An AI visibility tool asks AI assistants a fixed set of questions on a schedule and records whether your practice is named in the answers. For a medical practice, the useful version asks what local patients ask ("pediatric dentist in Tulsa that takes Delta Dental", not "best CRM software"), covers every major assistant, runs from your city and repeats every week. This hub explains what these tools measure, how to judge one, what the numbers mean, and how to see the visits AI assistants actually send to your website.
People are starting to look for this kind of tracking:
| Search (US, monthly) | Volume |
|---|---|
| ai visibility tool | 880 |
| ai search visibility tool | 720 |
| ai visibility tracking | 320 |
| ai rank tracker | 260 |
| llm visibility | 260 |
| ai brand monitoring | 170 |
Google Ads Keyword Planner data via DataForSEO, US, 12-month average to Sep 2026. A click on "ai visibility tool" costs about $67.53, a sign that software vendors bid for these searchers.
Most of what ranks for these searches today is a roundup of tools built for national brands that want to know whether ChatGPT mentions their sneakers or their accounting software. A practice has a narrower and more local question: when someone in my city asks an assistant for my kind of care, am I in the answer? The criteria below are written for that question.
What AI visibility tools measure
Every tool runs the same basic loop. It sends a list of prompts to one or more assistants, saves the answers, extracts the names and links in each answer, and repeats on a schedule so you can see change over time.
From each answer, a good tool records:
- Mention: whether your practice is named at all.
- Position: whether you're the first recommendation or the sixth in a list.
- Who appears instead: the other practices named for the same question.
- Citations: which pages the answer drew on, such as your website, a directory profile, a review site or a news story.
- Accuracy: whether the description of you is right (hours, services, insurance, location). Few tools judge this automatically, so plan to read the answers yourself.
What a tool can't measure is demand. It sees only the answers it requested, not the conversations your patients had, so it can tell you whether you were named for a question but not how many people asked it. Treat visibility data as a share of a conversation of unknown size, and pair it with the visit data covered further down.
For the bigger picture of why assistants name some practices and not others, start with AI visibility for practices. If you want to see where you stand today before reading further, the free marketing audit includes an AI visibility score.
The prompts that matter for a practice
The prompt list decides everything downstream. A tool that tracks the wrong questions will report precise numbers about nothing useful.
Practice prompts fall into four groups:
- Discovery: the patient doesn't know who to call. "Best family doctor in Boise", "pediatric dentist near Plano".
- Service: the patient knows the treatment. "Where can I get Invisalign in Columbus", "skin cancer screening Mississauga".
- Constraint: the patient has a filter. "Therapist in Denver who takes Aetna", "dentist open Saturday in Tampa", "Spanish-speaking pediatrician in Phoenix".
- Reputation: the patient already has your name. "Is Cedar Hill Family Dentistry good?", "Does Cedar Hill take Delta Dental?" These test accuracy, not discovery.
Illustrative example (fictional practice): Cedar Hill Family Dentistry in Columbus, Ohio, tracks 12 prompts: four discovery ("best family dentist in Columbus Ohio"), four service ("dental implants Columbus", "emergency dentist Columbus open now"), three constraint ("Columbus dentist that takes Medicaid for kids") and one reputation prompt about the practice by name.
Three rules keep the list useful. Write prompts the way patients phrase them, including the city. Keep the list fixed for months so trends mean something. And put the city in the prompt rather than relying on "near me": an assistant resolves "near me" from wherever the request comes from, which for a cloud-based tool may not be your town.
Evaluation criteria: assistants, locations, frequency, history
Use these questions to compare any AI visibility tool, including a spreadsheet you run yourself.
| Criterion | What to ask | Why it matters for a practice |
|---|---|---|
| Assistants | Which ones: ChatGPT, Gemini, Claude, Perplexity, Google AI Overviews? | Each uses different sources; patients don't all use the same one. |
| How answers are collected | Consumer apps or developer APIs? Web search on or off? | The setup can change the answer, so you want it stated. |
| Location | Can prompts run as a specific city, or do you name the city in the prompt? | Local answers depend on location. |
| Prompt control | Can you write specialty-specific prompts, or only use a preset list? | Generic prompts miss how your patients search. |
| Frequency | How often, and how many runs per prompt? | Answers vary; repeated runs smooth the noise. |
| History | How far back, and can you export it? | Trends matter more than any single week. |
| Competitors and citations | Does it show who was named instead and which pages were cited? | That is what tells you what to fix. |
| Reporting | Is there a plain-language summary an owner will read? | A chart nobody opens changes nothing. |
| Data needed | Does it ask for anything beyond your name, website and city? | It should never need patient information. |
Google AI Overviews deserve their own line. Google says AI Overviews "are only shown when our systems determine that it is additive to classic Search," and that AI Mode and AI Overviews "may use different models and techniques, so the set of responses and links they show will vary" (Google Search Central (opens in a new tab)). A tool should therefore record when no AI Overview appeared at all, which is different from appearing without you. Our guide to AI Overview tracking covers that case in detail.
Measuring visits from AI assistants
Mentions are upstream. Visits are what show whether being named sends anyone to your site.
Google Analytics 4 now has a default channel for this. Google defines it as "the channel by which users arrive at your site from sources like ChatGPT, Gemini, Deepseek, Copilot, or Grok," and notes that it "excludes Google's AI Overviews and AI Mode," which GA4 counts as Organic Search (Google Analytics Help (opens in a new tab)). Google doesn't publish the full list of referrers behind that channel, which is one reason some practices build their own channel group.
For AI Overviews and AI Mode, Google says the traffic is reported in Search Console's Performance report, "within the 'Web' search type" (Google Search Central (opens in a new tab)). It isn't separated from ordinary search clicks there.
