Does Google Analytics show traffic from ChatGPT?

Partially. Referrals from chatgpt.com and perplexity.ai appear in your referral report, but they undercount badly — many users read the answer without clicking, and some assistants strip referrer data. Treat referral numbers as a floor, not a measurement.

How often should I audit AI visibility?

Monthly is the right cadence for most practices. AI engines update on different cycles and individual answers vary between runs, so weekly checks mostly capture noise. Monthly over six months produces a trend you can actually act on.

Do AI assistants give the same answer to everyone?

No. Responses vary by phrasing, location, conversation history, and model version, and the same question asked twice can return different sources. This is why you track the pattern across a fixed question list over time rather than treating any single answer as definitive.

What's a realistic target for AI citation rate?

For a single-location practice in a competitive market, appearing in three to five of fifteen tracked questions within six months is solid progress. Treatment and cost questions are more winnable than broad 'best med spa in city' queries, which are dominated by directories.

Should I track branded search volume too?

Yes, it is one of the better proxies available. When AI assistants recommend you without a click, some people search your practice name afterward. A rising branded search trend in Search Console alongside flat organic clicks often indicates AI visibility working.

How to Measure Whether AI Search Recommends Your Med Spa

The Measurement Gap

You can tell exactly how many people found you through Google organic search last month, which queries brought them, and what they did next.

You cannot tell how many asked ChatGPT for a med spa recommendation and were told about you.

That gap is uncomfortable, and it is the most common objection to investing in AI visibility at all: how do you justify work you cannot measure?

The honest answer is that you can measure it, just not through a dashboard. The method is manual, takes about an hour a month, and produces a genuinely useful trend line.

Build a Fixed Question List

Everything depends on asking the same questions every time. A list that changes month to month measures nothing.

Aim for fifteen to twenty questions across four categories.

Provider selection. "What's the best med spa in [your city]?" "Where should I go for Botox in [city]?" These are hardest to win — directories and aggregators dominate — but they are the highest-value.

Treatment questions. "How long does lip filler last?" "Is Morpheus8 worth it?" Most winnable, and where your content work shows up first.

Cost questions. "How much does Botox cost in [city]?" Very high intent and frequently under-answered by practices.

Comparisons. "Morpheus8 vs microneedling." "Dysport or Botox?" Strong citation targets, since the model needs a source presenting both sides.

Write them the way a patient would speak them, not as keyword fragments. AI assistants receive full questions, and phrasing changes results.

Run the Audit

Once a month, work through the list on three surfaces.

Perplexity first, because it cites every source explicitly. You get an unambiguous answer per query and a list of who was cited instead of you. Our post on getting cited by Perplexity covers what to do with that.

ChatGPT second. Ask each question in a fresh conversation with no history, since prior context contaminates results.

Google third, recording whether an AI Overview appeared and which sources it cited. Do this logged out, from your service area.

For each question record four things: were you mentioned, were you cited with a link, which competitors appeared, and how many sources were used. A simple spreadsheet with one row per question and one column per month is enough.

The Supporting Signals

The manual audit tells you about visibility. Three other numbers tell you whether it is producing anything.

Referral traffic from AI domains. Filter your analytics for chatgpt.com, perplexity.ai, and similar. These numbers are small and undercount heavily, since most users read without clicking. Watch the direction rather than the magnitude.

Branded search volume. In Search Console, track impressions and clicks for your practice name. When an assistant recommends you without a link, a meaningful share of people search your name afterward. Rising branded search alongside flat non-branded organic is one of the clearer fingerprints of AI visibility.

Consultation source attribution. The most direct signal, and the one most practices skip. Add "AI assistant (ChatGPT, etc.)" to your intake form's "how did you hear about us" field. Answers will trickle in slowly and then less slowly.

Reading the Results

Expect variance. The same question asked twice can produce different sources, and a month where you appear in five of fifteen queries followed by a month at three is not necessarily a decline.

What matters is direction over three to six months, and category patterns. If you are cited on treatment questions but never on provider-selection queries, that is not a failure — it is a map. Treatment questions respond to content structure. Provider queries respond to reviews, listings, and third-party corroboration, which is a different body of work covered in how AI search recommends med spas.

Also read the competitors. If the same practice is cited across eight of your fifteen questions, their content is worth studying page by page. That is a free content-gap analysis performed by the engine you are trying to win.

What the Spreadsheet Should Look Like

Keep it simple enough that you will actually maintain it. One tab, one row per question, columns added each month.

For each question, record a short code rather than prose:

C — cited with a link to your site
M — mentioned by name but not linked
A — absent
Plus the competitors named, in a single column

A row might read: "How much does Botox cost in Nashville?" — Mar: A · Apr: A · May: M · Jun: C · Jul: C with a competitor column listing the two practices that keep appearing.

That single row tells a story a dashboard could not: the page went from invisible to mentioned in three months, then to cited once the cost section was rewritten with a specific range.

Add two summary figures at the bottom. Citation rate — how many of your fifteen questions produced a C or M this month, as a fraction. Competitor frequency — which practices appear most across all questions. The first is your trend line. The second is your reading list.

Resist the urge to make this elaborate. Practices that build a twelve-tab tracking workbook abandon it by month three. A single sheet that takes fifty minutes a month survives, and surviving is the entire value.

Setting Expectations Honestly

AI visibility moves slowly. Content published this month may not surface in citations for two or three months, and some engines update their retrieval and training on cycles nobody outside those companies can see.

A realistic first six months for a practice starting from near-zero looks like: months one and two, no visible change; month three, occasional citations on treatment questions; months four through six, steadier citation on treatment and cost questions, with provider-selection queries still dominated by directories.

Anyone promising faster than that is guessing.

Run Your First Audit This Month

Write fifteen questions. Run them across the three surfaces. Record the results.

It will take about an hour, and it is the only baseline you will ever get for free. Every month after is a comparison against it.

If you would rather have this run and reported for you, book a free visibility check and we will show you where you currently stand.

How to Measure Whether AI Search Is Recommending Your Med Spa