Grow My Therapy
Grow My TherapyGrow My Therapy — AI visibility
People now describe what they’re going through to an AI assistant and ask who can help. An assistant can only suggest a practice it can reach, read and understand. Making yours that practice is built into how we make a website, and our system checks the result before launch.
One homepage, as an AI crawler receives it
<title>Practice name</title>
<body>
<div class="loader">
<div class="progress"></div>
</div>
</body><title>Practice name</title>
<script type="application/ld+json">
{ "@type": "MedicalBusiness", … }
</script>
<h1>Therapy for women in …</h1>
<p>Every word of the page,
right here in the file.</p>Measured on one real therapist website, November 2026. Code simplified for reading.
How AI answers get made
When someone asks for a trauma therapist in their city, the assistant usually doesn’t answer from memory. It searches the web, reads a handful of pages, and writes its answer from what it finds. So there are several separate systems to be found by, and they don’t share a front door.
Google says these features are rooted in its core Search ranking systems and draw from its Search index. A page has to be indexed and eligible for a snippet first, so regular SEO still decides most of this. Google also names the Business Profile as a source for local details.
ChatGPT’s search features rely on OpenAI’s own crawler, OAI-SearchBot. OpenAI says sites opted out of it will not be shown in ChatGPT search answers, apart from navigational links. So a site can rank well on Google and still be shut out here by one line of configuration.
Microsoft’s answers come through Bing, which has its own index, its own tools and its own report of how often a site is cited. Claude and Perplexity run their own crawlers too. Each of these is a separate door, and each one has to be open.
The problem most checklists miss
Your browser runs a page’s JavaScript before showing it to you. In Vercel’s analysis of crawler traffic, the crawlers from OpenAI, Anthropic and Perplexity did not. They read the file the server sent and moved on. Google’s crawler can run JavaScript, so a site can rank well on Google and be nearly blank to the others.
We went looking for this on real therapist websites in November 2026, and we found it. One homepage that looked finished in any browser reached a crawler as a loading animation and 91 characters of text. A few small code changes brought it to 7,706 characters, with its heading and structured data in place.
A second later the page is there, so you, your developer and your clients all see a working website.
The tool that builds the site can report a page as ready-made even when what it made is the loading screen.
Popular structured data testers run JavaScript, so they can pass a page a crawler reads as empty.
On a computer, open your homepage, right-click and choose View Page Source (not Inspect). Then search that window for a sentence from your page. If it’s there, crawlers can read it. If it isn’t, they probably can’t.
What we actually do
None of this is an add-on we switch on at the end. These are steps in the checklist every new website goes through, alongside the design and the writing. See the full method.
We confirm that the crawlers behind search and AI answers can reach the site, in the site’s rules and on the server itself, since a firewall can refuse a crawler that the rules allow.
Every page is built ahead of time, so the first response from the server already holds its words, its heading and its structured data. Only the interactive parts, like forms and sliders, wait for JavaScript.
Structured data tells a machine, in its own format, who this practice is. Ours is written from your sources, and it connects the business, the therapist and the services into one picture.
Length is the easy part. We check every page in the sitemap against your real name, city and practice first, because the common failure is a tidy title with the wrong town in it.
Assistants lift individual passages and read them next to a dozen other sources. So we test each answer with its question covered: if you can’t tell what it’s about, it gets rewritten.
A person arriving on a service page wants to know within seconds whether it’s for them. Giving them that also gives an assistant a passage it can use accurately.
For anything location-based, assistants lean on the same business details that search engines do. Your Google Business Profile, and where it agrees with your site, matters as much as anything on this page.
Search engines and assistants need a reliable map of the site, and a clear statement on each page that it is the original.
Google says unique, useful content will likely matter more than any technical step here. Nothing above makes a thin site worth quoting, so the writing gets the same care.
Before launch
We learned this the expensive way. An About page once told Google it was a copy of the homepage for five months after launch, and a screenshot of that same task had been reviewed and approved.
So when a new website moves to review, and again when it goes live, our system fetches the deployed site the way a crawler does. If it finds any of these, the project stops, and the message names the page and what to fix.
Knowing what to ignore
AI search is new enough that a lot gets sold on very little evidence. Google has published a list of what you can ignore, and we’ve read it closely. Here’s what we leave out, and why.
No
It’s a second copy to keep in sync, and your visitors never see it. We fix the real page, so people and machines read the same thing.
Not by default
Google stopped showing FAQ rich results on May 7, 2026, and says there’s no special markup for its AI features. The written answer is what gets quoted, so that’s where we put the effort.
