More and more people start their search for treatment in AI search instead of a list of blue links. Recent estimates put AI platform use at more than 1 billion people, and Google now puts AI Overviews and AI Mode answers above the regular results for many addiction-related searches.
For treatment centers, that raises a practical question: when someone asks an AI about your facility, or about treatment in your area, what does it say?
An AI visibility audit answers that question. In this guide, we’ll walk through how to run one, how to score the results, what to fix first, and how to track it over time.
Why AI Visibility Matters Now
For years, visibility meant ranking on the first page of Google. AI search changes that in three ways:
- Fewer clicks, more direct answers. AI tools summarize the options instead of listing them. If your facility isn’t in the answer, the searcher may never see you, even if you rank well in the organic search results below it.
- The AI speaks for you. It describes your programs, your insurance, and your reputation in its own words, and it can be wrong.
- It’s already happening. Families are asking ChatGPT and Google’s AI which rehab to call today, whether or not you’re paying attention.
For a treatment center, an inaccurate AI answer isn’t just a branding issue. A wrong level of care or insurance plan creates calls your admissions team can’t convert and frustration for families who were already struggling.
What Is an AI Visibility Audit?
A traditional SEO audit looks at how your site performs in search engine results pages. An AI visibility audit, sometimes called an AI search visibility audit, looks at how AI engines talk about you: whether they mention your facility, how accurately and favorably they describe it, and which sources they rely on.
It’s the starting point for answer engine optimization (AEO), also called generative engine optimization (GEO): the work of getting large language models (LLMs) to find, trust, and correctly cite your facility in AI-generated answers. SEO and AEO overlap heavily, but the goal shifts from ranking a page to being the brand the answer is built around.
A complete audit checks four things:
- Presence: Does your facility appear for the questions families actually ask?
- Accuracy: Are your services, levels of care, locations, and insurance described correctly?
- Sentiment: Is the tone positive, neutral, or negative, and does it reflect your real reputation?
- Sources: Which pages and third-party sites is the AI pulling from?
How to Run an AI Visibility Audit
Step 1: Build Your Prompt List
Your audit is only as good as your prompts. Organize them by intent so you cover every stage of the search:
- Location prompts: “What addiction treatment centers are in [city]?” or “Where can I find alcohol detox near [city]?”
- Service-level prompts: “Where can I get outpatient rehab for opioids in [city]?” or “Is there a dual diagnosis program near [city]?”
- Insurance prompts: “Which rehabs in [city] accept [insurance provider]?” or “What insurance does [facility] accept?”
- Comparison prompts: “What’s the best rehab in [city]?” or “[Facility] vs. [competitor]: which is better?”
- Brand prompts: “Is [facility] a good rehab?”, “What levels of care does [facility] offer?”, or “What do people say about [facility]?”
For ideas, look at the “People also ask” questions in Google for your main services, and ask your admissions team what callers ask most. Keep prompts neutral, and don’t include your facility’s name in the location or service prompts, or you’ll get an answer that doesn’t reflect what a real searcher sees.
Step 2: Test Across the Platforms That Matter
Different AI platforms often give different answers to the same prompt, because each one draws on different data:
- Google AI Overviews and AI Mode draw from Google’s own index and, for local searches, often from Google Business Profiles and reviews. For most rehab searches, this is where the biggest exposure is, because it sits on the search results page itself.
- ChatGPT is the most widely used AI chatbot. When it searches the web, it pulls current sources. When it doesn’t, it answers from training data that may be months or years old.
- Perplexity searches the web for every answer and shows its sources, which makes it useful for seeing where information comes from.
- Gemini, Claude, and Copilot are worth including if you have the time, since each uses a different mix of sources.
A few habits make your results more reliable:
- Use a clean session. Log out or use a private window so your own history doesn’t shape the answer.
- Run each prompt more than once. AI responses vary from one run to the next. Two or three runs per prompt give you a truer picture.
- Mind your location. Local prompts return different answers depending on where the search comes from. Test from, or set the location to, the market you serve.
Step 3: Score the Results
A simple scorecard makes your results easy to compare across AI platforms and from one month to the next. It can look like this:
| Prompt | AI Overview | ChatGPT | Perplexity | Accuracy | Sentiment | Top Source |
|---|---|---|---|---|---|---|
| What addiction treatment centers are in [city]? | Yes | Yes | No | Moderate | Neutral | Directory listing |
| Where can I find alcohol detox near [city]? | No | Yes | No | Low | Neutral | Third-party directory |
| Does [facility] offer residential treatment? | Yes | Yes | Yes | High | Positive | /programs/residential-treatment/ |
| What levels of care does [facility] offer? | Yes | Yes | Yes | Low | Neutral | Homepage |
| Is [facility] a good rehab? | Yes | Yes | Yes | Moderate | Mixed | Google reviews |
For sentiment, note the actual words the AI uses. “Well-reviewed,” “limited information available,” and “some patients reported concerns” send very different signals to a family deciding who to call. Add a notes column for specific problems: an outdated address, a program you closed, a misstated level of care, or a link to a page that doesn’t exist.
From the scorecard, you can calculate a simple AI visibility score: the percentage of prompts where your facility appears. Track accuracy the same way. Compare both against your main competitors for a basic competitor benchmark and share of voice.