Two gaps remain. Visits that arrive without referrer information are counted as Direct, so some AI-assisted visits will hide there. And some patients read an answer, then search your name or call you without clicking anything, which no analytics tool can attribute to the assistant.
Medfy's first-party website analytics counts visits from AI assistants alongside calls, forms, bookings and directions, so they appear next to the actions that matter rather than in a separate report.
Technical access: crawlers and llms.txt
An assistant can only describe what it can read. Each major provider documents its own crawlers, and they do different jobs:
| Provider | Search or user-request crawler | Training crawler or control |
|---|---|---|
| OpenAI | OAI-SearchBot (ChatGPT search), ChatGPT-User | GPTBot |
| Anthropic | Claude-SearchBot, Claude-User | ClaudeBot |
| Perplexity | PerplexityBot, Perplexity-User | None documented; PerplexityBot "is not used to crawl content for AI foundation models" |
| Googlebot (AI Overviews and AI Mode use Search) | Google-Extended token |
Sources: OpenAI (opens in a new tab), Anthropic (opens in a new tab), Perplexity (opens in a new tab), Google (opens in a new tab), checked October 22, 2026.
The practical point: blocking a search crawler can keep your pages out of that assistant's answers, while blocking a training crawler is a separate choice. OpenAI, for example, says sites that opt out of OAI-SearchBot "will not be shown in ChatGPT search answers." Medfy's own site, medfy.ai, allows every crawler in its robots.txt; that's an example of the open setting, not a recommendation for every practice.
llms.txt is a newer idea: a plain Markdown file proposed by Jeremy Howard in September 2024 to help AI agents find a site's key pages (llmstxt.org (opens in a new tab)). Google says Search ignores such files and that you "don't need to create new machine readable files, AI text files, markup, or Markdown to appear in Google Search" (Google Search Central (opens in a new tab)). It costs little to add one, but don't expect it to change what assistants say.
The free audit checks this layer for you: it tests whether the crawlers behind ChatGPT, Perplexity, Claude and Gemini can read your site.
Build vs buy: manual tracking or a tool
You can track AI visibility by hand. Ten prompts across five assistants is 50 answers to read and log each week, in a private browser window, from the same location, on the same day. Many practices start this way to learn what the answers look like before paying for anything.
| Option | What you get | Cost |
|---|---|---|
| Manual spreadsheet | Full control of prompts; slow to run and easy to let lapse | Staff time |
| Free AI visibility checker | Builds patient-style questions for your specialty and city, opens each in ChatGPT, Perplexity, Claude or Google AI Mode, and keeps a scorecard in your browser | Free |
| Free marketing audit | An AI visibility score from 0 to 100: asks Claude and Gemini, with live web search, whether they name you, and checks that the crawlers behind ChatGPT, Perplexity, Claude and Gemini can read your site | Free |
| Medfy AI visibility tracking | Weekly checks of ChatGPT, Gemini, Claude, Perplexity and Google AI Overviews using the questions patients ask, such as "best family doctor near me" in your city; shows whether you're named, who is recommended instead and how that changes | Included in the $349/month base plan |
Buy a dedicated tool when the manual log starts slipping, or when you need history you can show an owner or partner. Compare vendors on the criteria table above and ask how the price changes as you add prompts, assistants or locations. Medfy's tracking comes bundled with the website, automated SEO, analytics and the AI CMO dashboard rather than as a standalone tool; see Medfy pricing for what the base plan includes.
No tool can make an assistant name you. AI answers vary by phrasing, location and day, and tracking only shows you what happened.
Guides in this section
- AI visibility score and share of voice: how tools calculate the number, why it swings week to week, and how to read the trend.
- ChatGPT referral traffic: the referrers to look for, a GA4 channel group for AI assistants, and what analytics still misses.
- llms.txt for practice websites: what the proposal is, who documents using it, and an example file for a practice.
- ClaudeBot, GPTBot and AI crawlers: what each crawler is for, the robots.txt tokens, and the allow-or-block trade-off.
Frequently asked questions
What is the best way to track AI visibility?
Fix a set of 10 to 20 patient-style questions that name your city and specialty. Run them on ChatGPT, Gemini, Claude, Perplexity and Google AI Overviews on the same day each week, and log whether you're named, who is named instead and which sources are cited. Read the trend over several weeks rather than any single answer, because responses shift with phrasing and timing.
How much do AI visibility tools cost?
Prices vary by vendor and change often, so compare current pricing pages and ask how the price moves as you add prompts, assistants or locations. Manual tracking costs staff time only. Medfy includes weekly AI visibility tracking in its $349/month base plan, alongside the website, SEO, analytics and dashboard. The free audit and the free AI visibility checker cost nothing.
Which AI assistants should a practice track?
Start with the major general assistants: ChatGPT, Google Gemini, Claude and Perplexity, plus Google AI Overviews, which appear inside ordinary Google searches. Each draws on different sources, so a practice named by one can be missing from another. Google also notes that AI Overviews and AI Mode may use different models, so the responses and links they show vary.
Can I track AI visibility for free?
Yes, if you accept some manual work. Medfy's free AI visibility checker builds patient-style questions for your specialty and city, opens each in ChatGPT, Perplexity, Claude or Google AI Mode, and keeps a scorecard in your browser. The free marketing audit adds an automated check: it asks Claude and Gemini, with live web search, whether they name your practice, and tests whether AI crawlers can read your site.
How often should AI visibility be checked?
Weekly suits most practices. Answers change from run to run, so daily checks mostly record noise, and monthly checks are too sparse to separate a real shift from a bad day. Keep the prompts, the day and the location the same each week, and judge movement over four to eight weeks. After a website relaunch or a listing change, keep the cadence and watch the trend.