We audited 30 treatment center websites. 16 showed the phone number SAMHSA lists.
PELORA audited 30 treatment center websites on October 24, 2026 for the signals AI search engines can read. All 30 served their main text without scripts, and none of the 30 robots.txt files (the file that tells crawlers what they may read) blocked an AI search crawler. The gaps sit in consistency and markup: 16 of 30 showed the phone number SAMHSA's locator lists, and 10 of 30 labeled an FAQ for search engines.
Source for every figure on this page: PELORA's audit of 30 treatment center websites, read October 24, 2026, Pacific time. Each figure shows the sites that passed and the sites we could check.
Last updated October 26, 2026. What we checked: 30 treatment center homepages, their robots.txt and llms.txt files, and the FAQ page and one blog article each homepage linked, all read October 24, 2026. By Preston Durnford, Founder and CEO, PELORA Marketing.
What did we find on 30 treatment center websites?
The 30 center websites we read were easy for AI crawlers to reach and read, and 14 of them show a different phone number than the federal locator. Either listing could be the out-of-date one. All 30 sites served their main text in the page itself, with a median of about 1,140 words on the homepage. That matters because some AI fetchers do not run scripts. Anthropic's documentation for Claude says: "The web fetch tool currently does not support websites dynamically rendered with JavaScript."
Most of these sites describe themselves in markup: 25 of 30 carry business schema (labels in the page code that tell machines who you are). The gaps are narrower. 14 of 30 show an FAQ, and 10 of 30 mark one up with FAQPage schema.
Can AI crawlers reach treatment center websites?
Yes: none of the 30 robots.txt files blocks an AI search crawler. We tested each file against Googlebot, Bingbot, OAI-SearchBot, Claude-SearchBot and PerplexityBot. Perplexity describes its crawler this way: "PerplexityBot is designed to surface and link websites in search results on Perplexity."
Few centers have taken a side. 3 of 30 files name any AI crawler at all, and 1 of 30 blocks a training crawler such as GPTBot or Google-Extended. Next, read how a clinic gets named in ChatGPT.
robots.txt is only half of access. Google's AI features guide lists "Ensuring that crawling is allowed in robots.txt, and by any CDN or hosting infrastructure" as a step. We saw 1 center site refuse our ordinary request, and 4 more we could not read at all; we did not count those as blocking AI crawlers, because we cannot tell how they treat a verified crawler.
12 of 30 sites publish an llms.txt file, a plain summary written for AI tools. Google says Search ignores these files: "Doing so will neither harm nor help your site's visibility or rankings in Google Search, as Google Search ignores them." We report the count and leave it out of the score. Our page on generative engine optimization covers where the effort pays instead.
Does the website's phone number match the federal locator?
About half the time: 16 of 30 homepages show the phone number SAMHSA's locator lists for the center. We counted a match when either the main or the intake number in the locator appeared on the homepage. The rest show a different number.
13 of 30 sites put a phone number in business schema, and 12 of those 13 match a number shown on the page. 22 of 30 homepages show both a phone number and a street address, or a state and ZIP code.
A caller may compare the locator, the website and the Business Profile. If you use a tracking number, set up with your counsel, make sure the locator, the website and your Business Profile all reach the same front desk. PELORA's study of the SAMHSA locator gap found a related problem: most sampled licensed California centers could not be confidently matched on the locator.
Do treatment center websites tell AI engines who they are?
Most do, in markup. 27 of 30 homepages carry JSON-LD structured data, 25 of 30 describe the center as an organization or local business, and 13 of 30 use a medical type such as MedicalClinic or MedicalOrganization.
"Structured data isn't required for generative AI search, and there's no special schema.org markup you need to add."
Google Search Central, Google's guide to optimizing for generative AI features, read October 24, 2026We treat markup as hygiene, not a ticket into AI answers: one clear statement of the center's name, phone and address that matches the page and the locator.
Do treatment center websites show their certifications?
