat. When AI encounters this structured data, it does not have to guess what your practice does — the information is declared in a format specifically designed for machine readability. Practices with proper structured data give AI a reliable data source that is faster and easier to parse than unstructured page content alone.

AI crawler access is a technical detail that many practices overlook entirely. AI platforms like ChatGPT (via GPTBot), Perplexity (via PerplexityBot), and Google Gemini use web crawlers to discover and index content. If your website's robots.txt file blocks these crawlers — which some website platforms do by default — AI literally cannot see your content. Check your robots.txt file and make sure GPTBot, PerplexityBot, and Google's crawlers are not blocked. Additionally, verify that your sitemap.xml is up to date and includes all of your condition and service pages. A clean sitemap helps AI crawlers find and index your content efficiently. This is a five-minute fix that can make the difference between being invisible to AI and being discoverable.

Page speed and mobile responsiveness also play a role, though they are more of a baseline requirement than a differentiator. AI crawlers, like search engine crawlers, have limited time to spend on any given website. If your pages load slowly, the crawler may not fully index your content before moving on. If your site is not mobile-responsive, some of your content may be inaccessible or poorly formatted for crawlers that evaluate the mobile version of your site. Most modern website platforms handle these basics reasonably well, but it is worth verifying. Run your site through Google PageSpeed Insights and address any critical issues. Ensure that your condition pages load quickly and display properly on mobile devices. These technical fundamentals support everything else in your AEO strategy.

You cannot improve what you do not measure, and AI visibility is no exception. The challenge is that tracking AI recommendations is fundamentally different from tracking Google rankings. There is no equivalent of checking your position for a keyword. AI responses vary based on how the question is phrased, the user's location, the time of day, and factors that are not publicly documented. What you can do is systematically query each AI platform with the kinds of questions your prospective patients ask and track whether your practice appears in the responses over time. This means running regular queries across ChatGPT, Perplexity, and Gemini for questions like "best physical therapist for knee pain in [your city]" and "who should I see for post-surgical rehab in [your area]" and recording the results.

Our research showed that cross-model agreement is very low — different AI platforms recommend different practices for the same query. This has a practical implication for monitoring: you cannot test on one platform and assume the others behave the same way. A practice that ChatGPT recommends consistently may be completely absent from Perplexity's answers. Monitoring needs to cover all three major platforms, with a variety of query types that reflect how patients actually ask for physical therapy. Track not just whether you appear, but which competitors appear, what the AI says about them, and what sources the AI cites. This competitive intelligence helps you understand what is working in your market and where the gaps are.

Establish a baseline before you make any changes. Run your monitoring queries before implementing any AEO optimizations, then track the same queries weekly or biweekly as you roll out changes. This allows you to see the direct impact of each optimization. Over time, you should see your practice appearing more frequently across more query types and more platforms. If a specific AI platform is not picking you up, that data tells you where to focus your next round of improvements. Monitoring is not a one-time activity — it is an ongoing practice that keeps your AEO strategy responsive to how AI platforms evolve.

The most common mistake we see is relying entirely on physician referrals and ignoring the patient discovery channel altogether. Direct access laws mean that patients can see a physical therapist without a referral in every state, and a growing number of patients are exercising that right. When those patients ask AI for a recommendation, practices that never invested in their online presence are invisible. A related mistake is assuming that hospital or health system affiliation will carry the day. Our research shows that AI does not give preferential treatment to hospital-affiliated practices. In fact, hospital system websites often work against AI visibility because they tend to use complex, multi-layered site architectures that make it harder for AI to find and extract information about individual providers and locations.

Another critical error is treating AI optimization like traditional SEO. Practices that pour resources into building backlinks, chasing domain authority, and optimizing for Google's ranking algorithm are investing in signals that have almost no effect on AI recommendations. Our data shows a correlation of r = 0.002 between domain authority and AI citations — that is statistically indistinguishable from zero. Similarly, practices that create a single "Services" page with a list of everything they offer are making themselves nearly invisible to AI. That format gives AI no way to match the practice to a specific patient query. A patient asking about knee pain treatment gets matched to practices with knee pain pages, not practices that mention knee pain in a bullet list.

A subtler mistake is optimizing for only one AI platform. Some practices test their visibility on ChatGPT, see that they appear, and assume they are covered. But our research shows that the overlap between AI platforms is remarkably low. Being visible on ChatGPT does not mean Perplexity or Gemini will recommend you. Each platform weighs different signals and draws from different data sources. Practices also frequently underestimate how much content their condition pages need. A page with two sentences and a contact form does not give AI enough information to cite. Finally, many practices never check whether AI crawlers can actually access their website. A single line in a robots.txt file can make your entire site invisible to AI — and you would never know unless you checked.

A practical AEO implementation for a physical therapy practice follows a phased approach over roughly 90 days. In the first two weeks, the focus is on auditing and planning. This means running a complete AI visibility audit — querying all three major AI platforms with relevant queries for your market and recording the baseline. It also means auditing your current website structure, identifying which condition pages exist and which need to be created, checking your technical setup (robots.txt, structured data, page speed), and reviewing your off-site presence for consistency. The output of this phase is a prioritized plan that tells you exactly what to build and in what order.

Weeks three through eight are the build phase. This is where you restructure your website navigation if needed, create or rewrite your condition pages, implement structured data markup, fix any technical issues identified in the audit, and clean up your directory listings and Google Business Profile. The condition pages are the most time-intensive part of this phase. Each page needs substantive, well-structured content that demonstrates real expertise. Depending on how many condition pages you need, this phase may involve creating eight to fifteen individual pages, each with unique content that addresses the specific condition, your treatment approach, and what patients should expect. The build phase also includes implementing proper internal linking between your pages and ensuring your heading hierarchy is clean across the entire site.

Weeks nine through twelve are about establishing your monitoring baseline and making refinements. Run the same AI queries from your initial audit and compare the results. You should see measurable improvements — new appearances in AI recommendations, more detailed citations, visibility on platforms where you were previously absent. Use this data to identify which areas need further work. Maybe your knee pain page is getting cited but your shoulder page is not — that tells you where to invest next. This phase also involves setting up an ongoing monitoring cadence so you can track progress over time and respond quickly if AI platforms change how they weight different signals. After the initial 90 days, AEO becomes an ongoing practice of monitoring, refining, and expanding your content as patient needs and AI platforms evolve.

AEO is not an overnight transformation. Most practices begin seeing changes in AI recommendations within four to eight weeks of implementing structural and content changes. The timeline depends on how quickly AI platforms re-crawl your site, how competitive your local market is, and how substantial the changes are. Practices in smaller markets or markets where no competitors have optimized for AI tend to see results faster. In competitive metro areas, it may take longer to break through, but the payoff is proportionally larger because more patients in those markets are using AI to find providers.

The results compound over time. Once AI starts recommending your practice, it creates a reinforcing cycle: more visibility leads to more patient inquiries, which leads to more reviews, which strengthens your online presence, which leads to more AI citations. Practices that invest early in AEO build an advantage that is genuinely difficult for competitors to replicate quickly. A competitor who decides to catch up six months from now will be starting from scratch while you have six months of compounding visibility. This first-mover advantage is one of the most compelling reasons to act now rather than waiting. The window where most PT practices are unaware of AI search is closing, and the practices that move during this window will be the ones AI has already learned to recommend by the time competitors start paying attention.