Step 4: Identify the Gaps and Errors
Now look for patterns. Common problems for treatment centers include:
- Levels of care listed that you don’t offer, or missing ones you do
- Insurance information that’s incomplete or out of date
- An old address, phone number, or closed location still showing up
- Your facility confused with a similarly named center
- Competitors recommended for services you offer too
- Links to pages on your site that don’t exist
That last one is more common than most people expect. AI tools often generate URLs rather than copying them, and those made-up links send real visitors to 404 pages. We cover how to find and fix them in our guide to reclaiming AI search traffic from hallucinated URLs.
Step 5: Trace Errors Back to Their Sources
When an AI gets something wrong, the error usually comes from somewhere. Look at the cited sources and citation patterns, and you’ll often find it’s not your website at all.
For treatment centers, AI tools frequently lean on third-party sources such as:
- SAMHSA’s FindTreatment.gov locator
- Psychology Today and other treatment directories
- Your Google Business Profile and review sites
- Business listing sites like the BBB, Yelp, and data aggregators
- Forums and community sites like Reddit and Quora
- Local news coverage and other earned media
Not all sources carry equal weight. A stale directory listing can outweigh your own website if the AI trusts that directory more. When independent, reputable sites agree on who you are and what you offer, that third-party validation makes AI answers more likely to mention you, and to get it right.
What Influences How AI Describes Your Facility
Reviews and Off-Site Discussion
For brand and comparison prompts, AI chatbots often summarize what patients and families say about you. Detailed Google reviews, responses from your team, and genuine discussion on forums shape both whether you’re recommended and the tone of the answer. A few vague five-star reviews carry less weight than detailed ones that mention specific programs and experiences.
This is also where sentiment problems usually start. If the AI says “some patients reported concerns,” check your reviews and forum mentions for the source.
Backlinks and Authority
Backlinks don’t directly buy you an AI citation. But the pages AI tools cite tend to be the same pages that rank well in organic search, and those pages usually have strong links and authority behind them. Mentions in local news, health publications, and trusted directories matter even when they don’t include a link, because they’re part of what the AI reads about you.
Structured Data and Clear Content
Organization or MedicalOrganization schema markup spells out your name, locations, contact details, and services in a format machines can read. Add it as JSON-LD, the structured data format Google recommends. It isn’t required to show up in AI answers, but it reduces the room for confusion, especially if another facility has a similar name.
Just as important is plain, explicit content. If you only offer outpatient care, say so clearly and consistently. Vague wording is exactly what leads an AI to fill in the blanks on its own. On the technical SEO side, check that your robots.txt file or firewall isn’t blocking AI crawlers like OAI-SearchBot (ChatGPT) or PerplexityBot, or those tools will have to rely on what other sites say about you.
What to Fix First
You won’t fix everything at once, so work in this order:
- Factual inaccuracies. Wrong levels of care, insurance, addresses, or phone numbers. These cost you admissions and can create compliance risk, so they come first.
- Third-party listings. Correct the directories, profiles, and listings the AI is pulling errors from. Claim any you don’t control yet.
- Your own high-priority pages. Update your homepage, locations, programs, insurance, admissions, and FAQ pages so each one states the basics clearly: who you treat, what levels of care you offer and where, which insurance you accept, and how to start admissions.
- Structured data. Add or clean up your schema so it matches your updated pages.
- Missing citations. For prompts where competitors are cited and you aren’t, create or improve the content that answers that question.
- Broader authority. Reviews, local media coverage, and earned mentions. These take longest, so start them early and let them build.
Then re-run the same prompts to confirm the fixes took hold.
Track Your AI Visibility Over Time
A single audit is a snapshot. AI answers change as models update and new sources get indexed, so the real value comes from running the same prompts on a regular schedule. Once a month is a good rhythm for most treatment centers, and quarterly works if you’re short on time.
Keep the prompt list the same so the results are comparable, and track:
- Brand mentions: how many prompts mention your facility
- Accuracy: how many of those mentions are correct
- Sentiment: whether the tone is improving, holding, or slipping
- AI citations: which of your pages are cited, and how often
- Share of voice: how often you appear compared with competitors
- Whether errors you fixed have disappeared
Manual auditing works for a short prompt list. If you’re tracking multiple locations or dozens of prompts, AI visibility tools like Ahrefs Brand Radar, BrightLocal’s AI Visibility reports, and Keyword.com’s AI visibility tracking can automate prompt monitoring, show changes over time, and roll results up into a visibility score.
Keep your regular SEO metrics in the picture too. Review your audit alongside keyword rankings and organic traffic, and check Google Analytics for referral traffic from ChatGPT, Perplexity, and other AI search engines. If AI visibility rises while those numbers hold steady, your content strategy is working in both places.
Bottom Line
AI search is already answering questions about your facility. For treatment centers, the stakes go beyond visibility. A wrong level of care, an outdated insurance list, or a negative summary affects real admissions and real families.
An AI visibility audit shows you what those tools are saying, why they’re saying it, and what to fix. Run it once to get a baseline, fix the errors in priority order, then keep running it. The centers that monitor and correct their AI presence now will be the ones AI recommends later.