Some do: 11 of 30 homepages reference LegitScript somewhere in their code, such as a seal or a link, and 7 of 30 name The Joint Commission or CARF in their visible text. A seal shown only as an image does not count toward the accreditation figure, so a missing mention does not mean a center lacks certification. LegitScript's own records are the place to confirm it.
Certification matters most for ads. Google's healthcare policy says: "In order to advertise for recovery-oriented drug and alcohol addiction services, you must be certified as an addiction services provider by the LegitScript Certification program." It adds: "Advertisers must also be certified with Google." (Read October 24, 2026.) Our guide to how LegitScript certification works and our page on ad rules for treatment centers cover the details. This is not legal advice; ask your health care attorney about your program.
Do treatment center websites answer questions and sign their articles?
Most answer early, and most articles are dated and signed. 25 of 30 homepages open with a full paragraph of 20 or more words inside the first 100 words of main text. 14 of 30 use a question as a section heading.
16 of 30 homepages link a blog or news section, and we reached an article on 16 of those 16. 15 of 16 articles show both a date and an author. Only 5 of 16 carry a "reviewed by" line, which matters on pages about withdrawal, medication and care.
How did the 30 sites score out of 10?
The median treatment center site passed 7 of the 10 checks. Scores ran from 4 to 10. 2 of 30 sites passed all 10, 11 of 30 passed 8 or more, and 7 of 30 passed 5 or fewer.
| Check | What we looked for | Sites that passed |
|---|---|---|
| Crawler access | robots.txt lets five search crawlers read the homepage | 30 of 30 |
| Text without scripts | 150 or more words in the homepage HTML as served | 30 of 30 |
| Answer first | A 20-word paragraph within the first 100 words | 25 of 30 |
| Business schema | Markup naming an organization, local business or medical type | 25 of 30 |
| Phone and address shown | A phone number and a street address, or a state and ZIP code | 22 of 30 |
| Phone in schema | A schema phone number that matches one on the page | 12 of 30 |
| FAQ shown | An FAQ section on the homepage, or a linked FAQ page | 14 of 30 |
| FAQ marked up | FAQPage schema on the homepage, FAQ page or article | 10 of 30 |
| Dated, signed article | A linked article showing a date and an author | 15 of 30 |
| Fast server response | Server starts sending the homepage in 0.8 seconds or less (our test) | 23 of 30 |
What should a treatment center fix first?
Start with the facts people use to reach you.
- 1Match your phone number everywhereCheck that your website, your locator listing and your Business Profile show the same main number.
- 2Mark up the questions you already answer6 of the 14 sites with an FAQ have no FAQPage markup. The answers exist; the label is missing.
- 3Put a clinical review line on clinical articlesName the clinician who reviewed the page, with a real date. Never invent a reviewer.
- 4Say your certifications in textIf the center holds LegitScript certification or accreditation, say so in words as well as in the seal.
- 5Choose your AI crawler rules on purposeRead your robots.txt file, then ask your host whether bot protection blocks crawlers the file allows.
None of this makes an AI engine recommend a center. It removes reasons for an engine, or a family, to pass a center by. PELORA offers AI search optimization and treatment center marketing to centers that want help with it.
How did we run this study?
- •
Sampling frame. SAMHSA's FindTreatment.gov locator, substance use facilities within about 25 miles of the center of 10 large US metro areas: Seattle, Phoenix, Denver, Dallas, Kansas City, Minneapolis, Nashville, Atlanta, Columbus and Boston.
- •
Who was eligible. Privately operated facilities, for profit or nonprofit, that offer a residential setting, are not listed as a hospital, and list a website on their own domain. One entry per website. The locator returned 1,043 substance use listings across the 10 areas, and 139 websites met these rules.
- •
The draw. In each metro we put the eligible websites in a random order fixed in advance, and measured them in that order until 3 per metro were read.
- •
Reading the sites. One plain request per page with an ordinary browser user agent: the page the locator lists as the website, robots.txt, llms.txt, a linked FAQ page and one linked blog article. No sign-ins, no forms, no retries.
