You don't need a paid platform to know whether AI is naming your practice. Ask the engines what your patients ask, write down the answers, and you'll see exactly where you sit. This takes ninety minutes and costs nothing.
An AI citation audit is a prompt log. You gather the questions patients actually ask about insurance, hours, conditions and neighborhoods, and run them through ChatGPT, Perplexity, Google AI Overviews, and Claude. You record which engines named you, which named competitors, and what sources they cited. That record is your audit.
The Questions Your Patients Ask
Start by listing the prompts patients use when they search for you. Not the keywords a marketer would write. The actual questions people say aloud or type into a chat box.
For an optometrist: "eye doctor near me that takes Medicaid," "best optometrist in Crown Heights Brooklyn," "optometrist open Saturday," "do I need an exam to get contacts," "how much does an eye exam cost."
For a physical therapist: "physical therapy without a referral in New York," "PT for lower back pain in Brooklyn," "can I see a PT directly," "physical therapy near Williamsburg."
For a chiropractor: "chiropractor in Brooklyn that takes insurance," "best chiropractor in my neighborhood," "chiropractic adjustment for neck pain," "chiropractor near the subway."
Write down eight to twelve questions. Mix specific neighborhood queries with broader ones. Include insurance and condition questions. Include the exact phrasing patients use: "near me," "take Medicaid," "open today."
Set Up Your Log
Create a simple spreadsheet or a text file with these columns:
Question: the exact prompt you asked
Engine: ChatGPT, Perplexity, Google AI, Claude
Named? Yes or No (your practice was named)
Position: First, second, third, not listed
Source cited: the URL the engine linked to
Competitor named: other practices that appeared
Date: when you ran the check
You can work in Google Sheets, Excel, Notion, or a text editor. The format doesn't matter. Consistency does.
Run the Audit
Ask each question through each engine on the same day. Use the web versions, signed out where you can, so your own history doesn't shape the answer. Write down exactly what each engine says.
Open ChatGPT. Paste your first question. Read the full answer. If your practice is named, note the position and the source URL. If a competitor is named instead, note their name and position. Copy the relevant text into your log or take a screenshot.
Repeat for Perplexity. Then Google: run the question as a normal search and note whether an AI Overview appears at the top, and what it says if it does. Then Claude.
Move to the next question. Run all four engines again.
This is methodical but not slow. Eight questions across four engines is thirty-two checks. If you spend three minutes per check, reading the answer and writing down the result, you're done in ninety minutes.
What Your Log Will Show
After you've logged ten questions across the four engines, patterns emerge.
You'll see which engines name you for which queries. You'll see which competitors are appearing instead. You'll notice which sources the engines choose: your website, Google Business Profile, review sites, Reddit, directories. You'll see whether your neighborhood queries get answered locally or bounce out to borough-level results.
This is not a ranking. A single day of checks shows you where you sit today, not a trend. Answers vary from one day to the next, which is why our own published checks give the date, the engine and the exact question for every answer. But it shows you the reality. If you're not named for "PT in Crown Heights Brooklyn," you'll see which practices are named instead, and you can study their sites to understand why. If you're named first for one question but not mentioned for another, you can adjust your content strategy toward the questions that retrieve you.
Make It Repeatable
Run this audit again in sixty days. Ask the same questions. Log the results in the same spreadsheet. Compare.
If you were named zero times in October and named twice in December, you've moved. If you stayed at zero, something in your site or schema needs attention. If you moved from position five to position two for a specific question, you can see which content changes happened around the same time: a new FAQ section, a schema update, or a neighborhood page you published. A log shows you what changed alongside a result; it can't prove the cause.
The log also becomes a record. If an engine's answer changes (your source URL shifts, a competitor disappears, a different business moves to first position), you'll see the shift. That visibility is worth more than a single month's snapshot.
What This Reveals About Your Schema
Your audit log tells you something your schema checker alone cannot: whether the markup you built is actually getting you retrieved and cited, or whether it's sitting on a page the engines aren't pulling from.
If you're not named for a question you should be ranking for, your next step is to check that you have the schema in place: LocalBusiness, MedicalBusiness, and FAQPage if you're answering common questions. Being named shows an engine found and trusted something about you, but it doesn't tell you which page or which markup did it. If you're not named for a question your patients ask, the audit tells you exactly which content needs to be built or which markup needs to be fixed.
Log It, Study It, Repeat
The audit works because you are asking real questions and recording real answers. You're not guessing. You're not paying for a tool that abstracts the data. You're seeing what patients see when they ask your engines.
Run the audit quarterly. Keep the log. Watch what changes. When you make a content update or add schema, you'll have a baseline to measure against. When you talk to a developer about your visibility, you'll have data that came from the engines themselves, not from a report.