Added, low priority
Google says its Search doesn’t use these files. Others may read them, and generating one from the site’s data takes minutes, so we include it without pretending it’s the work.
No
Google lists both as things you don’t need to do. We write for the person reading. A clear opening helps them first and happens to help a machine.
No
Google treats doing this to manipulate its results as scaled content abuse. Each location or service page has to earn its place, and each location plan has a cap.
Never
Google says inauthentic mentions aren’t as helpful as they seem, and fake reviews are a real risk for a licensed clinician. We help you ask real clients, within your board’s rules.
If someone has offered you a plugin
These are real tools doing mostly sensible work, usually for a monthly fee on a WordPress site. We don’t install one. We do that work in the site itself and check the result. Here’s what a typical plugin offers, next to how we handle it, including the parts we don’t do.
Removes what stops AI bots reading your site
Fixed in the site’s own code, then proven on the deployed pages. A project can’t launch until it passes.
FAQs and short summaries
Written from your own answers to the questions clients really ask, to a stricter standard, on the page your visitors see.
Structured data (JSON-LD)
Business, therapist and services connected as one identity, checked against your real details.
Meta titles and descriptions
Checked page by page against your name, city and practice, then tuned for length.
A crawler access report
The same check, run automatically as a condition of launch.
llms.txt and IndexNow
Both included, in the proportion they deserve.
Next steps: listings, reviews, mentions
Business Profile, a review process, and profiles linked from your structured data. No bought mentions.
Visits sent by AI tools
A dedicated “AI Assistants” channel in your Google Analytics.
Times AI cites you
Google’s and Bing’s own reports, where those platforms publish the data.
A dashboard of AI bot visits
We don’t provide this one. It tells you a bot came by, not that anyone was sent to you. We can read those visits from hosting logs when there’s a question to answer.
A separate AI-ready copy of each page
We don’t do this on purpose. See above.
How you’ll know
Search gives you positions and clicks. AI answers give you far less, so we use the first-party numbers that do exist and we’re straight about the rest.
A dedicated “AI Assistants” channel counts visitors arriving from ChatGPT, Perplexity, Claude, Gemini, Copilot and others, instead of burying them among ordinary referrals.
Its AI Performance report shows how often a site is cited across Copilot and Bing’s AI summaries, which pages are cited, and sample phrases used to find them.
Google’s Generative AI performance report shows how content performs in its AI features, alongside the regular Search data we already connect for you.
What nobody can see is every conversation. Someone can read your name in ChatGPT and call you the next day, and no report will show it. It’s worth asking new clients how they found you, and the answer is increasingly worth writing down.
Straight answers
This field attracts a lot of overclaiming. Here’s where we stop, so nothing on this page has to be walked back on a call.
Questions
AI visibility for a therapy practice means ChatGPT, Claude, Perplexity, Copilot and Google’s AI features can reach the website, read it, understand who the therapist is and what they treat, and quote it accurately. It rests on regular SEO, plus crawler access, pages readable without JavaScript, and answers that make sense on their own.
Grow My Therapy does not install an AI SEO plugin. A plugin adds things on top of an existing WordPress site. We do that work directly in the site as we build it: rendering, structured data, FAQs, crawler access and measurement. Then our system reads the deployed site to confirm it worked before launch.
AI visibility does not replace regular SEO. Google says its AI features are rooted in its core Search ranking systems, so a practice that is hard to find on Google usually won’t surface in its AI answers either. The AI-specific work sits on top: letting other companies’ crawlers in, and writing passages that can be quoted.
No one can guarantee that ChatGPT or any other assistant will recommend a practice, because each company’s system decides what to say and answers vary between people. What we can do is remove every technical reason you’d be left out, and give the assistant clear, accurate material to work with.
There is no fixed timeline for showing up in AI answers. Crawler access and page rendering are in place at launch, and crawlers can then read the site. Being chosen as a source tends to build over months, as search rankings do, and it depends on your specialty, your area and how many other practices compete.
Most AI visibility work is invisible to visitors, including rendering, structured data, crawler rules and reporting. The visible part is the writing: a clear opening on service pages, headings phrased as real questions, and FAQ answers that give straight information about fees and insurance. Those read better for people too.
Whether AI companies train on your content is a separate decision from whether their assistants can find you. OpenAI, for example, uses different crawlers for each, so a site can allow search and refuse training. Blocking training crawlers doesn’t remove you from search answers, so we can set this however you prefer.
A website that is already live can be read the same way a new one is. The checks that run before launch can be pointed at any site, which shows what crawlers receive today and what it would take to fix. To try a quick version yourself, open View Page Source and search for a sentence from your homepage.
SEO by specialty
The same service, written for how your license actually practices.