- •
Sites we could not read. 1 site refused our request, and 4 could not be read (a server error, a redirect loop, a TLS error and a proxy error). We recorded each one, did not retry it, and moved to the next website in that metro's order. We tried 35 sites to measure 30.
- •
Exclusions. We removed centers on PELORA's do-not-contact list, businesses tied to PELORA's founder and PELORA's clients. 0 listings in the frame and 0 sites in the draw needed to be dropped for that reason.
- •
Privacy. This page reports totals only and names no center. The site list stays in our audit file. We did not open admissions forms, chat windows or insurance checks, and we collected no patient information.
- •
Limits. This study measures signals on the sites, not answers from ChatGPT, Perplexity, Gemini or Google's AI features, and we queried no engine. Each check is a script reading of public HTML on one day. Call tracking can swap phone numbers, and our server response times come from a cloud server, so they are lab readings. On that point, web.dev says: "Good TTFB values are 0.8 seconds or less."
We did not ask any AI engine about these centers, judge their care, check their licenses or read their reviews. A missing signal is a fix, not a verdict on a center.
Why did PELORA run this study?
PELORA is a healthcare-only marketing agency. Its founder, Preston Durnford, built and ran Epic Journey Recovery, a behavioral health company. We serve clients nationwide, and we would rather publish the measurement than assert it.
What do treatment center owners ask most about AI visibility?
What does AI visibility mean for a treatment center website?
It means an AI search engine can reach the site, read its main text and tell who the center is, where it is, how to call it and which questions it answers. This study checks those signals on the sites. It does not show whether any engine names a center, which changes with the question, the place and the day.
Did this study test ChatGPT, Perplexity or Google's AI answers?
No. We did not query any AI engine. We measured signals that help an engine quote a site, on the center websites themselves, from public pages read on October 24, 2026. Engine answers vary from one run to the next, so one round of questions would not hold up as evidence.
Why does the phone number on the federal locator matter?
When SAMHSA's locator, the website and a Business Profile show different numbers, a caller sees two numbers for one center and has to guess. 16 of 30 centers in this study showed the locator's number on the homepage.
Does a LegitScript mention on the website help in AI search?
None of the AI engine documentation we read says so. LegitScript certification matters for ads: Google requires LegitScript certification, plus certification with Google, to advertise recovery-oriented drug and alcohol addiction services in the US. On the website, a plain-text mention lets a reader confirm it. This is not legal advice; ask your health care attorney about your program.
Should a treatment center block AI crawlers?
That is the owner's choice. Search crawlers such as OAI-SearchBot and PerplexityBot fetch pages to answer questions; training crawlers such as GPTBot collect text for models. None of the 30 robots.txt files we read blocked a search crawler.
Did you look at any patient information?
No. We read public pages only. We did not open admissions forms, chat windows or insurance checks, and we collected nothing about any patient.
Want these checks on your site?
Book a call, and we'll send a free one-page outside look at your website first. It reads public pages only and needs no logins.
Sources
- PELORA, AI citability audit of 30 treatment center websites, public pages read October 24, 2026. Aggregate results only; the site list is PELORA's audit record.
- SAMHSA, FindTreatment.gov locator, findtreatment.gov, queried October 24, 2026.
- Anthropic, "Web fetch tool", Claude Platform Docs, platform.claude.com, read October 24, 2026.
- Perplexity, "Perplexity Crawlers", docs.perplexity.ai, read October 24, 2026.
- Google Search Central, "AI Features and Your Website", developers.google.com, read October 24, 2026.
- Google Search Central, "Google's Guide to Optimizing for Generative AI Features on Google Search", developers.google.com, read October 24, 2026.
- Google Ads policy, "Healthcare and medicines", support.google.com, read October 24, 2026.
- web.dev, "Time to First Byte (TTFB)", web.dev, read October 24, 2026.